- Email: [email protected]
Fuzzy logic based process control strategy for effective sheeting of wheat dough in small and medium-sized enterprises
Journal of Food Engineering 199 (2017) 93e99
Contents lists available at ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
Fuzzy logic based process control strategy for effective sheeting of wheat dough in small and medium-sized enterprises Jyothi Mahadevappa*, Frauke Groß, Antonio Delgado €t Erlangen-Nurnberg (FAU), Cauerstraße 4, 91058, Fluid Mechanics, Department of Chemical and Biological Engineering, Friedrich-Alexander-Universita Erlangen, Germany
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 June 2016 Received in revised form 12 December 2016 Accepted 17 December 2016 Available online 19 December 2016
One of the prime aspects in small and medium-sized baking enterprises (SMEs) is to control and improve the dough sheeting process. Dough sheeting is an important phase in producing various baked products such as breads, pastries, pizza etc. This is one of the key feature that helps to achieve optimized timing and product quality. The deployment of a fuzzy control system is one of the feasible option when complexity of the process is high. The fuzzy control system (program) developed in this research can regulate the roll gaps based on the input parameters such as dough thickness, rheological measurements and surface cracks. The results demonstrated that by the implementation of a fuzzy system an optimized dough sheeting method is achieved. On comparison of the results with a non-fuzzy system, the dough was rolled in less time without making any compromise with the dough quality. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Dough sheeting Process control Surface cracks Rheological properties Fuzzy logic
1. Introduction Even for the small to medium sized industry, handling the flow of process information with an automated system is crucial not only to control the process, but to ensure optimum product quality and productivity. Computerized controls can be used to upgrade existing equipment as well as for new facilities. Digital or computer-based process controls optimize process and equipment efficiency (Mittel, 1997). In the baking industry achieving quality products consistently must be the goal of any baker striving to improve the business. Their greatest potential lies with the integration of bakery processes with computing environments, systems and techniques (Young, 2007). The use of advanced control techniques such as model-based predictive controllers, intelligent and software sensors, neuro-fuzzy control and expert systems lead to reduced costs, increased quality and improved safety along with an expectation to improve process profitability and business competitiveness (Linko & Linko, 1998). One of the well-known processes that helps in decision making based on human operator's experience and measured process variables is fuzzy logic (Zhang & Litchfield, 1993). Some applications of fuzzy logic in baking industry are given in
* Corresponding author. E-mail address: [email protected] (J. Mahadevappa). http://dx.doi.org/10.1016/j.jfoodeng.2016.12.013 0260-8774/© 2016 Elsevier Ltd. All rights reserved.
Table 1. The aim of the proposed work is to establish a fuzzy control system considering parameters such as dough rheology, thickness and surface cracks for designing efficient rolling process of wheat dough particularly in SMEs. The rheology of wheat dough is considered crucial in the successful manufacturing of bakery products (Faridi & Faubion, 1990). It plays an important role in determining the processing behavior and quality of baked products (Bloksma, 1990). Rheological principles and theory can be used as an aid in process control and design (Dobraszczyk & Morgenstern, 2003). In automated bakeries, knowledge of dough viscosity plays an important role in production control and equipment design. Dough viscosity may relate to the quality of the baked product and may control that quality in some instances (Sharma et al., 1993). In addition to dough rheology, dough thickness and surface cracks are also prominent parameters affecting the baked product quality (Quail et al., 1990; Mahadevappa et al., 2015). Moreover controlling dough sheeting processes has been a long standing challenge in the food industry. Due to the difficulties in controlling the dough thickness on the line, there has been issues in maintaining a consistent rate of production (Chakrabati-Bell et al., 2010). Efficiency of the sheeting process is achieved by using the output parameters provided by the fuzzy control system such as addition of flour and the roll gap (using which the next rolling step is adjusted). Addition of flour is an important aspect in reducing the stickiness of the dough. Stickiness is a common hindrance
94
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
Table 1 Applications of fuzzy logic in baking industry. SN
Application
Reference
1. 2. 3. 4. 5. 6. 7. 8.
Quality evaluation of biscuit during baking Bread baking process control Cookie baking Consumer ratings on chocolate chip cookies Controlling the browning process Prediction of chapati making quality from different Indian wheat varieties A sensory study for analysis of acceptability of millet-based dough and breads Effective control of temperature in the bread baking chamber
(Perrot et al., 1996) (Kim and Cho, 1997) (Perrot et al., 2000) (Davidson et al., 2001) (Ioannou et al., 2004) (Gangadharappa and Prabhasankar, 2011) (Singh et al., 2012) (Finaev et al., 2015)
interfering in the production process, causing considerable wastage and contamination of equipment (Dobraszczyk, 1997). The results of this research is an addendum to the inline measurement technique used to assess the quality of the wheat dough during sheeting process (Mahadevappa et al., 2015). 2. Materials Two types of wheat dough are considered for this study, namely ciabatta (dough with yeast) and puff pastry (dough without yeast). Two different types of wheat flours commercially known as “Brema flour” and “Keks flour” are used to prepare the dough samples along with different additives such as oxygen enriched water (80e115 ppm), tap water (2e7 ppm) and cysteine (0.001%, 0.005%, 0.01%). The wheat flour is supplied by Bremer Rolandmühle, Erling GmbH und Co KG, Bremen. The dough is prepared using the recipe as shown in Table 2. In this research the spiral kneader from the company WP Kemper, Germany and the sheeting machine from Rondo Burgdorf AG, Switzerland are used. The plate-plate rheometer from Anton Paar MCR 301, Ostfildern-Scharnhausen, Germany is used to carry out the rheological measurements. For every trial 15 kg dough is mixed using the spiral kneader. The dough is manually divided into approximately 2.5 kg portions. Each portion is sheeted within a roll gap range from 45 mm to 5 mm using two sheeting programs (Table 3). Program 1 with eight steps, corresponds to be a slow and gentle program similar to the one often used in small and medium sized bakeries. This program rolls the dough in a gentle manner using small increments of roll gaps and with least possible stress reduction. Program 2 with five steps is comparatively faster, due to the lesser number of rolling steps. In this case, the dough experiences significantly greater stress. These two are the basic programs using which the fuzzy system optimizes a program of its own.
