- Email: [email protected]
Practice and Evaluation of Model Based Development (MBD) Education
12th IFAC Symposium on Advances in Control Education 12th IFAC Symposium on Advances in Control Education July Philadelphia, PA, USAin 12th 7-9, IFAC2019. Symposium on Advances Advances in Control Control Education Education 12th IFAC Symposium on online at www.sciencedirect.com July 7-9, Philadelphia, PA, USAinAvailable 12th IFAC2019. Symposium on Advances July 2019. Philadelphia, PA, July 7-9, 7-9, 2019. Philadelphia, PA, USA USA Control Education July 7-9, 2019. Philadelphia, PA, USA
ScienceDirect
IFAC PapersOnLine 52-9 (2019) 206–211
Practice and Evaluation of Model Based Practice and Evaluation of Model Based Practice and Evaluation of Model Based Development (MBD) Education Practice and Evaluation of Model Based Development (MBD) Education Development (MBD) Education Development (MBD) Education S. Wakitani ∗ T. Yamamoto ∗ C. Morishige ∗∗ T. Adachi ∗∗
∗ ∗ ∗∗ ∗∗ S. Yamamoto ∗ C. Morishige ∗∗ T. Adachi ∗∗ ∗∗ ∗ ∗∗ ∗∗ S. Wakitani Wakitani ∗∗∗ T. T. Yamamoto C.Muraoka Morishige T. Adachi Adachi ∗∗ Harada S. Wakitani T.Y. Yamamoto C. Morishige ∗∗ ∗T. ∗∗ ∗∗ T. Harada ∗∗ T. ∗∗ S. Wakitani T.Y. C.Muraoka Morishige T. Adachi ∗∗ ∗∗ Y. Harada T. Muraoka Y.Yamamoto Harada ∗∗ ∗∗ T. Muraoka ∗∗ Y. Harada T. Muraoka ∗ ∗ Graduate School of Engineering, Hiroshima University, Hiroshima, School of Engineering, Hiroshima University, Hiroshima, ∗ ∗ Graduate Graduate School of Engineering, Hiroshima 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). School of Engineering, Hiroshima University, University, Hiroshima, Hiroshima, ∗ Graduate 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). ∗∗ Graduate School of Engineering, Hiroshima Hiroshima, 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). Motor Corporation, Hiroshima,University, 730-8670 Japan 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). ∗∗ Mazda Mazda Motor Corporation, Hiroshima, 730-8670 Japan ∗∗ 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). ∗∗ Mazda Motor Corporation, Hiroshima, 730-8670 Japan ∗∗ Mazda Motor Corporation, Hiroshima, 730-8670 Japan Mazda Motor Corporation, Hiroshima, 730-8670 Japan Abstract: This study presents the practical results concerning model-based development Abstract: This study presents the practical results concerning model-based development Abstract: study presents results concerning model-based development educational This practices implemented by practical Hiroshima University and Mazda Motor Corporation, Abstract: This study presents the the practical results concerning model-based development educational practices implemented by Hiroshima University and Mazda Motor Corporation, Abstract: This study presents the practical results concerning model-based development educational practices implemented by Hiroshima University and Mazda Motor Corporation, an automobile manufacturer, and the consequent educational effects. Importance of modeleducational practices implemented by Hiroshima University and Mazda Motor Corporation, an automobile manufacturer, and the consequent educational effects. Importance of modelmodeleducational practices implemented by Hiroshima University and Mazda Motor Corporation, an automobile manufacturer, andinthe the consequent educational effects. Importance of based development is increasing industrial world. Model-based development is performed an automobile manufacturer, and consequent educational effects. Importance of modelbased development is increasing in industrial world. Model-based development is performed an automobile andin consequent educational effects. Importance of modelbased development is industrial world. Model-based development is according to a manufacturer, development process called the ”V-type development process (V-process).” based development is increasing increasing inthe industrial world. Model-based development is performed performed according to a development process called the ”V-type development process (V-process).” based development is increasing in industrial world. Model-based development is performed according to a development process called the ”V-type development process (V-process).” However, several professional engineers involved in automobile development have insufficient according to a development process called the in ”V-type development process (V-process).” However, several professional engineers involved automobile development have insufficient according to a development process called the in ”V-type development process (V-process).” However, several professional engineers involved in automobile development have insufficient experience in conducting model-based development. In this study, the authors propose a program However, several professional engineers involved automobile development have insufficient experience in conducting model-based development. In this study, the authors propose program However, several professional engineers involved in automobile development have insufficient experience inthese conducting model-based development. In this this study, the the authors propose aa program program that allowsin professional engineers to learn the V-process using a motor control system experience conducting model-based development. In study, authors propose a that allows these professional engineers to learn the V-process using a motor control system experience inthese conducting model-based development. In this study, the authors propose aengineers program that professional to the V-process using a motor control system via aallows hands-on approach. Theengineers program is learn implemented to target 159 professional that allows these professional engineers to learn the V-process using a motor control system via a hands-on approach. Theengineers program is implemented to target 159 professional engineers that professional to V-process using motor by control system via a hands-on approach. program is implemented to 159 belonging tothese Mazda Motor The Corporation. From thethe questionnaire surveys taken theengineers learners, via aallows hands-on approach. The program From is learn implemented to target target 159a professional professional engineers belonging to Mazda Motor Corporation. the questionnaire surveys taken by the learners, via a hands-on approach. The program is implemented to target 159 professional belonging to Mazda Mazda Motor Corporation. From the questionnaire questionnaire surveys taken by the theengineers learners, it is revealed that the V-process exercises considerably contribute to the self-efficacy of the belonging to Motor Corporation. From the surveys taken by learners, it is is revealed revealed that the the V-process exercises considerably contribute to the the self-efficacy of the the belonging Mazda Motor Corporation. From the questionnaire surveys taken by the learners, it that V-process exercises considerably contribute to self-efficacy of learners fortooperational performance using model-based development. it is revealed that the V-process exercises considerably contribute to the self-efficacy of the learners for operational performance using model-based development. it is revealed that the V-process exercises considerablydevelopment. contribute to the self-efficacy of the learners for performance using learners for operational operational performance using model-based model-based development. © 2019, IFAC (International Federation ofusing Automatic Control) Hosting by Elsevier Ltd. All rights reserved. learners for operational performance model-based development. Keywords: Curriculum development; engineering education; model-based development; v-type Keywords: Curriculum development; engineering education; model-based development; v-type Keywords: Curriculum Curriculum development; engineering engineering education; education; model-based model-based development; development; v-type v-type development process development; Keywords: development process Keywords: Curriculum development process development; engineering education; model-based development; v-type development process development process 1. relevant systems are modeled, the modeled components 1. INTRODUCTION INTRODUCTION relevant systems systems are are modeled, modeled, the the modeled modeled components components 1. relevant are integrated. Accordingly, analysis of the components behavior of 1. INTRODUCTION INTRODUCTION relevant systems are modeled, the modeled are integrated. Accordingly, analysis of the behavior of 1. INTRODUCTION relevant systems are modeled, thedesign, modeled are integrated. analysis of the behavior of the system and Accordingly, controlled system arecomponents performed. are integrated. Accordingly, analysis of the behavior of Customer needs for products have become more diversified the system and controlled system design, are performed. are integrated. Accordingly, analysis of the behavior of the system and controlled system design, are performed. In HILS, controller algorithms designed through MILS are Customer needs for products have become more diversified the system and controlled system design, are performed. Customer needs for products have become more diversified and complex over the years.have Companies must diversified promptly In HILS, controller algorithms designed through MILS are Customer needs for products become more the system and controlled system design, are performed. In HILS, algorithms designed through are implemented for actual electronic control unitsMILS (ECUs), and complex over the years. Companies must promptly In HILS, controller controller algorithms designed through MILS are Customer for products have become more and complex over the years. Companies must promptly respond toneeds such demands. However, in regard todiversified intricate implemented for actual electronic control units (ECUs), and complex over the years. Companies mustto promptly In HILS, controller algorithms designed through MILS are implemented for actual electronic control units (ECUs), and their robustness and safety are verified via real-time respond to such demands. However, in regard intricate implemented for actual electronic control units (ECUs), and complex over the years. Companies must promptly respond to such demands. However, in to systems such as automobiles, countless combinations of and their robustness and safety are verified via real-time respond to such demands. However, in regard regard to intricate intricate implemented for actual electronic control units (ECUs), and their robustness and safety are verified via real-time systems such as automobiles, countless combinations of simulations. However,and in relation to verified using control targets and their robustness safety are via real-time respond such However, in regard toare intricate systems such as automobiles, countless combinations of optimumto components and adjustable parameters possiHowever, in relation to using control targets systems such as demands. automobiles, countless combinations of simulations. and their robustness and safety are verified via real-time simulations. However, in relation to using control targets for verification, a computer (i.e., hardware in the loop optimum components and adjustable parameters are possisimulations. However, in relation to using control targets systems such as automobiles, countless combinations of optimum components and adjustable parameters are possiin the loop ble. Long components times and high costs are required for formulating verification, a computer (i.e., optimum andcosts adjustable parameters are possi- for simulations. However, incan relation tohardware using control targets for verification, a computer (i.e., hardware in the (HIL) simulator) that mimic the behaviors of the ble. Long times and high are required for formulating for verification, a computer (i.e., hardware in the loop optimum components andcosts adjustable parameters are possi- (HIL) simulator) that can mimic the behaviors of loop ble. Long times and high are required for formulating designs and adjustments using actual samples. Moreover, the ble. Long times and high costs are required for formulating for verification, a computer (i.e., hardware in the loop (HIL) simulator) that can mimic the behaviors of the control targets is employed. Consequently, it has become designs and adjustments using actual samples. Moreover, (HIL) simulator) that can mimic the behaviors of the ble. Long times and high costs are required for formulating designs and adjustments