Product design logic for an intelligent product modelling system

Product design logic for an intelligent product modelling system

Robottcs & Computer-Integrated Manufactunng, Vol 4, No 3/4, pp 499-510, 1988 0736-5845/8853 00 + 0 00 © 1988 Pergamon Press plc Pnnted m Great Brita...

683KB Sizes 2 Downloads 167 Views

Robottcs & Computer-Integrated Manufactunng, Vol 4, No 3/4, pp 499-510, 1988

0736-5845/8853 00 + 0 00 © 1988 Pergamon Press plc

Pnnted m Great Britain

• Paper

PRODUCT

DESIGN LOGIC FOR AN INTELLIGENT SYSTEM

PRODUCT

MODELLING

EIJI A R A I * and K A Z U A K I I W A T A t *Department of Precision Engineering, Shizuoka University, Hamamatsu, 432 Japan and i'Department of Production Engineering, Kobe University, Kobe, 657 Japan Recent research in CAD systems has been conducted to realize intelligent processing. Several CAD systems and product modelling systems have been developed using AI techniques. However, in order to develop more intelligent CAD systems, the design logic which connects the functional requirement to the geometric and the technological information of the designed product must be evaluated. A product model used in such intelligent CAD systems has to include not only the geometric and the technological information of the product but also the designer's thought process which explains the design logic. Design logic is generally divided into two parts. One is the generalized design logic which is commonly used in the conceptual design of mechanical products. The other is the product specified design logic which is used in the fundamental and detailed design phase. Different logic is applied to each product. This type of design logic is often used in modification design and compilation design, where the dimensions of parts have to be modified according to different functional requirements. When the dimensions and accuracies of the products are defined in connection with the functional requirements through design logic, the CAD system can automatically make decisions according to the given requirements. In this paper, suitable presentation formats and processing functions for these two types of design logic are discussed. The importance of design logic in product modelling is proven through several case studies in this paper. As a conclusion, the intelligent product modelling system is developed, which should expedite the progress of design automation in the near future. In conceptual design, the design logic is processed in the modelling system and the product structure, with the technological information decided automatically from the functional requirement. Automation in the detailed design phase is also facilitated by the modelling system using the product specified design logic in the product model.

1. INTRODUCTION CAD/CAM systems have been developed mainly as tools for supporting routine work such as generation of engineering drawings and NC programs. On the other hand, CAE systems have been developed by connecting engineering analysis and computer systems. CAD/CAM systems and CAE systems are being integrated, which leads to the requirement to develop advanced design support systems to execute more intelligent information processing. In the field of manufacturing software for mechanical products, however, there are several problems remaining to be resolved in order to move present systems to the development of integrated and intelligent CAD/CAM systems. They are as follows: u No system can support the early stage of product design, such as conceptual design. • Systems have no concept of the designer's inten-

tion which is particularly important in the design modification process. a The lack of the product designer's intention, for instance process planning and operation planning, makes it impossible to generate production drawings automatically. • The ambiguity of the production process makes it difficult to decide the final design parameter values. Solutions for the third and fourth problems are essential when CAD and CAM systems are integrated for automatic processing from product design to production planning. These problems are caused by the fact that there are no software systems which can use logic to generate the product from the requirement. Present CAD/CAM systems process only the 2-D/3-D geometries of parts/products. They do not treat the 499

500

Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988

functional requirement and the geometric data in connection with the logic in the design process, which is required to realize the more effective and automatic product design. In order to realize a higher functional and more intelligent CAD/CAM system, it is vital to develop an intelligent product modelling system which includes the processing of logic in the design process. In this paper, first, the role and importance of the intelligent product modelling system are explained through an analysis of the product design process. Second, the methodology with which to treat and connect the design logic with the product modelling system is proposed. Last, some examples of design logic in the intelligent product modelling system are illustrated. 2. PRODUCT DESIGN PROCESS AND PRODUCT MODEL

The input of the product design process is generally the functional requirement for the product, and the output is the design solution. There are several discussions of the estabfishment of the product design process model. 1 The geometric modelling process, as shown in Fig. 1, is proposed in this paper. The product design process consists of three phases: the conceptual design process, the fundamental design process and the detailed design proceSS. The functional requirement is generally divided into a rough functional requirement and a detailed Functional Requireaent