Table 2 Dough recipe for ciabatta and puff pastry. Ingredient
Ciabatta (%)
Puff pastry (%)
Flour Water Salt Sugar Yeast Sunflower oil Fat
100 56 1.2 10 6 2 e
100 50 2 2 e e 5
3. Implementation of the fuzzy system A typical fuzzy system operates on input parameters, applies the defined rules and finally provides the output. Keeping the same framework, two fuzzy system variations - fuzzy variation 1 (FV 1) and fuzzy variation 2 (FV 2) are developed in the present research (Fig. 1). In FV 1, the dough thickness and surface cracks (obtained using inline measurement technique - refer sections 4.1 and 4.2) are selected as input parameters. FV 1 evaluates these input parameters against a set of defined rules in order to select the most efficient output parameters, which form the input for the next rolling step. The addition of flour and roll gap are the two output parameters. FV 2 is similar to FV 1 except for the fact that it considers rheological measurements (offline measurements e refer section 4.3) as an extra input parameter. The rules for these two variations of fuzzy system are based upon the findings of the image processing and rheological measurements (Mahadevappa et al., 2015). They are programmed using the software MATLAB (Fuzzy Logic Toolbox). The complete outlook of the setup for the dough sheeting process which is a combination of the current research and the setup explained in the paper by (Mahadevappa et al., 2015) is depicted in Fig. 2. 4. Input parameters Decision making in fuzzy logic is based upon the input parameters. The input parameters used in this research are dough thickness, surface cracks and rheological measurements. 4.1. Dough thickness Dough thickness is the thickness of the dough obtained when sheeted between two rollers with a specific roll gap (distance between the two rollers). The fuzzy system should output a suitable rolling profile for every dough being rolled. The belt speed of the sheeting machine is 26.4 cm/s. The first rolling step of 45 mm as well as the final thickness of 5 mm is achieved by default. The intermediate rolling steps are oriented between a gentle and a consistently stressful program depending on the nature of the dough. Based on the roll gap, there are four membership functions for the dough thickness input: thick, medium, thin and final thickness (Table 4). The dough thickness input is in the range of 5 mme45 mm.
Table 3 Representation of the basic predefined rolling programs. Sheeting program
Step 1
Step 2
Step 3
Step 4
Step 5
Step 6
Step 7
Step 8
Program 1 (P1) Program 2 (P2)
45 mm 45 mm
35 mm 30 mm
25 mm 15 mm
15 mm 5 mm
10 mm 5 mm
8 mm
6 mm
5 mm
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
95
Fig. 1. Overview of the fuzzy control system developed.
Fig. 2. Process flow diagram.
Table 4 Membership functions for the inputedough thickness. Membership functions
Range (mm)
Thick Medium Thin Final thickness
45e21.5 21.5e10 10e5 5
4.2. Surface cracks The surface cracks on laminated wheat dough are undesirable and generally affect the final quality of the baked products. Surface
cracks formed during the sheeting process are assessed using an inline measurement technique capable of automatic detection and evaluation of dough surface as well as its measurement in terms of crack ratio (Mahadevappa et al., 2015). Crack ratio is the percentage of the cracked surfaces to the total surface area of the dough. This crack ratio obtained forms the input (surface cracks) to the fuzzy variations FV 1 and FV2. The crack ratio for a very good dough surface is less than 0.12 and for an extremely cracked dough surface the value is greater than 0.45. Based on the crack ratio, there are five input membership functions for the surface crack input: none, low, slight, medium and high (Table 5). The surface cracks input is in the range of 0e1.5.