using actual samples. the cost and time required for manufacturing must be targets is employed. Consequently, it has become designs using samples. Moreover, Moreover, (HIL) simulator) that can mimic the of the control targets employed. Consequently, it the cost costand andadjustments time required required foractual manufacturing must be be control unnecessary to is produce prototypes for behaviors controller verificontrol targets isproduce employed. Consequently, it has has become become designs and adjustments using actual samples. the time for manufacturing reduced toand the maximum extent possible owing Moreover, tomust the inunnecessary to prototypes for controller verifithe cost and time required for manufacturing must be control targets is employed. Consequently, it has become unnecessary to produce prototypes for controller verification. Thus, the time and costs incurred for controller reduced to the maximum extent possible owing to the intothe produce prototypes for controller verifithe required forcurrently. manufacturing be unnecessary reduced to the maximum extent possible owing to the incation. Thus, time and costs incurred for controller tensecost international competition Given this backreduced toand thetime maximum extent possible owing this tomust the inunnecessary tothe produce prototypes for controller verification. Thus, and costs incurred for controller development can betime substantially reduced. The automotive tense international competition currently. Given backcation. Thus,can the time and costsreduced. incurred forautomotive controller reduced to the maximum extent possible owing to the intense international competition currently. Given this background, importance of model-based development (MBD) development be substantially The tense international competition currently. Given this backcation. Thus, the time and costs incurred for controller development can be substantially reduced. The automotive industry has can proactively adopted reduced. MBD forThe product develground, importance of model-based development (MBD) development be substantially automotive tense international competition currently. Given this back- industry ground, importance of development (MBD) (Larses and N.Adamsson (2004); Ohata and Kenneth has proactively adopted MBD for product develground, importance of model-based model-based development (MBD) development can bethere substantially reduced. The automotive industry has proactively adopted MBD for product devel(Larses and N.Adamsson (2004); Ohata and Kenneth opment. However, are only few professional engineers industry has proactively adopted MBD for product develground, importance of model-based development (MBD) (Larses and (2004); Ohata and (2008); Ogata and Katayama (2009); Jaikamal (2009); opment. However, there are only few professional engineers (Larses and N.Adamsson N.Adamsson (2004); Ohata and Kenneth Kenneth industry has proactively adopted MBD for product development. However, there only professional engineers involved in automobile development having sufficient ex(2008); Ogata and Katayama (2009); Jaikamal (2009); opment. there are are only few few professional engineers (Larses and N.Adamsson Ohata and Kenneth (2008); Ogata and Katayama (2009); (2009); involved However, in automobile development having sufficient exSingh (2014); Franco et al. (2004); (2016)) has Jaikamal increased. Thus, involved (2008); Ogata and Katayama (2009); Jaikamal (2009); opment. However, there are only few professional engineers in automobile development having sufficient experience for performing MBD. Singh (2014); Franco et al. (2016)) has increased. Thus, involved in automobile development having sufficient ex(2008); Ogata and Katayama (2009); Jaikamal (2009); Singh (2014); Franco et al. (2016)) has increased. Thus, for performing MBD. in recent years, difficult systems have been implemented perience Singh (2014); Franco et systems al. (2016)) hasbeen increased. Thus, perience involved for in automobile development having sufficient experforming MBD. in recent years, difficult have implemented for performing MBD. Singh et systems al.this (2016)) hasbeen increased. Thus, perience in years, difficult have implemented on recent the(2014); computer through methodology, and system The ʠ Hirojiren ʡ (i.e., Hiroshima Council for the Promoin recent years,Franco difficult systems have been implemented perience for performing MBD. The ʠ Hirojiren ʡ (i.e., Hiroshima Council the on the computer through this methodology, and system in recent years, difficult systems have beenApplication implemented on the computer through this methodology, and system The ʠ Hirojiren ʡ (i.e., (i.e., Hiroshima Council for for the PromoPromodesign and verification can be performed. of tion of Collaboration between Government, Academia, and on the computer through this methodology, and system The ʠ Hirojiren ʡ Hiroshima Council for the Promotion of Collaboration between Government, Academia, and design and verification can be performed. Application of on thehas computer through this methodology, and The ʠ Hirojiren ʡ (i.e., Hiroshima Council for the design and verification can be Application of tion of Collaboration between Government, Academia, MBD rapidly progressed inperformed. the industry (e.g. system Reedy Automobile Industry) was established in June 2015Promoin and Hidesign and verification can be performed. Application of tion of Collaboration between Government, Academia, and MBD has rapidly progressed in the industry (e.g. Reedy Automobile Industry) was established in June 2015 in Hidesign and verification can be performed. Application of of Collaboration between Government, Academia, and MBD has progressed in industry (e.g. Reedy Automobile Industry) was established in June 2015 in Hiand Lunzman (2010)). In MBD, in contrast to convenroshima Prefecture in regard to the ”Industry-AcademiaMBD has rapidly rapidly progressed in the the industry (e.g. Reedy tion Automobile Industry) was established in June 2015 in Hiand Lunzman (2010)). In MBD, in contrast to convenroshima Prefecture in regard to the ”Industry-AcademiaMBD has rapidly progressed in the industry (e.g. Reedy Automobile Industry) was established in June 2015 in Hiand (2010)). MBD, in to roshima Prefecture Prefecture in regard regardVision to the the2030.” ”Industry-AcademiationalLunzman source codes, each In component iscontrast defined as a convensimula- roshima Government Collaborative The Hirojiren, and Lunzman (2010)). In MBD, in contrast to convenin to ”Industry-AcademiaGovernment Collaborative Vision 2030.” The Hirojiren, tional source codes, each component is defined as a simulaand Lunzman (2010)). In MBD, in contrast to convenroshima Prefecture in regard to the ”Industry-Academiational source codes, each component is defined as a simulaGovernmentofCollaborative Collaborative Vision 2030.” The Hirojiren, Hirojiren, tion block (i.e., executable specification), and information comprising three committees and four expert committional source codes, each component is defined as a simula- Government Vision 2030.” The comprising of three committees and four expert committion block (i.e., executable specification), and information tional source each component as a simulaGovernment Collaborative Vision 2030.” The committee Hirojiren, tion block (i.e., executable specification), information of three committees and four expert commitregarding thecodes, relationships betweenis defined theand components is comprising tees, undertakes relevant activities. The expert tion block (i.e., executable specification), and information comprising of three committees and four expert committees, undertakes relevant activities. The expert committee regarding the relationships between the components is tion block (i.e., executable specification), and information comprising of three committees and four expert commitregarding the relationships between the components is tees, undertakes relevant activities. The expert committee described via a block diagram. Thus, it is possible to verify for MBD established ”Fundamentals of MBD Laboratory regarding the relationships between the components is tees, undertakes relevant activities. The expert committee described via a block diagram. Thus, it is possible to verify for MBD established ”Fundamentals of MBD Laboratory regarding the relationships between the components is tees, undertakes relevant activities. The expert committee described via diagram. Thus, possible to for MBD MBD established ”Fundamentals ofinMBD MBD Laboratory various operations using the computer. MBD is executed (MBD lab)” in Hiroshima Universityof 2016,Laboratory Hiroshima described via aa block block diagram. Thus, it it is isMBD possible to verify verify for established ”Fundamentals various operations using the computer. is executed lab)” in Hiroshima University 2016, Hiroshima described via the a block diagram. Thus, it is possible to for MBD established ”Fundamentals ofin Laboratory various operations using the computer. MBD is (MBD lab)” in Mazda Hiroshima University inMBD 2016, Hiroshima in line with development process known as executed theverify ”V- (MBD University and Motor Corporation, the Japanese various operations using the computer. MBD is executed (MBD lab)” in Hiroshima University in 2016, Hiroshima in line with the development process known as the ”VUniversity and Mazda Motor Corporation, the Japanese various operations using the(V-process).” computer. MBD executed (MBD lab)”and in Mazda Hiroshima University in 2016, Hiroshima in line with the development process known as the ”VUniversity Motor Corporation, the Japanese type development process Theis V-process automobile manufacturer, played a central role in the in line with the development process known as the ”VUniversity and Mazda Motor Corporation, the Japanese type development process (V-process).” The V-process automobile manufacturer, played a central role in the in line with the development process known as the ”VUniversity and Mazda Motor Corporation, the Japanese type development process (V-process).” The V-process automobile manufacturer, played a central role in the comprises the following operations for which computer establishment of this course. In relation to the MBD type development process (V-process).” The V-process automobile manufacturer, played a central role inLab., the comprises the following operations for which computer establishment of this course. In relation to the MBD Lab., type development process (V-process).” The V-process automobile manufacturer, played a central role in the comprises operations for which computer establishment of ofof this course. InMBD relation to the the MBD MBD Lab., simulationsthe playfollowing an important role: (i) model in the loop implementation an effective education program is comprises the following operations for which computer establishment this course. In relation to Lab., role: (i) model in the loop simulations play an important implementation of an effective MBD education program is comprises the following operations for which computer establishment of this course. In relation to the MBD Lab., simulations play an important role: (i) model in the loop implementation of an an effective effective MBD(specifically education program program is simulation (MILS) and (ii) hardware in the loopin simulation required for professional engineers for those simulations play an important role: (i) model the loop implementation of MBD education is required for professional engineers (specifically for those simulation (MILS) and (ii) hardware in the loop simulation simulations play anafter important role: (i)the model in the loop implementation of an effective MBD(specifically education program is simulation (MILS) and (ii) hardware in loop simulation required for professional engineers for those (HILS). In MILS, all the components comprising of simulation (MILS) and (ii) hardware in the loop simulation required for professional engineers (specifically for those In MILS, after all the components comprising of (HILS). simulation and (ii) in the loop simulation (HILS). In (MILS) MILS, after after all hardware the components components comprising of required for professional engineers (specifically for those (HILS). In MILS, all the comprising of (HILS). MILS, after all the components of Hosting by Elsevier Ltd. All rights reserved. 2405-8963In © 2019, IFAC (International Federation of comprising Automatic Control)