[ Conceptual Design I I Rough Structure of Product i Funda|ental besign

I

Detailed Structure vith

undecided design paraseters [

Detailed Design I

1 Design Solution All design paraseters are decided. Fig. 1. Product design process.

functional requirement. The detailed requirement is rarely given at the beginning of a new design. In the conceptual design phase, the rough functional requirement is input to decide the rough structure or adequate mechanism of the design solution. The output of the process is the rough sketch or the name of the adopted mechanism of the solution. In the fundamental and detailed design phases, more detailed requirements are input which determine the output of the conceptual design phase. The design solution includes the design parameters such as the geometric shapes, accuracies and materials. The process to output the design solution with some undecided design parameters is called the fundamental design. The process to output the design solution with all parameters decided is called the detailed design. In order to decide all design parameters, the designer should take the production design process into consideration. Some parameter values decided in the product design process may be modified in production design, including process planning and operation planning. The product designer must take this modification into account: the product design process and the production design process essentially cannot be divided, although they seem to be independent at present. In other words, product design and production design should be integrated. To integrate CAD and CAM systems, first, the information commonly used in CAD and CAM must be integrated and processed in the same manner. The design solution is presented by a computerinternal model in CAD/CAM systems. The computer-internal model of the product is simply called the product model in this paper. The information most closely related to the functions of the mechanical products is the geometric information, and the product model usually adopted by present CAD/CAM systems describes the geometric shape of the parts]products. However, CAM systems which support production design require that technological information such as dimensions, tolerances, accuracies and materials is included in the product model. Moreover, when CAD/CAM systems support the designer's creative tasks, the designer's logic for getting the design solution has to be included in the product model. The development of the software system to treat the logic is required to connect the functional requirement with the design parameters. 2 The product model including all required product specified information used in the intelligent CAD/ CAM systems is called the integrated product model in this paper; the structure is shown in Fig. 2. 3 The

501

Product design logier or product modelEng • E. ARAIand K. IWATA

Integrated Product godel

Functional

Design

requirement ~ ' ~ - - - - - ~

Geometric godel

logic -

I

Geometric informstion4~f~Techological

Notes

information

I

I

P a r t shapes

Arrangement informat ion

)

1

[

Connecting information

Accuracies

I Dimensions and To ! erances

I Part mater ia I s

Fzg. 2. Information included m the product model. Note: <--, shows that there is a relation between two pzeces of information.

intelligent product modelling system constructs and manages the integrated product model. Figure 3 shows the role of the integrated modelling system in relation to other CAD/CAM application software. The intelligent product modelling system manages the integrated product model including the design logic, and processes the design logic in connection CAD A p p l i c a t i o n Subsystems

I

I

Inteli

with the functional requirement and the design parameters in the design solution. Each CAD/CAM application accepts the models constructed in the modelling system whether they are correct or not. Data integrity (for example, that the geometric shape of the product is well-formed) should be ensured by the integrated product modelling system. CAN A p p l i c a t i o n Subsystems

!E1 E!E 1 agent Product Modelling System

Fzg. 3. Intelhgent product modelling system m CAD/CAM

502

Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988

In most product modelling systems, however, technological information--including accuracies and materials--is treated as notes or comments, and such systems seldom ensure matching between the geometric and the technological information or among the technological information. The system has not yet been developed to process the design logic for the real automation of product design. The intelligent product modelling system is one of the most important fundamental software systems for future intelligent CAD/CAM systems. 3. DESIGN LOGIC Design logic is that needed to generate a design solution from a given functional requirement and is roughly divided into two types. The first type is used independently of products or designers where the method of design evaluation or the design procedure itself is established in the organization. Logic is seen in conceptual design when the rough structure of the product is decided by the designer's common sense or in fundamental/detail design when the designer selects standard parts and units. The second type is peculiar to each designer or each product. This logic is indispensable for other people to know how the designer generates the solution and decides the

design parameter values. These two types of design logic are called here generalized design logic and specified design logic, respectively. Generalized design logic can be shared by designers in an organization, and it is preferable that the logic is described independently of each product model. On the other hand, specified design logic should be described for each product model so that the design logic connects the functional requirement to the design parameter data of the product. The intelligent product modelling system consists of three main subsystems shown in Fig. 4. The first subsystem is the requirement processing system, which treats the generation and the management of the functional requirement mostly described by natural language. The second is the geometric modelling system, which constructs and manages the geometric model, including geometric and technological parts/products information. The third subsystem is the knowledge base system which processes the design logic. Each subsystem contains the modelling and data management information. Functional requirements, geometric models and design logic are, however, not independent--especially design logic, which connects the functional requirement with the geometric