96
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99 Table 5 Membership functions for the inputesurface cracks. Membership functions
Crack ratio
None Low Slight Medium High
<0.12 0.12e0.20 0.20e0.32 0.32e0.45 >0.45
4.3. Rheological measurements In order to establish a fuzzy system for efficient rolling process of wheat dough, focus is also on identifying the rheological properties of dough mixtures with yeast (ciabatta) and without yeast (puff pastry). To assess the dough processability, additives such as oxygen enriched water and cysteine are used. Addition of these additives effect the rheological properties in both negative and positive manner. The interpretation of rheological oscillation and rotation measurements is intended to draw conclusions on the dough processability. To determine the rheological properties of the wheat dough, oscillation and rotation measurements are carried out using a rheometer. Based on the oscillation measurements, input sets for the fuzzy system could not be defined as the results are relatively in a smaller range. Rotation measurements are done with a shear rate of 2 1/s in order to determine viscosity with respect to time (Fig. 3). Here the viscosity increases to maximum and then starts decreasing. From this point of time the dough begins to flow. In contrast to the oscillation measurements, a significant difference between various types of dough can be evidently recognized. As a result, rotation measurements (viscosity vs time) were used as an additional input parameter in fuzzy variation 2. At a shear rate of 2 1/s dough prepared with Keks flour, yeast and 0.01% cysteine displayed higher viscosity implying a good quality dough. On the contrary, dough prepared with Keks flour, yeast and
oxygen enriched water showed lower viscosity implying a low dough quality. The dough classification is primarily based on the viscosity values obtained during experimental procedures and the user's perspective (stickiness, vision). It is categorized into good, medium, poor and bad (refer Fig. 3). The dough with higher viscosity, good machine processability, smooth and crack free surface is considered to be good. Whereas, the sticky dough with low viscosity values, poor processability and high surface irregularities (cracks) is considered to be bad and not acceptable. The dough classified into medium and poor do not necessarily matchup to the parameters of a good dough, but are also not in the bad category of the dough so as to be discarded. The mid-range viscosity values infer that the medium and poor quality dough are also processable. In addition, opinion of expert bakers over the viscosity values and the experimental results are also considered for the above classification. Based on this categorization of viscosity, there are four input membership functions for the time measurements: good, medium, poor and bad (Table 6). 5. Output parameters The output parameters of the developed fuzzy system are mainly the control variables. These indicate to the user, the next steps to be followed in order to increase the efficiency of the dough processing.
Table 6 Categorization of membership functions based on viscosity values. Membership functions
Viscosity range
Good Medium Poor Bad
>6616 4900e6616 3400e4899 <3400
Fig. 3. Graphical representation of rotation measurements (Viscosity vs Time) with a shear rate of 2 1/s for different dough compositions.
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
5.1. Addition of flour Addition of flour also known as dusting flour, helps to reduce the stickiness of dough and also the formation of the surface cracks. It is generally observed that, in bakeries the stickiness of the dough is compensated by sprinkling of flour. How much flour is to be added is a matter of experience, it can vary from 30 g to 113 g. It is important to note here that often more flour than necessary is applied in bakeries to prevent adhesion of the dough to the machine. Excessive addition of dusting flour can also lead to a reduction in dough quality. As a step towards optimization, this variable is set up as an output parameter, through which we can get a dough profile with a final thickness of 5 mm in a few steps thereby limiting the excessive usage of flour. The addition of flour is gauged in three categories namely, no flour, little amount of flour (half a fistful ~ 30 g) and medium amount of flour (fistful ~ 75 g). Based on this categorization, there are three output membership functions for the addition of flour: none, little and medium. 5.2. Roll gap This output parameter provided by the fuzzy system is the extent by which the next rolling step needs to be reduced. For example, if the dough is rolled with the first rolling step of 45 mm, and 14 mm is the roll gap given as an output by the fuzzy system then the next rolling step is 45 mm-14 mm ¼ 31 mm. 6. Implementation of the fuzzy logic rules Mamdani's fuzzy inference method is used in this implementation (supported by MATLAB). Once the input and output parameters are set the next step is to connect them with the IFTHEN conditional rules, so that an inference can be made. The implementation of the fuzzy system is based on valid if-then rules. The input and output parameters are defined using trapezoidal membership functions.
97
Once a good quality dough surface is achieved, there is no further necessity to either add any flour or to roll the dough further. This rule could be represented as below: If (dough thickness ¼ final thickness) and (surface cracks ¼ slight) then (addition of flour ¼ none) and (roll gap ¼ 0 mm) Thus the first variant of fuzzy system helps to achieve a good quality dough surface by defining a set of fuzzy logic rules as show in the examples above. 6.2. Fuzzy system- variation 2 In FV 2 along with the dough thickness and surface cracks, rheological measurements (Time measurements) is also considered as an input parameter. Here again the addition of flour and roll gap form the two output parameters. In the below rule set, depending on the varying degree of surface cracks and time measurements suitable actions are taken with respect to the output parameters namely addition of flour and roll gap. If (dough thickness ¼ thick) and (surface cracks ¼ none) and (time measurements ¼ good) then (addition of flour ¼ none) and (roll gap ¼ 15 mm) If (dough thickness ¼ thick) and (surface cracks ¼ high) and (time measurements ¼ bad) then (addition of flour ¼ medium) and (roll gap ¼ 11 mm) If (dough thickness ¼ thick) and (surface cracks ¼ low) and (time measurements ¼ medium) then (addition of flour ¼ none) and (roll gap ¼ 14 mm) The below rule would suffice for a good quality dough achieving the desired thickness with negligible surface cracks. If (dough thickness ¼ thick) and (surface cracks ¼ slight) and (time measurements ¼ medium) then (addition of flour ¼ none) and (roll gap ¼ 0 mm)