Copyright © 2019 IFAC 219 Copyright 2019 IFAC 219 Control. Peer review© under responsibility of International Federation of Automatic Copyright © 219 Copyright © 2019 2019 IFAC IFAC 219 10.1016/j.ifacol.2019.08.197 Copyright © 2019 IFAC 219
2019 IFAC ACE June 1-3, 2016. Bratislava, Slovakia
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
who are beginners, without sufficient experience in development using MBD). Professional engineers undertake the learning program in parallel with performing their work. Thus, a program implementation period of 4 days (i.e., 6 h a day) has been set. An MBD education program that would allow for educational effects with the learners being beginners and the implementation period as constraints was proposed for the MBD lab. Such a program was implemented for 159 people for 4 months. According to the above-mentioned program, control experiments based on physical modeling and motor control systems, as well as basic exercises using MATLAB/Simulink were performed. Moreover, in order to learn HILS via a hands-on approach, a simple hardware in the loop (HIL) simulator using Arduino was developed, which was then utilized for practical training. The questionnaire surveys implemented for each lecture revealed that several learners gained confidence for operational performance via MBD before and after participating in the learning program. In this study, changes of the educational effects on the learners will also be analyzed based on the changes in the results of the questionnaires. The structure of this paper is as follows. In Section 2, an outline of the V-process is first described, and explanations of the results of prior questionnaire surveys performed with targeted learners will be provided. Subsequently, the proposed MBD education program is presented. In Section 3, the motor control system, which is necessary for exercises for the learning program and the simulation models thereof, is explained and the simple HIL simulator using Arduino will be provided. In Section 4, educational effects concerning the learning program and changes of selfefficacy of the learners are observed. Finally, a summary of this work and future issues are described. 2. MODEL-BASED DEVELOPMENT LEARNING PROGRAM 2.1 Model-based Development MBD focuses on visibility compared with conventional development methods based on source codes. Thus, various analyses and verifications can be performed through relevant models. Therefore, it is possible to discover serious systemic defects at an early stage. As a result, the development efficiency can be improved. Case examples in which the development efficiency could be reduced more than 30% based on MBD adoption have been reported by Krasner (2010). MBD is performed in line with the development process named, the V-process. The V-process varies depending upon the number of items according to the degree of details for the descriptions. Briefly, this is described as per Fig. 1. According to MILS, first the requirements for an intended system operations are determined. The combination of components and parameters necessary to satisfy the aforementioned requirements is verified and designed via simulation using a simulation block. In such a case, feasibility for intended specifications may be also reviewed and modifications may be added. According to HILS, controller algorithms designed through MILS are implemented for actual ECUs, and operation verifications are performed by using a HIL simulator. After the aforementioned designing and verification, a modelbased calibration (MBC) that allows for satisfaction of 220
207
Fig. 1. V-Type Development Process. various requirements relating to mass production, such as vehicle performance and law adaptation, is performed for the combined use of actual components and models. The final products are completed after relevant products are tested. In relation to the operations described on the left-hand side of Fig. 1 in the V-process, no verification using actual prototypes is performed. Additionally, at this point, prior extraction and treatment of almost all fatal errors including human errors in product development, are conducted. Therefore, it is possible to prevent unnecessary rework in the operations described on the right-hand side of Fig. 1. Consequently, new product development is possible in a short period. In regard to the learning program, MILS and HILS which are particularly important operations described on the left-hand side of Fig. 1, the V-process will be primarily examined. 2.2 Learners 159 professional engineers of Mazda Motor Corporation undertook the basic course of MBD. The program was implemented for 4 classes (i.e., a single class comprised 40 professional engineers (and 39 professional engineers for some classes)). The learners undertook the learning program concurrently with their operations. Therefore, a 4-day (i.e., 6 h a day and 2 days a week) implementation period for the program was set in advance. Many learners were primarily responsible for operations for vehicle design, vehicle equipment, electrical equipment, and vehicle analysis, instead of development of the control algorithm. Therefore, it could be inferred that such learners were not very familiar with the MBD development process accompanying the controller development. The questionnaire survey shown in TABLE I was provided to the learners prior to implementation of the learning program in order to evaluate their knowledge and predict the degree of understanding of MBD as well as the degree of motivation for MBD operational performance. In relation to all the question items, the learners selected the correct answer from 4 choices of ”1: Not at all,” ”2: Slightly capable,” ”3: Nearly capable,” or ”4: Sufficiently capable” using a 4-point scale. Fig. 2 shows the distribution of responses by the learners prior to implementation of the program. It is observed that although several learners had knowledge about MBD to some extent, they did not have skills or were not confident in the performance of MBD (the details thereof will be discussed in Section 4). Moreover, questionnaire surveys with the same content as that mentioned above were implemented on all 4 days of the lectures. The changes thereof were surveyed as well. The survey results will be discussed in Section 4.
2019 IFAC ACE 208 1-3, 2016. Bratislava, Slovakia June
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
Table 1. QUESTIONNAIRE SURVEY ITEMS. Question Number Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8
The Number of Respondents (%)
Q9 Q10
Question
Evaluation Item
Are you capable of explaining MBD (i.e., Model-Based Development)? Are you capable of explaining the V-Type Development Process? Are you capable of explaining the significance of the modelling? Are you capable of modelling with the use of these software, such as Matlab and Simulink? Are you capable of explaining MILS (Model in the Loop Simulation)? Are you capable of explaining simulation based on MILS through the use of software, such as Matlab and Simulink? Are you capable of explaining HILS (Hardware in the Loop Simulation)? Are you capable of performing a simulation based on HILS by the using a software, such as Matlab and Simulink? Are you capable of explaining the necessity for MBC (Model-Based Calibration)? Are you capable of explaining the operational performance based on the v-type development process or are you confident in the performance?
Knowledge Knowledge Understanding Skills
100 80 60 40 20 0
Q1 1: Not at all
Q2
Q3
Q4
2: Slightly capable
Q5 Q6 Question Items 3: Nearly capable
Q7
Q8
Q9
Q10
4: Sufficiently capable
Fig. 2. Questionnaires Prior to Implementation of the Program. 2.3 Organization of the Learning Program According to the educational program, the following educational objectives have to be achieved: (i) the learners understand the positioning of their own duties relating to MBD after implementation of the program, (ii) they will become interested in proactive use of MBD, and (iii) they will become confident in implementing MBD. In the light of the outline of MBD and the situation of the learners described above, the program is organized as per TABLE 2. On the first half of the 1st day, overview of MBD was provided, which motivated the learners to learn. In the second half of the 1st day, exercises using Simulink allowed the learners to understand the basic method for using MATLAB/Simulink. On the 2nd day, physical modeling was performed based on differential equations for a basic learning of plant modeling, which was inevitable for MBD. Learning regarding the Laplace transform and transfer function as a separate method for expression of plant models was performed. The learners understood that there existed various methods for plant expressions. Moreover, establishment and execution of the aforementioned models via MATLAB/Simulink could lead to the effect of repetitive learning with the methods available with the software. On the 3rd day, application of the knowledge gained thus far allowed modeling of the motor control system. Explanations of the details of the motor control system will be provided in the following section. In here, the entire motor (and the load), sensor, actuator, interface (AD/DA converters), and control logic (Proportional-Integral-Derivative (PID) Controller (Visioli (2006); Vilanova and Visioli (2012))) are described as simulation blocks. Thereby, simulation could be conducted. Moreover, the learners adjusted PID parameters 221
Knowledge Skills Knowledge Skills Understanding Self-efficacy
included in the control logics so that such parameters could satisfy the requirements specifications (i.e., settling time and overshoot) defined by the learners. On the 4th day, real-time simulation was conducted using a simple HIL simulator. Explanations regarding the details of the simple HIL simulator will be provided in the following section. Through this process, the learners were able to discover failures (i.e., response to noises and occurrence of winding-up) that they could not notice through parameter adjustments for MILS anew, and readjust the parameters. In this way, the learners were able to learn that a method for development in line with the V-process could resolve problems relating to failures urgently through a handson approach. Furthermore, the learners implemented controllers obtained through the HILS operations as controllers for experimental devices, and found that results equivalent to those regarding HILS could be obtained. Thereby, the learners were able to achieve success using the V-process. 3. LEARNING MATERIALS Fig. 3 shows the external appearance of the motor control system for the experiment, which was used for exercises on the 3rd and 4th days. Three aluminum disks having diameters different from those of the pulley were connected to the motor shaft. Such disks were connected by a pulley (connected to the dynamo-shaft) and belt. Moreover, the Arduino Mega 2560, motor driver, tachometer, and current sensor were mounted integral to the housing. In the exercises, controlling of rotational speed was set as an assignment. Control parameter adjustment and verification of the controller safety performance were performed via MILS and HILS. 3.1 Model in the Loop Simulation The block diagram of the motor control system is shown in Fig. 4. It allows the learners to understand the relationship between all the components (i.e., blocks). Next, mathematical models corresponding to the aforementioned blocks were obtained and Simulink blocks were constructed. Mathematic models for all the components within Fig. 4 were obtained and Simulink blocks were designed. Consequently, the Simulink model corresponding to the system shown in Fig. 4 was acquired as per Fig. 5. Here, the
2019 IFAC ACE June 1-3, 2016. Bratislava, Slovakia
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
209
Table 2. MBD EDUCATION PROGRAM Day 1 Day 2 Day 3 Day 4
MBD overview and MATLAB/Simulink exercises First-principle modeling (examples of the spring mass damper system) Basics of Laplace transform (first-order lag system) V-Type development process exercises (MILS) V-Type development process exercises (HILS and actual-machine experiment)
Fig. 6. Results of PID control via MILS Fig. 3. Motor Control System (on the Load Side)
Fig. 4. Block Diagram regarding the Motor Control System Fig. 7. Simulink model for controller via HILS via MILS were extracted as per Fig. 7. Direct download to Arduino Mega 2560 took place through the use of a MATLAB/Simulink add-on.