CIg Application Softwares I

Design solution

Geometric

Modelling System

I I

~

knowledge

Base

Requirement

Processing

System

esi~

System

I

~ R~ui~nH

t I

L I

Integrated Product Model

I

Fig. 4 Intelhgent product modelhng system processing general and speofied design logic.

Product design logic for product modelling • E. ARAI and K. IWATA

503

Designer

Designer's Detailed

Functional

leot'

Requirement

1

1

* Requiresent-

Logic

L

Design

--I Design

T Structure Rough

Detailed I

I Produ l

Design

Proved Logic and Solution

Model

Model

I

I n t e l l i g e n t Product Modelling

System

I

Fig. 5. Flow of a new design using an intelligent product modelling system

model and which is required to integrate the geometric modelling system, knowledge base system and other application software. Generally, the rough structure of the product is decided in the conceptual design, while the detailed design parameters of each part are decided in the fundamental/detail design. Figure 5 shows the flow of new design using generalized and specific design logic. In the following sections, the design of a bearing structure and a robot arm are taken as examples to explain generalized and specified design logic and their processing in the intelligent product modelling system. 4. GENERALIZED DESIGN LOGIC Design logic is a kind of knowledge used in the design process. Knowledge representation methods include: the production rule, semantic network, and frames among others. In this study, the production rule is adopted because of the ease of understanding and the capability of treating fragmentary knowledge.

In the knowledge base system, production rules are described by the form "If-Then-Else" as shown in Fig. 6. This new form is an effective means of presenting design logic and powerful in forward chaining. In order to use the new form of production rules, each rule has to declare the restricted variables which are used in the rule and must be evaluated before the rule is applied. The structure of the knowledge base system to treat the design logic is shown in Fig. 7. The knowledge base system is combined with the geometric modelling system and the requirement processing system so that the intelligent product modelling system can process the geometric modelling process or decide the design parameters described in the design logic in connection with the functional requirement. In this section, the presentation and information processing of a generalized design logic are explained by taking the conceptual design of a bearing structure as an example. The conceptual design flow using the intelligent product modelling system is shown in Fig. 8. The system input is the rough requirement of a

504

Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988

rRESTRICED /PREDICATEI |PREDICATE ~LvARIABLE "-~ LOGIC -~ LOGIC t FORMULA [ FORMULA ._..LRECOGNIZED _~ CALCULATION SUBROUTINE Fig. 6. "If-Then-Else" production rule.

Des i gner

Geometric Modelling System

Requirement Processing S~stee

Intelligent Product Modelling System

Reason Explain Mechanism

Inference Engine

Fig. 7. Knowledge base system.

505

Product design logicfor product modelling• E. AP.AIand K. IWATA Rough

Functional

Rough Specification lnowledgej Structu~ Geometric

Speclflcation I/*Accuracy~ ProcesslngSystem Requirement

Base System

i",°r,ue !

I "IntrumeotsI [ '.oad ,J

Accuracy I Modelling Material| System Descriotion

Data

Rough Oeoletr?c Model of Bearing Structure

Fig. 8. Conceptualdesign flow. rotational function which consists of the following four items: • Accuracy of rotational movement. • Starting torque. • Description of connection instruments at the edges of the rotary axis. • Load to the rotary axis. The relative value is selected instead of the absolute value for the first, second and the fourth items shown above. For the third item, the description is input whether or not there exists a connected instrument at each edge. The output of the system is the rough bearing structure. The geometric structure of the design solution candidates provided in the system is roughly divided into 24 types as shown in Fig. 9; however, up

to 124 variations are also provided including the use of ball bearings and so on. The output bearing structure also includes technological information such as surface roughness, coaxiality, fit tolerance and so on. Fifty-seven rules are used in this system, several of which are shown in Fig. 10. qlae decided geometric data of the rough structure are automatically sent to the geometric modelling system, where the geometric model of the bearing mechanism is constructed by using the three-dimensional solid model. Figure 11 is an output example which consists of used rules plus the type and value of accuracies, and a sectional view of the three-dimensional geometric model of the bearing structure. The geometric model constructed in the concep-