6.1. Fuzzy system- variation 1 In FV 1, the dough thickness and surface cracks are selected as input parameters. This variation provides us two output parameters: addition of flour and roll gap. In FV 1, examples of the rules set for the dough with different crack surfaces are shown below. When the dough thickness is classified as thick (refer Table 4) and there are no surface cracks then it implies a good quality dough, hence there is no necessity of adding any flour and next rolling step can be drastically reduced to 15 mm. This can be represented by the following rule:
Thus the second variant of fuzzy system helps to achieve a good quality dough surface by defining a set of fuzzy logic rules by taking in an extra input parameter as show in the examples above. 7. Results and discussions In the presented research the test setup from the inline measurement technique (Mahadevappa et al., 2015) is used in conjunction with the fuzzy system developed using MATLAB (refer Fig. 2). 7.1. Analysis and assessment of the fuzzy control system
If (dough thickness ¼ thick) and (surface cracks ¼ none) then (addition of flour ¼ none) and (Roll gap ¼ 15 mm) When the dough thickness is classified as thick (refer Table 4) but there are varying degrees of surface cracks observed (slight, high) then it implies that the dough needs more processing. Hence there may be a necessity to add flour and the next rolling step is set accordingly. This can be represented by the following rules: If (dough thickness ¼ thick) and (surface cracks ¼ high) then (addition of flour ¼ medium) and (roll gap ¼ 11 mm) If (dough thickness ¼ thick) and (surface cracks ¼ slight) then (addition of flour ¼ none) and (roll gap ¼ 14 mm)
The two variations of fuzzy control system (FV 1 & FV 2) are tested on two types of dough: dough prepared using Keks flour, tap water without yeast (puff pastry) and dough prepared using Brema flour, tap water and yeast (ciabatta). According to the rheological measurements these two dough exhibited mid-range viscosity values. Two basic program cycles are considered as a reference for the fuzzy system (Table 3). These two programs provide the basic minimum number of steps required to achieve a good quality dough. Program 1 (P1) consisting of eight steps can be envisioned for a dough with very bad structure whereas, program 2 (P2) consisting of five steps can be considered for a dough having
98
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
medium to good quality structure. The job of the fuzzy control system is to optimize the sheeting program by regulating between these two programs. Thus giving a new optimized and efficient program (this program can consist a maximum of eight steps to a minimum of four steps depending on the dough quality). Both types of dough were rolled using FV 1 and FV 2. The sheeting profile of the tested dough types is shown in Fig. 4. In case of puff pastry both the fuzzy variations FV1 and FV2 exhibits similar sheeting profile (program). The dough sheeting process starts with the thickness of 45 mm followed by 30 mm, 15 mm and finally ends with 5 mm (four step program). In case of ciabatta, both the variations of fuzzy system showed different sheeting profiles with equal number of rolling steps (six step program). This infers that the fuzzy system selects the best optimum suited program based on the input quality of the dough. Adopting similar system in the bakeries can help the bakers to achieve optimum processing time. To understand the effects of FV1 and FV2, ciabatta dough results are considered as it showed two different sheeting profiles (programs). Here the difference between FV 1 and FV 2 is seen mainly in the values of the crack ratio obtained (Fig. 5). FV 1 processes the ciabatta dough in six rolling steps indicating good dough quality with the final crack ratio of 0.12. FV 2 also processes the ciabatta dough in six rolling steps again indicating a very good dough quality with the final crack ratio of 0.09. FV 2 displays a better crack ratio when compared to FV 1. This is mainly attributed to the rheological parameter that is taken into account by FV 2 during the dough sheeting process. From the above results obtained, it can be inferred that a fuzzy control system is capable of deciding the best possible program with least number of rolling steps depending on the dough quality. FV 1 and FV 2 are found to be working as designed to roll any type of dough, finally resulting in a smooth surface exhibiting least crack ratio. The fuzzy systems FV 1 and FV 2 were tested on two commonly used types of dough (puff pastry and ciabatta). However, it can be extended to test various other types of dough combinations with the limitation being that the dough consistency should be such that it is able to pass through the first rolling step. A tabulated summary on the effects of the four programs P1, P2, FV 1 and FV 2 on both the dough types with respect to the number
Fig. 5. Difference in the crack ratio between FV1 and FV2 on the ciabatta dough surface.
of rolling steps, time and crack ratio is presented in Tables 7a and 7b. The above results (Tables 7a and 7b) emphasize: FV 1 and FV 2 requires less time for rolling compared to P1 and P2. FV 1 and FV2 help in achieving a better or equal crack ratio compared to P1 and P2 but, with minimal steps. FV 2 displays a better crack ratio due to the fact that rheological measurement is taken into consideration. This design of fuzzy logic system can be extended to SMEs since, it provides adequate advantage over the existing manual processes. Implementation of such a fuzzy system results in achieving a good and preferred dough surface during dough processing stage. The advantages mainly being having an optimized method to roll the dough thus saving time and in turn saving energy and costs when compared to a manual setup. Generally in a manual setup the dough displaying high surface cracks is considered unfit for processing and is discarded. On the contrary, a fuzzy control system will analyze the dough surface and help to reduce the surface cracks and increase the quality by the addition of an accurate amount of dusting flour. In this way the dough that would have been discarded is now modified and its usability is increased. This fuzzy system design provides an optimized, time saving and efficient control on the sheeting process compared to manual intervention which can lead to wastage and is also prone to manual errors such as decisions on the dough quality based on manual senses (vision and touch).