Fig. 5. Simulink model for the motor control system ”Scope Real” block within Fig. 5 directly indicates the signals from the motor. On the other hand, in relation to ”Scope Measure,” only the signals measured inside the computer are indicated. Additionally, in relation to an Arduino block, the PID control law represented by the following formula is implemented. t 1 de(t) u(t) = kc e(t) + (1) e(τ )dτ + TI 0 dt In relation to the PID controller, the duty ratio (u(t) [%]) for the motor driver has been determined to track the target number of rotations (r(t) [rpm]). An example of control results regarding the motor control system is shown in Fig. 6. In Fig. 6, ”N ∗ ” represents the measured rotations [rpm], ”i∗ ” is the measured current [A], and ”duty” is the duty ratio [%]. In the exercises, the learners designed PID parameters via simulations that satisfied the control specifications established by the learners. 3.2 Hardware in the Loop Simulation In here, a PID controller designed via MILS was mounted to Arduino Mega 2560 as a type of micro controllers and operation verifications thereof were performed. Only the Simulink models of the interface and controller section 222
The computer containing the interface revealed that the HIL simulator was used as a control target (plant) of the controller instead of actual prototypes via HILS. The HIL simulator received electrical signals that were output from the controller as digital data via the interface, and computed behaviors regarding plant models inside the simulator based on such received signals. After calculation of such behaviors, electrical signals corresponding to sensor outputs were output from the HIL simulator. In relation to the HIL simulator, in addition to normal systemic behaviors, a realistic situation accompanying dangers could be reproduced on the computer. Thus, it was possible to verify the controller performance under various conditions. Here, the simple HIL simulator shown in Fig. 8 exclusively developed for the learning program is used. The left-hand side section in Fig. 8 represents the controller and the right-hand side section the HIL simulator. The power, ground (GND), and signal lines were interconnected via 5 wires in total. Additionally, the systemic conceptual diagram is shown in Fig. 9. The simple HIL simulator was configured based on ”Arduino Due” that allowed for faster calculation than that of Arduino Mega 2560 used for a controller. The logic level for Arduino Mega 2560 differed from that of Arduino Due (Arduino Mega 2560: 5 V; Arduino Due: 3.3 V). Shield boards that allowed for conversion of a logic level were installed to the Arduino Due. Moreover, there were 3 buttons on the shield board. Pushing such buttons allowed for conversion of the motor condition within the HIL simulator. The motor con-
2019 IFAC ACE 210 1-3, 2016. Bratislava, Slovakia June
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
Fig. 11. Results of PID control via HILS
Fig. 8. Overview of the HILS System (Left Side: Controller; Right Side: HIL Simulator) Duty ratio
Controller Algorithm Angular velocity Current
Arduino Due (HIL Simulator)
Arduino Shield
D/A Converter 5V GND
A/D Converter Level Converter
A/D Converter
3.3V GND
D/A Converter
Duty ratio
Driver Motor & Load
State transition
Sensors Angular velocity Current
: Simulation model
Fig. 12. PID Control Results of the Motor Control System
3 Switches The Number of Respondents (%)
Arduino Mega 2560 (Controller)
Fig. 9. Conceptual Diagram for HILS System
100 80 60 40 20 0
Q1 1: Not at all
Q2
Q3
Q4
2: Slightly capable
Q5 Q6 Question Items
Q7
3: Nearly capable
Q8
Q9
Q10
4: Sufficiently capable
Fig. 13. Results of Questionnaires following Implementation of the Program
Fig. 10. Simulink Model for the HIL Simulator. dition could be described based on 3 conditions: (i) normal condition in which the current was flown based on the applied voltage of the motor, (ii) disconnection condition in which energization did not take place by the addition of the specified voltage, and (iii) degradation condition that caused the doubling of the internal resistance of the motor. The Simulink model for the simple HIL simulator is shown in Fig. 10. The motor condition could be converted through the use of Stateflow Toolbox of Simulink. According to the learning program, the learners conducted conversion of the HIL simulator condition at any time and verified whether or not the controller could operate normally. The HILS results based on PID controllers used for MILS during normal operation of the motor are shown in Fig. 11. The waveform is slightly unstable owing to the output resolution for the HIL simulator and influence of the noise between the interfaces. However, it is revealed that the HILS results were almost the same as those of MILS. 3.3 Actual-machine Experiment Control of the actual motor control system was achieved by using the controller whose safety was verified via HILS. In relation to HILS and an actual machine, there was no difference at the interface between the controller and 223
control target system in the case of HILS. Thus, the controller verified through HILS could be used for an actual machine. The control results of an actual machine with the same PID controllers as those used for MILS and HILS are shown in Fig. 12. Based on the information shown in Fig. 12, despite the fact that input values were influenced by the sensor noise, in relation to the output rotations, it is revealed that the results are almost equivalent to those obtained via MILS and HILS. 4. RESULTS OF THE QUESTIONNAIRES AND EDUCATIONAL EFFECTS In this section, educational effects from the learning program will be identified. The results of the questionnaires for the learners after implementation of the learning program are shown in Fig. 13. In relation to the items of the questionnaire surveys shown in TABLE 1; Q1, Q2, Q5, and Q7 represent the items pertaining to knowledge about MBD; Q4, Q6, and Q8 represent the items on skills necessary for MBD performance; Q3 and Q9 represent the items on understanding the connection between the questions and MBD, and Q10 represents the item on confidence gained through learning (self-efficacy). First, the results of the questionnaires prior to implementation of the program shown in Fig. 2 are discussed. Based on the results of answers regarding Q1 and Q2, the number of persons who answered that no explanations could be provided regarding MBD or the V-process accounted for less than 21% of the total (Q1: 32 persons
2019 IFAC ACE June 1-3, 2016. Bratislava, Slovakia
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
(20.9%) in maximum). Therefore, it can be inferred that the MBD method is recognized internally to some extent even when the learners were in charge of operations other than those relating to controller designing. Moreover, the answer distributions of Q1 and Q3 are remarkably similar. Therefore, even when some learners were beginners, it is revealed that recognition to the effect that simulation models were necessary for MBD has been gained to some extent. In contrast, in addition to specific explanations regarding MILS and HILS that make up the V-process, more than 60% of the learners gave the answer ”1” (Not at all) regarding the items on skills using MATLAB/Simulink (Q8: 142 persons (92.2%) in maximum). Given such a situation, despite the fact that the relevant learners had knowledge of MBD, they were not able to understand the entire picture of how MBD is performed because they have not had any experience in development in line with the V-process (or even when they are in charge of some operations in the V-process, they do not recognize so). Consequently, as they had little operational experience (or successful experience) with V-type processes, it can be assumed that their self-efficacy in being capable of carrying out MBD operations is low. Next, educational effects from the learning program are discussed based on a comparison of Figs. 2 and 13. In relation to the question items Q1 - Q8, less than 3% of the learners gave the answer ”1” regarding questions on using MATLAB/Simulink (Q8: 4 persons (2.63%)). Based on this result, high educational effects can be recognized. Additionally, in relation to Q9, despite the fact that no treatment was made through the learning program, the answer results are improved. Thus, it can be assumed that a knowledge transfer has been made based on the understanding of the content of the Vprocess. Moreover, based on the result of Q10, understanding of the content of learning and successful experience via exercises have led to confidence gained by the learners for their operational performance. Thus, educational effects from the learning program have been recognized. In the following section, the content of the learning process that remarkably contributed to improvement of self-efficacy by the learners is inferred based on analysis of the changes of the questionnaire survey results tallied for each lecture. 5. CONCLUSION The MBD education program was proposed to 159 professional engineers who belonged to Mazda Motor Corporation, and as such the program was implemented. Observation was made regarding educational effects from the program based on the results of the questionnaires for the learners. First, prior questionnaire surveys were given to the learners, and situations for acquisition of knowledge, understanding, skills, etc. regarding MBD were understood. The learning program was organized under the constraint of the schedule for program implementation. Furthermore, the motor control system and simple HIL simulator were developed for exercises and the resultants were used thereof. As a result of the questionnaire surveys following implementation of the program, exercises regarding the V-process significantly urged the learners to understand MBD. It has come to conclusion that successful experience in practical training led to improvement of selfefficacy for the future operational performance. We would 224
211
like to progress observation on a method for objective evaluation regarding skills of learners in the future. ACKNOWLEDGEMENT This work was supported by JSPS KAKENHI Grant Number JP17K12803. Mr. Takashi Kobayashi at Kobayashi Seisakujo Ltd. designed and manufactured the motor control system used for the learning program. Ms. Naomi Maegaki, a clerical assistant of this course provided cooperation on tallying the questionnaire survey results. We would like to express our deepest gratitude to them. REFERENCES Franco, F., Stella, G.D., Neme, J.H., Santos, M.M.D., Stevan, S.L., and Rosa, J.N.H. (2016). Teaching modelin-the loop: A case study for controller of distributed dashboard in a road vehicle. In 2016 IEEE 25th International Symposium on Industrial Electronics (ISIE), 863–868. Jaikamal, V. (2009). Model-based ecu development - an integrated mil-sil-hil approach. SAE Technical Paper. JMAAB (2008). JMAAB MBD Technology Skill Standards Version 1.1 (in Japanese). Japan MATLAB Automotive Advisory Board (JMAAB). Krasner, J. (2010). Comparing embedded design outcomes with and without model-based design. American Technology International, ʡ White Paper, 10. Lano, K., Yassipour-Tehrani, S., and Alfraihi, H. (2015). Experiences of teaching model-based development. CEUR Workshop Proceedings, 1555, 43–54. Larses, O. and N.Adamsson (2004). Drivers for model based development of mechatronic systems. Proceedings of DESIGN 2004, the 8th International Design Conference, 865–870. Ogata, H. and Katayama, T. (2009). Skill standards for model based development engineers in the automotive industry. IFAC Proceedings Volumes (IFACPapersOnline), 8(PART 1), 240–244. Ohata, A. and Kenneth, B.R. (2008). Improving modelbased design for automotive control systems development. Proc. 17th World Congr.. 2008, 1062–1065. Prabhu, S., Callanan, L., Chambers, Z., and Herniter, M. (2007). Development of model based design curriculum. ASEE Annual Conference and Exposition, Conference Proceedings. Reedy, J. and Lunzman, S. (2010). Model based design accelerates the development of mechanical locomotive controls. In SAE Technical Paper. SAE International. doi:10.4271/2010-01-1999. Singh, G. (2014). Experiential learning with respect to model based design applied to advanced vehicle development. URL http://hdl.handle.net/10012/8138. Vilanova, R. and Visioli, A. (2012). PID control in the Third Millennium : lessons learned and new approaches. Springer. Visioli, A. (2006). Practical PID control. Springer Science & Business Media. Yang, B. and Chen, C.Y. (2011). Development of a robotic platform for teaching model-based design techniques in dynamics and control program. ASEE Annual Conference and Exposition, Conference Proceedings.