11o. I

11o.2 11o.3

No.4

11o.13

11o.14 11o.15

No.16 1to.17 11o.18 11o.19 tto.20

No.5

No.6

No.7

No.8

F:g. 9. Fundamentaltypes of beanng.

N o . 9 No.lO

No.ll

No.12

11o.21 11o.22 No.23 !to.24

Robotics and Computer-IntegratedManufacturing• Volume4, Number 3/4, 1988

506 I

RULE16 &RESTRIC

{52} *DIAMETER.*SUPPORT-RO

&IF ^sEQ(*CONTACT-NO.$ONE) ^sEQ(,DIRECT-F(*SUPPORT-RO).$FORCE1) &THEN

^SlS(?TOUCH-CASE($ONE).$PIPE) ^#1S(?TOUCH-CASE($TWO).SHOTHING) ^#IS(?TOHCH-SHAP($ONE).$CY)

^#IS(?TOUCH-SHAP($TWO),$NOTHING)

^slS(*PATTERH,$EOTATIONO1)

[ RULE17 &RESTRIC

{52} *DIAMETER.*SUPPORT-RO

&IF

^#EQ(*CONTACT-NO,$TWO) ^#EQ(,DIRECT-F(*SUPPORT-RO).$FORCEI) &THEN ^s IS (?TOUCH-CAgE ($ONE). $P IPE) ^sIS (TTOUCH-CASE ($TWO). SP IPE) ^# IS (?TOUCH-SHAP($ONE),$CY)

^SIS (?TOUCH-SHAP ($TWO), $CY) ^# IS (?TI (*SUPPORT-RO. $SYHHETRY). SIN) ^sIS (*PATTERN,SROTATION02)

{52} &RESTRIC *DIAMETER.*SUPPORT-RO

! RULE18 &IF

^#EQ (*DIRECT-F (*SUPPORT-RO). SFORCE2) ^~EQ (*CONTACT-NO. $ONE) ^#EQ (*THRUSTPATI. $5) vSEQ (*THRUSTPATI. $4) ^#EQ (*OBJECT. STERN) &THEN ^#IS(?TOUCH-CASE($ONE).$CAP) ^~IS(?TOUCH-CASE($TWO).$NOTHING) ^~IS(?TOUCH-SNAP($ONE).$CY) ^#IS(?TOUCH-SHAP(STWO).SNOTHING) ^~IS(*PATTERN.$ROTATION03)

Fig 10. Example rules of generahzed design logic used m conceptual demgn tual design is sent to the fundamental/detailed design phase, where the detailed structure, shape, technological information and other design parameters are decided. The developed knowledge base system enables the designers to add design logic easily, which supports more effective and intelligent

product design. At the same time, random input of the functional requirement can be processed by using backward chaining. 5. SPECIFIED DESIGN LOGIC Specified design logic is also described by production rules, and is processed in the knowledge base system as well as the generalized design logic. The product design process, using specified design logic, is explained taking the design of robot arm as an example. Figure 12 shows the fundamental design flow of the robot arm. The input to the system is the fundamental structure of the multi-joint robot where the number of joints, type of each joint and length of each arm are decided. In the fundamental design, the designer makes a first approximate solution in which the design parameters describing the sectional view and material of each arm are given through the designer's experience. According to the required motion, the positional accuracy of the robot hand and the force/torque on each joint are calculated by the kinematic simulation system connected with the intelligent product modelling system. Then they are compared with the detailed functional requirement, which includes both the accuracy and the torque/ force on each joint. When the calculated values satisfy the requirement, fundamental design ends and the design parameters are output. In general, the approximate solution does not fulfil the requirement, and the parameter values must be modified. The knowledge of parts selection and design parameter modification is stored in the knowledge base system as the specified design logic in connection with the robot's geometric model and the description of the functional requirement. The following items are included as the detailed functional requirement in the developed system: • Robot weight. • Accuracy of the movement of the hand. • Force and torque on each joint. When the item values, calculated by the kinematic simulation of the candidate solution, do not satisfy the requirement, the system decides the goal values of the weight and the stiffness for each robot arm in the first step. The decision method is described by the production rules and stored as the specified design logic. In the second step, the decided weight and stiffness become the functional requirement for each arm. Four design parameters are taken for each robot arm in the developed system. These are: • Sectional view pattern (circle, square, solid and hollow).