Table 7a Result summary for puff pastry. Puff Pastry
P1
P2
FV 1
FV 2
Rolling steps Rolling time Crack ratio
8 4min 20s 0.10
5 2min 36s 0.12
4 1min 40s 0.11
4 1min 50s 0.10
Table 7b Result summary for ciabatta.
Fig. 4. Representation of the sheeting profile of wheat dough using fuzzy control system.
Ciabatta
P1
P2
FV 1
FV 2
Rolling steps Rolling time Crack ratio
8 4min 15s 0.17
5 2min 35s 0.17
6 2min 30s 0.12
6 2min 40s 0.09
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
8. Conclusions A novel method envisaged to improve the dough sheeting process in SMEs using a fuzzy control system is presented in this paper. The fuzzy system is essentially implemented as a closed loop control system where, the entire process is repeated until an acceptable value of the crack ratio is obtained. Two formats of fuzzy systems were developed (FV 1 & FV 2) differentiated by the input parameter sets. Considering the rheological properties of the dough as an input parameter gave an added advantage to comprehensively assess the dough. Developing a fuzzy system gives an edge in the sheeting process by effective time management and minimizing losses leading to an efficient dough processing method. The target group which would be benefited by implementation of this research are the small and medium scale baking industries. This approach would help them to save time, cost, minimize dough wastage and reduce manual errors. The market studies shows that more and more baking enterprises are moving towards adapting fuzzy control system. Although there are many successful fuzzy logic applications in baking enterprises, there is still a need to transfer and adapt it in the sheeting process of the wheat dough. This pioneering approach also caters to manufacturers of sheeting machine in enhancing their machine design. Acknowledgements The authors wish to thank the German Ministry of Economics and Technology (via AiF) and the FEI (Forschungskreis der Ernahrungsindustrie e.V., Bonn) for the financial support (AiF 16755 N) and the members of Bremerhaven Institute of Food Technology and Bioprocess Engineering (BILB) for their cooperation. The acknowledgment would be incomplete without thanking Nicola Cebulla €lfert for their assistance in accomplishing this work. and Kathrin Wo References Bloksma, A.H., 1990. Dough structure, dough rheology, and baking quality. Cereal Foods World 35, 237e244.
99
€ m, J.S., Lindskog, E., Sridhar, T., 2010. Computational Chakrabati-Bell, S., Bergstro modeling of dough sheeting and physical interpretation of the non-linear rheological behavior of wheat flour dough. J. Food Eng. 100, 278e288. Davidson, V.J., Ryks, J., Chu, T., 2001. Fuzzy models to predict consumer ratings for biscuits based on digital features. IEEE Trans. Fuzzy Syst. 9, 62e67. Dobraszczyk, B.J., 1997. The rheological basis of dough stickiness. J. Texture Stud. 28, 139e162. Dobraszczyk, B.J., Morgenstern, M.P., 2003. Rheology and the breadmaking process. J. Cereal Sci. 38, 229e245. Faridi, H., Faubion, J.M., 1990. Dough Rheology and Baked Product Texture. AVI Van Nostrand Reinhold, New York. Finaev, V.I., Sinajvskay, E.D., Pushnina, I.V., 2015. Fuzzy control model of temperature at the bread baking chamber. Izestiya SFedU. Eng. Sci. 150e157. Gangadharappa, G.H., Prabhasankar, P., 2011. Spreadsheet aided fuzzy model for prediction of chapati making quality. J. food Sci. Technol. 48 (3), 344e348. Ioannou, I., Perrot, N., Mauris, G., Trystram, G., 2004. Development of a control system using the fuzzy set theory applied to a browning processeetowards a control system of the browning process combining a diagnosis model and a decision modeleepart II. J. Food Eng. 64, 507e514. Kim, S., Cho, I., 1997. Neural network modeling and fuzzy control simulation for bread-baking process. Trans. ASAE 40, 671e676. Linko, S., Linko, P., 1998. Developments in monitoring and control of food processes. Food Bioprod. Process. 76, 127e137. Mahadevappa, J., Groß, F., Benning, R., Delgado, A., 2015. Development of an inline measurement technique to assess the quality of wheat dough during the sheeting process. J. Cereal Sci. 64, 183e188. Mittel, G.S., 1997. Computerized Control Systems in the Food Industry. Marcel Dekker, Inc, USA (New York, NY). Perrot, N., Trystram, G., Le Guennecb, D., Guely, F., 1996. Sensor fusion for real time quality evaluation of biscuit during baking. Comparison between Bayesian and fuzzy approaches. J. Food Eng. 29, 301e315. Perrot, N., Trystram, G., Chevrie, F., Guely, F., Schoesetters, N., Dugre, E., 2000. Feedback quality control in the baking industry using fuzzy sets. J. Food Process Eng. 23, 249e279. Quail, K.J., McMaster, G.J., Tomlinson, J.D., 1990. Effect of baking temperature/time conditions and dough thickness on arabic bread quality. J. Sci. Food Agric. 53, 527e540. Sharma, N., Hanna, M.A., Chen, Y.R., 1993. Flow behavior of wheat flour-water dough using a capillary rheometer. I. Effect of capillary geometry. Cereal Chem. 70, 59e63. Singh, K., Mishra, A., Mishra, H., 2012. Fuzzy analysis of sensory attributes of bread prepared from millet-based composite flours. LWT - Food Sci. Technol. 48, 276e282. Young, L.S., 2007. Application of baking knowledge in software systems. In: Cauvain, S., Young, L.S. (Eds.), Technology of Breadmaking. Springer, New York, pp. 207e222. Zhang, Q., Litchfield, J., 1993. Fuzzy logic control for a continuous crossflow grain dryer. J. Food Process Eng. 16, 59e77.