ScienceDirect
IFAC PapersOnLine 52-9 (2019) 206–211
Practice and Evaluation of Model Based Practice and Evaluation of Model Based Practice and Evaluation of Model Based Development (MBD) Education Practice and Evaluation of Model Based Development (MBD) Education Development (MBD) Education Development (MBD) Education S. Wakitani ∗ T. Yamamoto ∗ C. Morishige ∗∗ T. Adachi ∗∗
∗ ∗ ∗∗ ∗∗ S. Yamamoto ∗ C. Morishige ∗∗ T. Adachi ∗∗ ∗∗ ∗ ∗∗ ∗∗ S. Wakitani Wakitani ∗∗∗ T. T. Yamamoto C.Muraoka Morishige T. Adachi Adachi ∗∗ Harada S. Wakitani T.Y. Yamamoto C. Morishige ∗∗ ∗T. ∗∗ ∗∗ T. Harada ∗∗ T. ∗∗ S. Wakitani T.Y. C.Muraoka Morishige T. Adachi ∗∗ ∗∗ Y. Harada T. Muraoka Y.Yamamoto Harada ∗∗ ∗∗ T. Muraoka ∗∗ Y. Harada T. Muraoka ∗ ∗ Graduate School of Engineering, Hiroshima University, Hiroshima, School of Engineering, Hiroshima University, Hiroshima, ∗ ∗ Graduate Graduate School of Engineering, Hiroshima 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). School of Engineering, Hiroshima University, University, Hiroshima, Hiroshima, ∗ Graduate 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). ∗∗ Graduate School of Engineering, Hiroshima Hiroshima, 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). Motor Corporation, Hiroshima,University, 730-8670 Japan 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). ∗∗ Mazda Mazda Motor Corporation, Hiroshima, 730-8670 Japan ∗∗ 739-8527 Japan (e-mail: (wakitani/yama)@hiroshima-u.ac.jp). ∗∗ Mazda Motor Corporation, Hiroshima, 730-8670 Japan ∗∗ Mazda Motor Corporation, Hiroshima, 730-8670 Japan Mazda Motor Corporation, Hiroshima, 730-8670 Japan Abstract: This study presents the practical results concerning model-based development Abstract: This study presents the practical results concerning model-based development Abstract: study presents results concerning model-based development educational This practices implemented by practical Hiroshima University and Mazda Motor Corporation, Abstract: This study presents the the practical results concerning model-based development educational practices implemented by Hiroshima University and Mazda Motor Corporation, Abstract: This study presents the practical results concerning model-based development educational practices implemented by Hiroshima University and Mazda Motor Corporation, an automobile manufacturer, and the consequent educational effects. Importance of modeleducational practices implemented by Hiroshima University and Mazda Motor Corporation, an automobile manufacturer, and the consequent educational effects. Importance of modelmodeleducational practices implemented by Hiroshima University and Mazda Motor Corporation, an automobile manufacturer, andinthe the consequent educational effects. Importance of based development is increasing industrial world. Model-based development is performed an automobile manufacturer, and consequent educational effects. Importance of modelbased development is increasing in industrial world. Model-based development is performed an automobile andin consequent educational effects. Importance of modelbased development is industrial world. Model-based development is according to a manufacturer, development process called the ”V-type development process (V-process).” based development is increasing increasing inthe industrial world. Model-based development is performed performed according to a development process called the ”V-type development process (V-process).” based development is increasing in industrial world. Model-based development is performed according to a development process called the ”V-type development process (V-process).” However, several professional engineers involved in automobile development have insufficient according to a development process called the in ”V-type development process (V-process).” However, several professional engineers involved automobile development have insufficient according to a development process called the in ”V-type development process (V-process).” However, several professional engineers involved in automobile development have insufficient experience in conducting model-based development. In this study, the authors propose a program However, several professional engineers involved automobile development have insufficient experience in conducting model-based development. In this study, the authors propose program However, several professional engineers involved in automobile development have insufficient experience inthese conducting model-based development. In this this study, the the authors propose aa program program that allowsin professional engineers to learn the V-process using a motor control system experience conducting model-based development. In study, authors propose a that allows these professional engineers to learn the V-process using a motor control system experience inthese conducting model-based development. In this study, the authors propose aengineers program that professional to the V-process using a motor control system via aallows hands-on approach. Theengineers program is learn implemented to target 159 professional that allows these professional engineers to learn the V-process using a motor control system via a hands-on approach. Theengineers program is implemented to target 159 professional engineers that professional to V-process using motor by control system via a hands-on approach. program is implemented to 159 belonging tothese Mazda Motor The Corporation. From thethe questionnaire surveys taken theengineers learners, via aallows hands-on approach. The program From is learn implemented to target target 159a professional professional engineers belonging to Mazda Motor Corporation. the questionnaire surveys taken by the learners, via a hands-on approach. The program is implemented to target 159 professional belonging to Mazda Mazda Motor Corporation. From the questionnaire questionnaire surveys taken by the theengineers learners, it is revealed that the V-process exercises considerably contribute to the self-efficacy of the belonging to Motor Corporation. From the surveys taken by learners, it is is revealed revealed that the the V-process exercises considerably contribute to the the self-efficacy of the the belonging Mazda Motor Corporation. From the questionnaire surveys taken by the learners, it that V-process exercises considerably contribute to self-efficacy of learners fortooperational performance using model-based development. it is revealed that the V-process exercises considerably contribute to the self-efficacy of the learners for operational performance using model-based development. it is revealed that the V-process exercises considerablydevelopment. contribute to the self-efficacy of the learners for performance using learners for operational operational performance using model-based model-based development. © 2019, IFAC (International Federation ofusing Automatic Control) Hosting by Elsevier Ltd. All rights reserved. learners for operational performance model-based development. Keywords: Curriculum development; engineering education; model-based development; v-type Keywords: Curriculum development; engineering education; model-based development; v-type Keywords: Curriculum Curriculum development; engineering engineering education; education; model-based model-based development; development; v-type v-type development process development; Keywords: development process Keywords: Curriculum development process development; engineering education; model-based development; v-type development process development process 1. relevant systems are modeled, the modeled components 1. INTRODUCTION INTRODUCTION relevant systems systems are are modeled, modeled, the the modeled modeled components components 1. relevant are integrated. Accordingly, analysis of the components behavior of 1. INTRODUCTION INTRODUCTION relevant systems are modeled, the modeled are integrated. Accordingly, analysis of the behavior of 1. INTRODUCTION relevant systems are modeled, thedesign, modeled are integrated. analysis of the behavior of the system and Accordingly, controlled system arecomponents performed. are integrated. Accordingly, analysis of the behavior of Customer needs for products have become more diversified the system and controlled system design, are performed. are integrated. Accordingly, analysis of the behavior of the system and controlled system design, are performed. In HILS, controller algorithms designed through MILS are Customer needs for products have become more diversified the system and controlled system design, are performed. Customer needs for products have become more diversified and complex over the years.have Companies must diversified promptly In HILS, controller algorithms designed through MILS are Customer needs for products become more the system and controlled system design, are performed. In HILS, algorithms designed through are implemented for actual electronic control unitsMILS (ECUs), and complex over the years. Companies must promptly In HILS, controller controller algorithms designed through MILS are Customer for products have become more and complex over the years. Companies must promptly respond toneeds such demands. However, in regard todiversified intricate implemented for actual electronic control units (ECUs), and complex over the years. Companies mustto promptly In HILS, controller algorithms designed through MILS are implemented for actual electronic control units (ECUs), and their robustness and safety are verified via real-time respond to such demands. However, in regard intricate implemented for actual electronic control units (ECUs), and complex over the years. Companies must promptly respond to such demands. However, in to systems such as automobiles, countless combinations of and their robustness and safety are verified via real-time respond to such demands. However, in regard regard to intricate intricate implemented for actual electronic control units (ECUs), and their robustness and safety are verified via real-time systems such as automobiles, countless combinations of simulations. However,and in relation to verified using control targets and their robustness safety are via real-time respond such However, in regard toare intricate systems such as automobiles, countless combinations of optimumto components and adjustable parameters possiHowever, in relation to using control targets systems such as demands. automobiles, countless combinations of simulations. and their robustness and safety are verified via real-time simulations. However, in relation to using control targets for verification, a computer (i.e., hardware in the loop optimum components and adjustable parameters are possisimulations. However, in relation to using control targets systems such as automobiles, countless combinations of optimum components and adjustable parameters are possiin the loop ble. Long components times and high costs are required for formulating verification, a computer (i.e., optimum andcosts adjustable parameters are possi- for simulations. However, incan relation tohardware using control targets for verification, a computer (i.e., hardware in the (HIL) simulator) that mimic the behaviors of the ble. Long times and high are required for formulating for verification, a computer (i.e., hardware in the loop optimum components andcosts adjustable parameters are possi- (HIL) simulator) that can mimic the behaviors of loop ble. Long times and high are required for formulating designs and adjustments using actual samples. Moreover, the ble. Long times and high costs are required for formulating for verification, a computer (i.e., hardware in the loop (HIL) simulator) that can mimic the behaviors of the control targets is employed. Consequently, it has become designs and adjustments using actual samples. Moreover, (HIL) simulator) that can mimic the behaviors of the ble. Long times and high costs are required for formulating designs and adjustments using actual samples. the cost and time required for manufacturing must be targets is employed. Consequently, it has become designs using samples. Moreover, Moreover, (HIL) simulator) that can mimic the of the control targets employed. Consequently, it the cost costand andadjustments time required required foractual manufacturing must be be control unnecessary to is produce prototypes