Product design logic for product m o d e l h n g • E. ARAI and K. IWATA ..............................

-1 -2 -3 -4 -5

1 -1 -1 0 -1

~psir ($1111 gguld ($par~l gstart-truq($smatt ~put-2slde ($part2 ~=s (?dtrec~-f

.............................. STEP[ STEP[ STEP[ STEP[ STEP[ STEP[ STEP[ STEP[

1 2 3 4 5 6 ? 8

]= ]= ]= ]= ]= ]= ]= ]=

<

specification

>

Spar~l Spart2 ). )$motor ($partl

< process

507

Spart2 Srotat=on

,

).

). )$force3

of

reason

).

>

rule06 rule07 ruLeOS ruLe12 rule14 rule30 rule41 rule45

..............................

< predtc,&'i:,e L o g i c >

%gutd($par~1,$pSr~2,$rotatlon) gstar~-truq($small) ~Ioad($part2,$small) ~speed($part2,$taw)

~use($1111,Sbearlng-b) ~concentlod($partl,$large)

technical

E

inforlati0n

partl

part2

circularity

o

o

eylindrieity

o

o

coaxiality

o

sYInetry

o

circle-run-out

o

all-run-out

o

o

fitlness

k7

n6

dgre3

dgre3

roughness vaviness

o

o

o

Fig. 11. O u t p u t example of conceptual design.

• Outer dimensions of the sectional view. • Thickness. • Material. The relation between the change of each design parameter value and the attributes of the product (weight and stiffness of robot arm) is described as the specified design logic in the knowledge base system by the production rules. By using the knowledge base system, it is possible to process the complicated relation between the design parameters and the attributes and to introduce feasible solutions to satisfy the requirement, which results in making a progress for automatic and effective product design. Figure 13 shows examples among the 100 rules used

in the system. The design parameters are modified as shown in Fig. 14. In the product design process, the designers generate and apply their specified design logic. The specified design logic includes design parameters such as geometric pattern, accuracies, materials and dimensions. Most of the present three-dimensional solid modelling systems cannot treat the dimensions in connection directly with the geometric data. Dimensions are defined in these systems by using the parameters which appear in the geometric modelling process, like the generation of primitives and set operations. The developed intelligent modelling system

508

Robotics and Computer-Integrated Manufacturmg • Volume 4, Number 3/4, 1988 Detailed Functional

from Conceptual Design

-1

Rough Robot

Structure

Requirement

*Accuracy -Force/Torque

Kinematic

]

Simulator Force Torque Accuracy etc.

satisfied

Design Solution im

Hodified Design Solution

to Detailed Design

~not satisfied Design Logic to decide the specification

Decision of goal I specification for ~ each arm I Weight for each arm Rigidity etc. [

Reasoning

for each era I

[ ~

for each arm

=

_

~

parameters| Shape / I . . ~ _ ~

Design Logic to decide the design parameters of each arm

~Haterial)

Modification I of Geometric Hodel

Fig. 12. Fundamental design flow for robot arm. treats the following dimensions, which are often seen in mechanical products, so that they are described directly in the geometric model data. The dimensions are defined as: • Distance between two vertices. • Distance between two parallel planes. • Distance between parallel edge and plane. • Distance between a vertex and a plane. • Diameter or radius of cylinder. • Angle between two planes. In order to process the dimensions, the intelligent product modelling system has the following functions: • The system secures matching between the defined dimenslon value and the geometric data of the object. • The range for each dimension can be defined so as not to destroy the topology of the object.