Contents lists available at ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
Fuzzy logic based process control strategy for effective sheeting of wheat dough in small and medium-sized enterprises Jyothi Mahadevappa*, Frauke Groß, Antonio Delgado €t Erlangen-Nurnberg (FAU), Cauerstraße 4, 91058, Fluid Mechanics, Department of Chemical and Biological Engineering, Friedrich-Alexander-Universita Erlangen, Germany
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 June 2016 Received in revised form 12 December 2016 Accepted 17 December 2016 Available online 19 December 2016
One of the prime aspects in small and medium-sized baking enterprises (SMEs) is to control and improve the dough sheeting process. Dough sheeting is an important phase in producing various baked products such as breads, pastries, pizza etc. This is one of the key feature that helps to achieve optimized timing and product quality. The deployment of a fuzzy control system is one of the feasible option when complexity of the process is high. The fuzzy control system (program) developed in this research can regulate the roll gaps based on the input parameters such as dough thickness, rheological measurements and surface cracks. The results demonstrated that by the implementation of a fuzzy system an optimized dough sheeting method is achieved. On comparison of the results with a non-fuzzy system, the dough was rolled in less time without making any compromise with the dough quality. © 2016 Elsevier Ltd. All rights reserved.
Keywords: Dough sheeting Process control Surface cracks Rheological properties Fuzzy logic
1. Introduction Even for the small to medium sized industry, handling the flow of process information with an automated system is crucial not only to control the process, but to ensure optimum product quality and productivity. Computerized controls can be used to upgrade existing equipment as well as for new facilities. Digital or computer-based process controls optimize process and equipment efficiency (Mittel, 1997). In the baking industry achieving quality products consistently must be the goal of any baker striving to improve the business. Their greatest potential lies with the integration of bakery processes with computing environments, systems and techniques (Young, 2007). The use of advanced control techniques such as model-based predictive controllers, intelligent and software sensors, neuro-fuzzy control and expert systems lead to reduced costs, increased quality and improved safety along with an expectation to improve process profitability and business competitiveness (Linko & Linko, 1998). One of the well-known processes that helps in decision making based on human operator's experience and measured process variables is fuzzy logic (Zhang & Litchfield, 1993). Some applications of fuzzy logic in baking industry are given in
* Corresponding author. E-mail address: [email protected] (J. Mahadevappa). http://dx.doi.org/10.1016/j.jfoodeng.2016.12.013 0260-8774/© 2016 Elsevier Ltd. All rights reserved.
Table 1. The aim of the proposed work is to establish a fuzzy control system considering parameters such as dough rheology, thickness and surface cracks for designing efficient rolling process of wheat dough particularly in SMEs. The rheology of wheat dough is considered crucial in the successful manufacturing of bakery products (Faridi & Faubion, 1990). It plays an important role in determining the processing behavior and quality of baked products (Bloksma, 1990). Rheological principles and theory can be used as an aid in process control and design (Dobraszczyk & Morgenstern, 2003). In automated bakeries, knowledge of dough viscosity plays an important role in production control and equipment design. Dough viscosity may relate to the quality of the baked product and may control that quality in some instances (Sharma et al., 1993). In addition to dough rheology, dough thickness and surface cracks are also prominent parameters affecting the baked product quality (Quail et al., 1990; Mahadevappa et al., 2015). Moreover controlling dough sheeting processes has been a long standing challenge in the food industry. Due to the difficulties in controlling the dough thickness on the line, there has been issues in maintaining a consistent rate of production (Chakrabati-Bell et al., 2010). Efficiency of the sheeting process is achieved by using the output parameters provided by the fuzzy control system such as addition of flour and the roll gap (using which the next rolling step is adjusted). Addition of flour is an important aspect in reducing the stickiness of the dough. Stickiness is a common hindrance
94
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
Table 1 Applications of fuzzy logic in baking industry. SN
Application
Reference
1. 2. 3. 4. 5. 6. 7. 8.