for behaviors controller verificontrol targets isproduce employed. Consequently, it has has become become designs and adjustments using actual samples. the time for manufacturing reduced toand the maximum extent possible owing Moreover, tomust the inunnecessary to prototypes for controller verifithe cost and time required for manufacturing must be control targets is employed. Consequently, it has become unnecessary to produce prototypes for controller verification. Thus, the time and costs incurred for controller reduced to the maximum extent possible owing to the intothe produce prototypes for controller verifithe required forcurrently. manufacturing be unnecessary reduced to the maximum extent possible owing to the incation. Thus, time and costs incurred for controller tensecost international competition Given this backreduced toand thetime maximum extent possible owing this tomust the inunnecessary tothe produce prototypes for controller verification. Thus, and costs incurred for controller development can betime substantially reduced. The automotive tense international competition currently. Given backcation. Thus,can the time and costsreduced. incurred forautomotive controller reduced to the maximum extent possible owing to the intense international competition currently. Given this background, importance of model-based development (MBD) development be substantially The tense international competition currently. Given this backcation. Thus, the time and costs incurred for controller development can be substantially reduced. The automotive industry has can proactively adopted reduced. MBD forThe product develground, importance of model-based development (MBD) development be substantially automotive tense international competition currently. Given this back- industry ground, importance of development (MBD) (Larses and N.Adamsson (2004); Ohata and Kenneth has proactively adopted MBD for product develground, importance of model-based model-based development (MBD) development can bethere substantially reduced. The automotive industry has proactively adopted MBD for product devel(Larses and N.Adamsson (2004); Ohata and Kenneth opment. However, are only few professional engineers industry has proactively adopted MBD for product develground, importance of model-based development (MBD) (Larses and (2004); Ohata and (2008); Ogata and Katayama (2009); Jaikamal (2009); opment. However, there are only few professional engineers (Larses and N.Adamsson N.Adamsson (2004); Ohata and Kenneth Kenneth industry has proactively adopted MBD for product development. However, there only professional engineers involved in automobile development having sufficient ex(2008); Ogata and Katayama (2009); Jaikamal (2009); opment. there are are only few few professional engineers (Larses and N.Adamsson Ohata and Kenneth (2008); Ogata and Katayama (2009); (2009); involved However, in automobile development having sufficient exSingh (2014); Franco et al. (2004); (2016)) has Jaikamal increased. Thus, involved (2008); Ogata and Katayama (2009); Jaikamal (2009); opment. However, there are only few professional engineers in automobile development having sufficient experience for performing MBD. Singh (2014); Franco et al. (2016)) has increased. Thus, involved in automobile development having sufficient ex(2008); Ogata and Katayama (2009); Jaikamal (2009); Singh (2014); Franco et al. (2016)) has increased. Thus, for performing MBD. in recent years, difficult systems have been implemented perience Singh (2014); Franco et systems al. (2016)) hasbeen increased. Thus, perience involved for in automobile development having sufficient experforming MBD. in recent years, difficult have implemented for performing MBD. Singh et systems al.this (2016)) hasbeen increased. Thus, perience in years, difficult have implemented on recent the(2014); computer through methodology, and system The ʠ Hirojiren ʡ (i.e., Hiroshima Council for the Promoin recent years,Franco difficult systems have been implemented perience for performing MBD. The ʠ Hirojiren ʡ (i.e., Hiroshima Council the on the computer through this methodology, and system in recent years, difficult systems have beenApplication implemented on the computer through this methodology, and system The ʠ Hirojiren ʡ (i.e., (i.e., Hiroshima Council for for the PromoPromodesign and verification can be performed. of tion of Collaboration between Government, Academia, and on the computer through this methodology, and system The ʠ Hirojiren ʡ Hiroshima Council for the Promotion of Collaboration between Government, Academia, and design and verification can be performed. Application of on thehas computer through this methodology, and The ʠ Hirojiren ʡ (i.e., Hiroshima Council for the design and verification can be Application of tion of Collaboration between Government, Academia, MBD rapidly progressed inperformed. the industry (e.g. system Reedy Automobile Industry) was established in June 2015Promoin and Hidesign and verification can be performed. Application of tion of Collaboration between Government, Academia, and MBD has rapidly progressed in the industry (e.g. Reedy Automobile Industry) was established in June 2015 in Hidesign and verification can be performed. Application of of Collaboration between Government, Academia, and MBD has progressed in industry (e.g. Reedy Automobile Industry) was established in June 2015 in Hiand Lunzman (2010)). In MBD, in contrast to convenroshima Prefecture in regard to the ”Industry-AcademiaMBD has rapidly rapidly progressed in the the industry (e.g. Reedy tion Automobile Industry) was established in June 2015 in Hiand Lunzman (2010)). In MBD, in contrast to convenroshima Prefecture in regard to the ”Industry-AcademiaMBD has rapidly progressed in the industry (e.g. Reedy Automobile Industry) was established in June 2015 in Hiand (2010)). MBD, in to roshima Prefecture Prefecture in regard regardVision to the the2030.” ”Industry-AcademiationalLunzman source codes, each In component iscontrast defined as a convensimula- roshima Government Collaborative The Hirojiren, and Lunzman (2010)). In MBD, in contrast to convenin to ”Industry-AcademiaGovernment Collaborative Vision 2030.” The Hirojiren, tional source codes, each component is defined as a simulaand Lunzman (2010)). In MBD, in contrast to convenroshima Prefecture in regard to the ”Industry-Academiational source codes, each component is defined as a simulaGovernmentofCollaborative Collaborative Vision 2030.” The Hirojiren, Hirojiren, tion block (i.e., executable specification), and information comprising three committees and four expert committional source codes, each component is defined as a simula- Government Vision 2030.” The comprising of three committees and four expert committion block (i.e., executable specification), and information tional source each component as a simulaGovernment Collaborative Vision 2030.” The committee Hirojiren, tion block (i.e., executable specification), information of three committees and four expert commitregarding thecodes, relationships betweenis defined theand components is comprising tees, undertakes relevant activities. The expert tion block (i.e., executable specification), and information comprising of three committees and four expert committees, undertakes relevant activities. The expert committee regarding the relationships between the components is tion block (i.e., executable specification), and information comprising of three committees and four expert commitregarding the relationships between the components is tees, undertakes relevant activities. The expert committee described via a block diagram. Thus, it is possible to verify for MBD established ”Fundamentals of MBD Laboratory regarding the relationships between the components is tees, undertakes relevant activities. The expert committee described via a block diagram. Thus, it is possible to verify for MBD established ”Fundamentals of MBD Laboratory regarding the relationships between the components is tees, undertakes relevant activities. The expert committee described via diagram. Thus, possible to for MBD MBD established ”Fundamentals ofinMBD MBD Laboratory various operations using the computer. MBD is executed (MBD lab)” in Hiroshima Universityof 2016,Laboratory Hiroshima described via aa block block diagram. Thus, it it is isMBD possible to verify verify for established ”Fundamentals various operations using the computer. is executed lab)” in Hiroshima University 2016, Hiroshima described via the a block diagram. Thus, it is possible to for MBD established ”Fundamentals ofin Laboratory various operations using the computer. MBD is (MBD lab)” in Mazda Hiroshima University inMBD 2016, Hiroshima in line with development process known as executed theverify ”V- (MBD University and Motor Corporation, the Japanese various operations using the computer. MBD is executed (MBD lab)” in Hiroshima University in 2016, Hiroshima in line with the development process known as the ”VUniversity and Mazda Motor Corporation, the Japanese various operations using the(V-process).” computer. MBD executed (MBD lab)”and in Mazda Hiroshima University in 2016, Hiroshima in line with the development process known as the ”VUniversity Motor Corporation, the Japanese type development process Theis V-process automobile manufacturer, played a central role in the in line with the development process known as the ”VUniversity and Mazda Motor Corporation, the Japanese type development process (V-process).” The V-process automobile manufacturer, played a central role in the in line with the development process known as the ”VUniversity and Mazda Motor Corporation, the Japanese type development process (V-process).” The V-process automobile manufacturer, played a central role in the comprises the following operations for which computer establishment of this course. In relation to the MBD type development process (V-process).” The V-process automobile manufacturer, played a central role inLab., the comprises the following operations for which computer establishment of this course. In relation to the MBD Lab., type development process (V-process).” The V-process automobile manufacturer, played a central role in the comprises operations for which computer establishment of ofof this course. InMBD relation to the the MBD MBD Lab., simulationsthe playfollowing an important role: (i) model in the loop implementation an effective education program is comprises the following operations for which computer establishment this course. In relation to Lab., role: (i) model in the loop simulations play an important implementation of an effective MBD education program is comprises the following operations for which computer establishment of this course. In relation to the MBD Lab., simulations play an important role: (i) model in the loop implementation of an an effective effective MBD(specifically education program program is simulation (MILS) and (ii) hardware in the loopin simulation required for professional engineers for those simulations play an important role: (i) model the loop implementation of MBD education is required for professional engineers (specifically for those simulation (MILS) and (ii) hardware in the loop simulation simulations play anafter important role: (i)the model in the loop implementation of an effective MBD(specifically education program is simulation (MILS) and (ii) hardware in loop simulation required for professional engineers for those (HILS). In MILS, all the components comprising of simulation (MILS) and (ii) hardware in the loop simulation required for professional engineers (specifically for those In MILS, after all the components comprising of (HILS). simulation and (ii) in the loop simulation (HILS). In (MILS) MILS, after after all hardware the components components comprising of required for professional engineers (specifically for those (HILS). In MILS, all the comprising of (HILS). MILS, after all the components of Hosting by Elsevier Ltd. All rights reserved. 2405-8963In © 2019, IFAC (International Federation of comprising Automatic Control)
Copyright © 2019 IFAC 219 Copyright 2019 IFAC 219 Control. Peer review© under responsibility of International Federation of Automatic Copyright © 219 Copyright © 2019 2019 IFAC IFAC 219 10.1016/j.ifacol.2019.08.197 Copyright © 2019 IFAC 219