• The system modifies the geometric shape of the object according to the change of the dimension values. These functions are required when the designers describe their specified design logic in the integrated product model. The specified design logic is presented by using the production rules, and stored in the file area of the intelligent product model. Specified design logic plays an important role, especially in automation of the modification design process and the optimization process of the design parameters. The designer utilizes the specified design logic as the description of the qualitative and quantitative relations between the design parameters and the attributes/functions of the product, which is modelled in the integrated product model by the intelligent product modelling system. In order to select the design parameters to be modified and to

Product design logic for product modelhng • E. ARA1and K. IWATA { rule61

509

{12}

&if "%patern($a) ^,eq (?judge($width),$float) ^~eq(?judge($thiek),$fixed) ^Seq(?judge($faee2).$float) ^~eq(*faee-st,$eubid-stop) &then

^~is(*marka4,8+l . . . . i+01) ^~is(*faee-new,$cubid-hole) I ruleS2 &if

{12)

^aeq(*marka4,8+l . . . . i+Ol) &then ^ffdelete(*reprel) ^R~lete(*repre2) ^~delete(*repre3) ^=operatel(~faee-new,~width-st,~thiek-st.~elastie-st,~grv/vol-st .gwidth-lm,gthiek-lm,:ridity-lm,~gravity-lm ,~reprel,~repFe2,~repre3) ^~is(*&rka5,*reprel)

^~is(~width-new,~repre2) ^Wis(~thick-new,~repre3) I rule63 &if

{12} ^lleq (:~a rkaS, @+1. . . . i *01)

&then

"%display($face,*faee-new) ^%display($width,*width-new) ^%display($thiek,*thiek-new) ^%end { rule54

{12}

&if ^~eq(*marka4,0-1. . . . i+01)v~eq(*marka5,Ù-I. . . . i+01) ^,eq(?judge($elastie),$float) ^~eq(?judge($grv/vol),$float) &then

^%di s pIay ($faee, *face -new) ^%disp 1ay ($wi dth, ¢wi dth-n ew) ^%display ($thiek,*thiek-new) ^~display ($elastie, Zelast ie-st) ^%display ($grv/vo 1, $grv/vo 1-st) ^%end Fig. 13. Example rules of speofied design logic used in fundamental design decide the revised values, the stored design logic is processed as shown in Fig. 11.

6. C O N C L U S I O N The d e v e l o p m e n t of the intelligent product modelling system to process not only the geometric information, but also the design logic is necessary in order to realize an intelligent C A D / C A M system which supports both routine and intelligent tasks in the

product design process. Design logic can be divided into generalized design logic and specific design logic. In this paper, a methodology is proposed to describe the design logic by production rules and to process the design logic in the product modelling system. The developed intelligent product modelling system consists of three subsystems: • A geometric modelling system to process the geometric model of the products.

510

Robotics and Computer-Integrated Manufacturing • Volume 4, Number 3/4, 1988

J

~

-I-or~ue [ him3 ~

~ 49.5149.0 !

i~.l eeti°nal I ~'] View of ~

Q

8rI

21.0/24.5 J

L48"8149"0 !/~ ~=~]"~! © ////

Fig. 14. Change m geometry of arm section to satisfy funcnonal reqmrement. Note: left side of slash shows simulated value and right side shows required upper hmlt value. • A specification processing system to treat the functional requirement for the product. • A knowledge base system to treat the design logic. Two kinds of design logic are explained through examples. The design solutions can be generated automatically through the intelligent product modelling system by using design logic. Through these examples, the importance of design logic in the product model is clarified and the availability of the

developed system in intelligent C A D system in the near future is evaluated. REFERENCES 1. For example, see Preprints of IFIP W.G.5.2 Working Conference on Design Theory for CAD, 1985. 2. Spur, G., Krause, F.L., Research problems of CADsystems. A n n . C I R P 34: 183-186, 1985. 3. Aral, E., Iwata, K.: Development of integrated product model for CIM. Proceedings o f the 18th C I R P M F S S e m i n a r R & D on the W a y to C I M . pp. 1-15, 1986.