Quality evaluation of biscuit during baking Bread baking process control Cookie baking Consumer ratings on chocolate chip cookies Controlling the browning process Prediction of chapati making quality from different Indian wheat varieties A sensory study for analysis of acceptability of millet-based dough and breads Effective control of temperature in the bread baking chamber
(Perrot et al., 1996) (Kim and Cho, 1997) (Perrot et al., 2000) (Davidson et al., 2001) (Ioannou et al., 2004) (Gangadharappa and Prabhasankar, 2011) (Singh et al., 2012) (Finaev et al., 2015)
interfering in the production process, causing considerable wastage and contamination of equipment (Dobraszczyk, 1997). The results of this research is an addendum to the inline measurement technique used to assess the quality of the wheat dough during sheeting process (Mahadevappa et al., 2015). 2. Materials Two types of wheat dough are considered for this study, namely ciabatta (dough with yeast) and puff pastry (dough without yeast). Two different types of wheat flours commercially known as “Brema flour” and “Keks flour” are used to prepare the dough samples along with different additives such as oxygen enriched water (80e115 ppm), tap water (2e7 ppm) and cysteine (0.001%, 0.005%, 0.01%). The wheat flour is supplied by Bremer Rolandmühle, Erling GmbH und Co KG, Bremen. The dough is prepared using the recipe as shown in Table 2. In this research the spiral kneader from the company WP Kemper, Germany and the sheeting machine from Rondo Burgdorf AG, Switzerland are used. The plate-plate rheometer from Anton Paar MCR 301, Ostfildern-Scharnhausen, Germany is used to carry out the rheological measurements. For every trial 15 kg dough is mixed using the spiral kneader. The dough is manually divided into approximately 2.5 kg portions. Each portion is sheeted within a roll gap range from 45 mm to 5 mm using two sheeting programs (Table 3). Program 1 with eight steps, corresponds to be a slow and gentle program similar to the one often used in small and medium sized bakeries. This program rolls the dough in a gentle manner using small increments of roll gaps and with least possible stress reduction. Program 2 with five steps is comparatively faster, due to the lesser number of rolling steps. In this case, the dough experiences significantly greater stress. These two are the basic programs using which the fuzzy system optimizes a program of its own.
Table 2 Dough recipe for ciabatta and puff pastry. Ingredient
Ciabatta (%)
Puff pastry (%)
Flour Water Salt Sugar Yeast Sunflower oil Fat
100 56 1.2 10 6 2 e
100 50 2 2 e e 5
3. Implementation of the fuzzy system A typical fuzzy system operates on input parameters, applies the defined rules and finally provides the output. Keeping the same framework, two fuzzy system variations - fuzzy variation 1 (FV 1) and fuzzy variation 2 (FV 2) are developed in the present research (Fig. 1). In FV 1, the dough thickness and surface cracks (obtained using inline measurement technique - refer sections 4.1 and 4.2) are selected as input parameters. FV 1 evaluates these input parameters against a set of defined rules in order to select the most efficient output parameters, which form the input for the next rolling step. The addition of flour and roll gap are the two output parameters. FV 2 is similar to FV 1 except for the fact that it considers rheological measurements (offline measurements e refer section 4.3) as an extra input parameter. The rules for these two variations of fuzzy system are based upon the findings of the image processing and rheological measurements (Mahadevappa et al., 2015). They are programmed using the software MATLAB (Fuzzy Logic Toolbox). The complete outlook of the setup for the dough sheeting process which is a combination of the current research and the setup explained in the paper by (Mahadevappa et al., 2015) is depicted in Fig. 2. 4. Input parameters Decision making in fuzzy logic is based upon the input parameters. The input parameters used in this research are dough thickness, surface cracks and rheological measurements. 4.1. Dough thickness Dough thickness is the thickness of the dough obtained when sheeted between two rollers with a specific roll gap (distance between the two rollers). The fuzzy system should output a suitable rolling profile for every dough being rolled. The belt speed of the sheeting machine is 26.4 cm/s. The first rolling step of 45 mm as well as the final thickness of 5 mm is achieved by default. The intermediate rolling steps are oriented between a gentle and a consistently stressful program depending on the nature of the dough. Based on the roll gap, there are four membership functions for the dough thickness input: thick, medium, thin and final thickness (Table 4). The dough thickness input is in the range of 5 mme45 mm.
Table 3 Representation of the basic predefined rolling programs. Sheeting program
Step 1
Step 2
Step 3
Step 4
Step 5
Step 6
Step 7
Step 8
Program 1 (P1) Program 2 (P2)
45 mm 45 mm
35 mm 30 mm
25 mm 15 mm
15 mm 5 mm
10 mm 5 mm
8 mm
6 mm
5 mm
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
95
Fig. 1. Overview of the fuzzy control system developed.
Fig. 2. Process flow diagram.
Table 4 Membership functions for the inputedough thickness. Membership functions
Range (mm)
Thick Medium Thin Final thickness
45e21.5 21.5e10 10e5 5
4.2. Surface cracks The surface cracks on laminated wheat dough are undesirable and generally affect the final quality of the baked products. Surface
cracks formed during the sheeting process are assessed using an inline measurement technique capable of automatic detection and evaluation of dough surface as well as its measurement in terms of crack ratio (Mahadevappa et al., 2015). Crack ratio is the percentage of the cracked surfaces to the total surface area of the dough. This crack ratio obtained forms the input (surface cracks) to the fuzzy variations FV 1 and FV2. The crack ratio for a very good dough surface is less than 0.12 and for an extremely cracked dough surface the value is greater than 0.45. Based on the crack ratio, there are five input membership functions for the surface crack input: none, low, slight, medium and high (Table 5). The surface cracks input is in the range of 0e1.5.