2019 IFAC ACE June 1-3, 2016. Bratislava, Slovakia
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
who are beginners, without sufficient experience in development using MBD). Professional engineers undertake the learning program in parallel with performing their work. Thus, a program implementation period of 4 days (i.e., 6 h a day) has been set. An MBD education program that would allow for educational effects with the learners being beginners and the implementation period as constraints was proposed for the MBD lab. Such a program was implemented for 159 people for 4 months. According to the above-mentioned program, control experiments based on physical modeling and motor control systems, as well as basic exercises using MATLAB/Simulink were performed. Moreover, in order to learn HILS via a hands-on approach, a simple hardware in the loop (HIL) simulator using Arduino was developed, which was then utilized for practical training. The questionnaire surveys implemented for each lecture revealed that several learners gained confidence for operational performance via MBD before and after participating in the learning program. In this study, changes of the educational effects on the learners will also be analyzed based on the changes in the results of the questionnaires. The structure of this paper is as follows. In Section 2, an outline of the V-process is first described, and explanations of the results of prior questionnaire surveys performed with targeted learners will be provided. Subsequently, the proposed MBD education program is presented. In Section 3, the motor control system, which is necessary for exercises for the learning program and the simulation models thereof, is explained and the simple HIL simulator using Arduino will be provided. In Section 4, educational effects concerning the learning program and changes of selfefficacy of the learners are observed. Finally, a summary of this work and future issues are described. 2. MODEL-BASED DEVELOPMENT LEARNING PROGRAM 2.1 Model-based Development MBD focuses on visibility compared with conventional development methods based on source codes. Thus, various analyses and verifications can be performed through relevant models. Therefore, it is possible to discover serious systemic defects at an early stage. As a result, the development efficiency can be improved. Case examples in which the development efficiency could be reduced more than 30% based on MBD adoption have been reported by Krasner (2010). MBD is performed in line with the development process named, the V-process. The V-process varies depending upon the number of items according to the degree of details for the descriptions. Briefly, this is described as per Fig. 1. According to MILS, first the requirements for an intended system operations are determined. The combination of components and parameters necessary to satisfy the aforementioned requirements is verified and designed via simulation using a simulation block. In such a case, feasibility for intended specifications may be also reviewed and modifications may be added. According to HILS, controller algorithms designed through MILS are implemented for actual ECUs, and operation verifications are performed by using a HIL simulator. After the aforementioned designing and verification, a modelbased calibration (MBC) that allows for satisfaction of 220
207
Fig. 1. V-Type Development Process. various requirements relating to mass production, such as vehicle performance and law adaptation, is performed for the combined use of actual components and models. The final products are completed after relevant products are tested. In relation to the operations described on the left-hand side of Fig. 1 in the V-process, no verification using actual prototypes is performed. Additionally, at this point, prior extraction and treatment of almost all fatal errors including human errors in product development, are conducted. Therefore, it is possible to prevent unnecessary rework in the operations described on the right-hand side of Fig. 1. Consequently, new product development is possible in a short period. In regard to the learning program, MILS and HILS which are particularly important operations described on the left-hand side of Fig. 1, the V-process will be primarily examined. 2.2 Learners 159 professional engineers of Mazda Motor Corporation undertook the basic course of MBD. The program was implemented for 4 classes (i.e., a single class comprised 40 professional engineers (and 39 professional engineers for some classes)). The learners undertook the learning program concurrently with their operations. Therefore, a 4-day (i.e., 6 h a day and 2 days a week) implementation period for the program was set in advance. Many learners were primarily responsible for operations for vehicle design, vehicle equipment, electrical equipment, and vehicle analysis, instead of development of the control algorithm. Therefore, it could be inferred that such learners were not very familiar with the MBD development process accompanying the controller development. The questionnaire survey shown in TABLE I was provided to the learners prior to implementation of the learning program in order to evaluate their knowledge and predict the degree of understanding of MBD as well as the degree of motivation for MBD operational performance. In relation to all the question items, the learners selected the correct answer from 4 choices of ”1: Not at all,” ”2: Slightly capable,” ”3: Nearly capable,” or ”4: Sufficiently capable” using a 4-point scale. Fig. 2 shows the distribution of responses by the learners prior to implementation of the program. It is observed that although several learners had knowledge about MBD to some extent, they did not have skills or were not confident in the performance of MBD (the details thereof will be discussed in Section 4). Moreover, questionnaire surveys with the same content as that mentioned above were implemented on all 4 days of the lectures. The changes thereof were surveyed as well. The survey results will be discussed in Section 4.
2019 IFAC ACE 208 1-3, 2016. Bratislava, Slovakia June
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
Table 1. QUESTIONNAIRE SURVEY ITEMS. Question Number Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8
The Number of Respondents (%)
Q9 Q10
Question
Evaluation Item
Are you capable of explaining MBD (i.e., Model-Based Development)? Are you capable of explaining the V-Type Development Process? Are you capable of explaining the significance of the modelling? Are you capable of modelling with the use of these software, such as Matlab and Simulink? Are you capable of explaining MILS (Model in the Loop Simulation)? Are you capable of explaining simulation based on MILS through the use of software, such as Matlab and Simulink? Are you capable of explaining HILS (Hardware in the Loop Simulation)? Are you capable of performing a simulation based on HILS by the using a software, such as Matlab and Simulink? Are you capable of explaining the necessity for MBC (Model-Based Calibration)? Are you capable of explaining the operational performance based on the v-type development process or are you confident in the performance?
Knowledge Knowledge Understanding Skills
100 80 60 40 20 0
Q1 1: Not at all
Q2
Q3
Q4
2: Slightly capable
Q5 Q6 Question Items 3: Nearly capable
Q7
Q8
Q9
Q10
4: Sufficiently capable
Fig. 2. Questionnaires Prior to Implementation of the Program. 2.3 Organization of the Learning Program According to the educational program, the following educational objectives have to be achieved: (i) the learners understand the positioning of their own duties relating to MBD after implementation of the program, (ii) they will become interested in proactive use of MBD, and (iii) they will become confident in implementing MBD. In the light of the outline of MBD and the situation of the learners described above, the program is organized as per TABLE 2. On the first half of the 1st day, overview of MBD was provided, which motivated the learners to learn. In the second half of the 1st day, exercises using Simulink allowed the learners to understand the basic method for using MATLAB/Simulink. On the 2nd day, physical modeling was performed based on differential equations for a basic learning of plant modeling, which was inevitable for MBD. Learning regarding the Laplace transform and transfer function as a separate method for expression of plant models was performed. The learners understood that there existed various methods for plant expressions. Moreover, establishment and execution of the aforementioned models via MATLAB/Simulink could lead to the effect of repetitive learning with the methods available with the software. On the 3rd day, application of the knowledge gained thus far allowed modeling of the motor control system. Explanations of the details of the motor control system will be provided in the following section. In here, the entire motor (and the load), sensor, actuator, interface (AD/DA converters), and control logic (Proportional-Integral-Derivative (PID) Controller (Visioli (2006); Vilanova and Visioli (2012))) are described as simulation blocks. Thereby, simulation could be conducted. Moreover, the learners adjusted PID parameters 221
Knowledge Skills Knowledge Skills Understanding Self-efficacy
included in the control logics so that such parameters could satisfy the requirements specifications (i.e., settling time and overshoot) defined by the learners. On the 4th day, real-time simulation was conducted using a simple HIL simulator. Explanations regarding the details of the simple HIL simulator will be provided in the following section. Through this process, the learners were able to discover failures (i.e., response to noises and occurrence of winding-up) that they could not notice through parameter adjustments for MILS anew, and readjust the parameters. In this way, the learners were able to learn that a method for development in line with the V-process could resolve problems relating to failures urgently through a handson approach. Furthermore, the learners implemented controllers obtained through the HILS operations as controllers for experimental devices, and found that results equivalent to those regarding HILS could be obtained. Thereby, the learners were able to achieve success using the V-process. 3. LEARNING MATERIALS Fig. 3 shows the external appearance of the motor control system for the experiment, which was used for exercises on the 3rd and 4th days. Three aluminum disks having diameters different from those of the pulley were connected to the motor shaft. Such disks were connected by a pulley (connected to the dynamo-shaft) and belt. Moreover, the Arduino Mega 2560, motor driver, tachometer, and current sensor were mounted integral to the housing. In the exercises, controlling of rotational speed was set as an assignment. Control parameter adjustment and verification of the controller safety performance were performed via MILS and HILS. 3.1 Model in the Loop Simulation The block diagram of the motor control system is shown in Fig. 4. It allows the learners to understand the relationship between all the components (i.e., blocks). Next, mathematical models corresponding to the aforementioned blocks were obtained and Simulink blocks were constructed. Mathematic models for all the components within Fig. 4 were obtained and Simulink blocks were designed. Consequently, the Simulink model corresponding to the system shown in Fig. 4 was acquired as per Fig. 5. Here, the
2019 IFAC ACE June 1-3, 2016. Bratislava, Slovakia
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
209
Table 2. MBD EDUCATION PROGRAM Day 1 Day 2 Day 3 Day 4
MBD overview and MATLAB/Simulink exercises First-principle modeling (examples of the spring mass damper system) Basics of Laplace transform (first-order lag system) V-Type development process exercises (MILS) V-Type development process exercises (HILS and actual-machine experiment)
Fig. 6. Results of PID control via MILS Fig. 3. Motor Control System (on the Load Side)
Fig. 4. Block Diagram regarding the Motor Control System Fig. 7. Simulink model for controller via HILS via MILS were extracted as per Fig. 7. Direct download to Arduino Mega 2560 took place through the use of a MATLAB/Simulink add-on.
Fig. 5. Simulink model for the motor control system ”Scope Real” block within Fig. 5 directly indicates the signals from the motor. On the other hand, in relation to ”Scope Measure,” only the signals measured inside the computer are indicated. Additionally, in relation to an Arduino block, the PID control law represented by the following formula is implemented. t 1 de(t) u(t) = kc e(t) + (1) e(τ )dτ + TI 0 dt In relation to the PID controller, the duty ratio (u(t) [%]) for the motor driver has been determined to track the target number of rotations (r(t) [rpm]). An example of control results regarding the motor control system is shown in Fig. 6. In Fig. 6, ”N ∗ ” represents the measured rotations [rpm], ”i∗ ” is the measured current [A], and ”duty” is the duty ratio [%]. In the exercises, the learners designed PID parameters via simulations that satisfied the control specifications established by the learners. 3.2 Hardware in the Loop Simulation In here, a PID controller designed via MILS was mounted to Arduino Mega 2560 as a type of micro controllers and operation verifications thereof were performed. Only the Simulink models of the interface and controller section 222
The computer containing the interface revealed that the HIL simulator was used as a control target (plant) of the controller instead of actual prototypes via HILS. The HIL simulator received electrical signals that were output from the controller as digital data via the interface, and computed behaviors regarding plant models inside the simulator based on such received signals. After calculation of such behaviors, electrical signals corresponding to sensor outputs were output from the HIL simulator. In relation to the HIL simulator, in addition to normal systemic behaviors, a realistic situation accompanying dangers could be reproduced on the computer. Thus, it was possible to verify the controller performance under various conditions. Here, the simple HIL simulator shown in Fig. 8 exclusively developed for the learning program is used. The left-hand side section in Fig. 8 represents the controller and the right-hand side section the HIL simulator. The power, ground (GND), and signal lines were interconnected via 5 wires in total. Additionally, the systemic conceptual diagram is shown in Fig. 9. The simple HIL simulator was configured based on ”Arduino Due” that allowed for faster calculation than that of Arduino Mega 2560 used for a controller. The logic level for Arduino Mega 2560 differed from that of Arduino Due (Arduino Mega 2560: 5 V; Arduino Due: 3.3 V). Shield boards that allowed for conversion of a logic level were installed to the Arduino Due. Moreover, there were 3 buttons on the shield board. Pushing such buttons allowed for conversion of the motor condition within the HIL simulator. The motor con-
2019 IFAC ACE 210 1-3, 2016. Bratislava, Slovakia June
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
Fig. 11. Results of PID control via HILS
Fig. 8. Overview of the HILS System (Left Side: Controller; Right Side: HIL Simulator) Duty ratio
Controller Algorithm Angular velocity Current
Arduino Due (HIL Simulator)
Arduino Shield
D/A Converter 5V GND
A/D Converter Level Converter
A/D Converter
3.3V GND
D/A Converter
Duty ratio
Driver Motor & Load
State transition
Sensors Angular velocity Current
: Simulation model
Fig. 12. PID Control Results of the Motor Control System
3 Switches The Number of Respondents (%)
Arduino Mega 2560 (Controller)
Fig. 9. Conceptual Diagram for HILS System
100 80 60 40 20 0
Q1 1: Not at all
Q2
Q3
Q4
2: Slightly capable
Q5 Q6 Question Items
Q7
3: Nearly capable
Q8
Q9
Q10
4: Sufficiently capable
Fig. 13. Results of Questionnaires following Implementation of the Program
Fig. 10. Simulink Model for the HIL Simulator. dition could be described based on 3 conditions: (i) normal condition in which the current was flown based on the applied voltage of the motor, (ii) disconnection condition in which energization did not take place by the addition of the specified voltage, and (iii) degradation condition that caused the doubling of the internal resistance of the motor. The Simulink model for the simple HIL simulator is shown in Fig. 10. The motor condition could be converted through the use of Stateflow Toolbox of Simulink. According to the learning program, the learners conducted conversion of the HIL simulator condition at any time and verified whether or not the controller could operate normally. The HILS results based on PID controllers used for MILS during normal operation of the motor are shown in Fig. 11. The waveform is slightly unstable owing to the output resolution for the HIL simulator and influence of the noise between the interfaces. However, it is revealed that the HILS results were almost the same as those of MILS. 3.3 Actual-machine Experiment Control of the actual motor control system was achieved by using the controller whose safety was verified via HILS. In relation to HILS and an actual machine, there was no difference at the interface between the controller and 223
control target system in the case of HILS. Thus, the controller verified through HILS could be used for an actual machine. The control results of an actual machine with the same PID controllers as those used for MILS and HILS are shown in Fig. 12. Based on the information shown in Fig. 12, despite the fact that input values were influenced by the sensor noise, in relation to the output rotations, it is revealed that the results are almost equivalent to those obtained via MILS and HILS. 4. RESULTS OF THE QUESTIONNAIRES AND EDUCATIONAL EFFECTS In this section, educational effects from the learning program will be identified. The results of the questionnaires for the learners after implementation of the learning program are shown in Fig. 13. In relation to the items of the questionnaire surveys shown in TABLE 1; Q1, Q2, Q5, and Q7 represent the items pertaining to knowledge about MBD; Q4, Q6, and Q8 represent the items on skills necessary for MBD performance; Q3 and Q9 represent the items on understanding the connection between the questions and MBD, and Q10 represents the item on confidence gained through learning (self-efficacy). First, the results of the questionnaires prior to implementation of the program shown in Fig. 2 are discussed. Based on the results of answers regarding Q1 and Q2, the number of persons who answered that no explanations could be provided regarding MBD or the V-process accounted for less than 21% of the total (Q1: 32 persons
2019 IFAC ACE June 1-3, 2016. Bratislava, Slovakia
S. Wakitani et al. / IFAC PapersOnLine 52-9 (2019) 206–211
(20.9%) in maximum). Therefore, it can be inferred that the MBD method is recognized internally to some extent even when the learners were in charge of operations other than those relating to controller designing. Moreover, the answer distributions of Q1 and Q3 are remarkably similar. Therefore, even when some learners were beginners, it is revealed that recognition to the effect that simulation models were necessary for MBD has been gained to some extent. In contrast, in addition to specific explanations regarding MILS and HILS that make up the V-process, more than 60% of the learners gave the answer ”1” (Not at all) regarding the items on skills using MATLAB/Simulink (Q8: 142 persons (92.2%) in maximum). Given such a situation, despite the fact that the relevant learners had knowledge of MBD, they were not able to understand the entire picture of how MBD is performed because they have not had any experience in development in line with the V-process (or even when they are in charge of some operations in the V-process, they do not recognize so). Consequently, as they had little operational experience (or successful experience) with V-type processes, it can be assumed that their self-efficacy in being capable of carrying out MBD operations is low. Next, educational effects from the learning program are discussed based on a comparison of Figs. 2 and 13. In relation to the question items Q1 - Q8, less than 3% of the learners gave the answer ”1” regarding questions on using MATLAB/Simulink (Q8: 4 persons (2.63%)). Based on this result, high educational effects can be recognized. Additionally, in relation to Q9, despite the fact that no treatment was made through the learning program, the answer results are improved. Thus, it can be assumed that a knowledge transfer has been made based on the understanding of the content of the Vprocess. Moreover, based on the result of Q10, understanding of the content of learning and successful experience via exercises have led to confidence gained by the learners for their operational performance. Thus, educational effects from the learning program have been recognized. In the following section, the content of the learning process that remarkably contributed to improvement of self-efficacy by the learners is inferred based on analysis of the changes of the questionnaire survey results tallied for each lecture. 5. CONCLUSION The MBD education program was proposed to 159 professional engineers who belonged to Mazda Motor Corporation, and as such the program was implemented. Observation was made regarding educational effects from the program based on the results of the questionnaires for the learners. First, prior questionnaire surveys were given to the learners, and situations for acquisition of knowledge, understanding, skills, etc. regarding MBD were understood. The learning program was organized under the constraint of the schedule for program implementation. Furthermore, the motor control system and simple HIL simulator were developed for exercises and the resultants were used thereof. As a result of the questionnaire surveys following implementation of the program, exercises regarding the V-process significantly urged the learners to understand MBD. It has come to conclusion that successful experience in practical training led to improvement of selfefficacy for the future operational performance. We would 224
211
like to progress observation on a method for objective evaluation regarding skills of learners in the future. ACKNOWLEDGEMENT This work was supported by JSPS KAKENHI Grant Number JP17K12803. Mr. Takashi Kobayashi at Kobayashi Seisakujo Ltd. designed and manufactured the motor control system used for the learning program. Ms. Naomi Maegaki, a clerical assistant of this course provided cooperation on tallying the questionnaire survey results. We would like to express our deepest gratitude to them. REFERENCES Franco, F., Stella, G.D., Neme, J.H., Santos, M.M.D., Stevan, S.L., and Rosa, J.N.H. (2016). Teaching modelin-the loop: A case study for controller of distributed dashboard in a road vehicle. In 2016 IEEE 25th International Symposium on Industrial Electronics (ISIE), 863–868. Jaikamal, V. (2009). Model-based ecu development - an integrated mil-sil-hil approach. SAE Technical Paper. JMAAB (2008). JMAAB MBD Technology Skill Standards Version 1.1 (in Japanese). Japan MATLAB Automotive Advisory Board (JMAAB). Krasner, J. (2010). Comparing embedded design outcomes with and without model-based design. American Technology International, ʡ White Paper, 10. Lano, K., Yassipour-Tehrani, S., and Alfraihi, H. (2015). Experiences of teaching model-based development. CEUR Workshop Proceedings, 1555, 43–54. Larses, O. and N.Adamsson (2004). Drivers for model based development of mechatronic systems. Proceedings of DESIGN 2004, the 8th International Design Conference, 865–870. Ogata, H. and Katayama, T. (2009). Skill standards for model based development engineers in the automotive industry. IFAC Proceedings Volumes (IFACPapersOnline), 8(PART 1), 240–244. Ohata, A. and Kenneth, B.R. (2008). Improving modelbased design for automotive control systems development. Proc. 17th World Congr.. 2008, 1062–1065. Prabhu, S., Callanan, L., Chambers, Z., and Herniter, M. (2007). Development of model based design curriculum. ASEE Annual Conference and Exposition, Conference Proceedings. Reedy, J. and Lunzman, S. (2010). Model based design accelerates the development of mechanical locomotive controls. In SAE Technical Paper. SAE International. doi:10.4271/2010-01-1999. Singh, G. (2014). Experiential learning with respect to model based design applied to advanced vehicle development. URL http://hdl.handle.net/10012/8138. Vilanova, R. and Visioli, A. (2012). PID control in the Third Millennium : lessons learned and new approaches. Springer. Visioli, A. (2006). Practical PID control. Springer Science & Business Media. Yang, B. and Chen, C.Y. (2011). Development of a robotic platform for teaching model-based design techniques in dynamics and control program. ASEE Annual Conference and Exposition, Conference Proceedings.