96
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99 Table 5 Membership functions for the inputesurface cracks. Membership functions
Crack ratio
None Low Slight Medium High
<0.12 0.12e0.20 0.20e0.32 0.32e0.45 >0.45
4.3. Rheological measurements In order to establish a fuzzy system for efficient rolling process of wheat dough, focus is also on identifying the rheological properties of dough mixtures with yeast (ciabatta) and without yeast (puff pastry). To assess the dough processability, additives such as oxygen enriched water and cysteine are used. Addition of these additives effect the rheological properties in both negative and positive manner. The interpretation of rheological oscillation and rotation measurements is intended to draw conclusions on the dough processability. To determine the rheological properties of the wheat dough, oscillation and rotation measurements are carried out using a rheometer. Based on the oscillation measurements, input sets for the fuzzy system could not be defined as the results are relatively in a smaller range. Rotation measurements are done with a shear rate of 2 1/s in order to determine viscosity with respect to time (Fig. 3). Here the viscosity increases to maximum and then starts decreasing. From this point of time the dough begins to flow. In contrast to the oscillation measurements, a significant difference between various types of dough can be evidently recognized. As a result, rotation measurements (viscosity vs time) were used as an additional input parameter in fuzzy variation 2. At a shear rate of 2 1/s dough prepared with Keks flour, yeast and 0.01% cysteine displayed higher viscosity implying a good quality dough. On the contrary, dough prepared with Keks flour, yeast and
oxygen enriched water showed lower viscosity implying a low dough quality. The dough classification is primarily based on the viscosity values obtained during experimental procedures and the user's perspective (stickiness, vision). It is categorized into good, medium, poor and bad (refer Fig. 3). The dough with higher viscosity, good machine processability, smooth and crack free surface is considered to be good. Whereas, the sticky dough with low viscosity values, poor processability and high surface irregularities (cracks) is considered to be bad and not acceptable. The dough classified into medium and poor do not necessarily matchup to the parameters of a good dough, but are also not in the bad category of the dough so as to be discarded. The mid-range viscosity values infer that the medium and poor quality dough are also processable. In addition, opinion of expert bakers over the viscosity values and the experimental results are also considered for the above classification. Based on this categorization of viscosity, there are four input membership functions for the time measurements: good, medium, poor and bad (Table 6). 5. Output parameters The output parameters of the developed fuzzy system are mainly the control variables. These indicate to the user, the next steps to be followed in order to increase the efficiency of the dough processing.
Table 6 Categorization of membership functions based on viscosity values. Membership functions
Viscosity range
Good Medium Poor Bad
>6616 4900e6616 3400e4899 <3400
Fig. 3. Graphical representation of rotation measurements (Viscosity vs Time) with a shear rate of 2 1/s for different dough compositions.
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
5.1. Addition of flour Addition of flour also known as dusting flour, helps to reduce the stickiness of dough and also the formation of the surface cracks. It is generally observed that, in bakeries the stickiness of the dough is compensated by sprinkling of flour. How much flour is to be added is a matter of experience, it can vary from 30 g to 113 g. It is important to note here that often more flour than necessary is applied in bakeries to prevent adhesion of the dough to the machine. Excessive addition of dusting flour can also lead to a reduction in dough quality. As a step towards optimization, this variable is set up as an output parameter, through which we can get a dough profile with a final thickness of 5 mm in a few steps thereby limiting the excessive usage of flour. The addition of flour is gauged in three categories namely, no flour, little amount of flour (half a fistful ~ 30 g) and medium amount of flour (fistful ~ 75 g). Based on this categorization, there are three output membership functions for the addition of flour: none, little and medium. 5.2. Roll gap This output parameter provided by the fuzzy system is the extent by which the next rolling step needs to be reduced. For example, if the dough is rolled with the first rolling step of 45 mm, and 14 mm is the roll gap given as an output by the fuzzy system then the next rolling step is 45 mm-14 mm ¼ 31 mm. 6. Implementation of the fuzzy logic rules Mamdani's fuzzy inference method is used in this implementation (supported by MATLAB). Once the input and output parameters are set the next step is to connect them with the IFTHEN conditional rules, so that an inference can be made. The implementation of the fuzzy system is based on valid if-then rules. The input and output parameters are defined using trapezoidal membership functions.
97
Once a good quality dough surface is achieved, there is no further necessity to either add any flour or to roll the dough further. This rule could be represented as below: If (dough thickness ¼ final thickness) and (surface cracks ¼ slight) then (addition of flour ¼ none) and (roll gap ¼ 0 mm) Thus the first variant of fuzzy system helps to achieve a good quality dough surface by defining a set of fuzzy logic rules as show in the examples above. 6.2. Fuzzy system- variation 2 In FV 2 along with the dough thickness and surface cracks, rheological measurements (Time measurements) is also considered as an input parameter. Here again the addition of flour and roll gap form the two output parameters. In the below rule set, depending on the varying degree of surface cracks and time measurements suitable actions are taken with respect to the output parameters namely addition of flour and roll gap. If (dough thickness ¼ thick) and (surface cracks ¼ none) and (time measurements ¼ good) then (addition of flour ¼ none) and (roll gap ¼ 15 mm) If (dough thickness ¼ thick) and (surface cracks ¼ high) and (time measurements ¼ bad) then (addition of flour ¼ medium) and (roll gap ¼ 11 mm) If (dough thickness ¼ thick) and (surface cracks ¼ low) and (time measurements ¼ medium) then (addition of flour ¼ none) and (roll gap ¼ 14 mm) The below rule would suffice for a good quality dough achieving the desired thickness with negligible surface cracks. If (dough thickness ¼ thick) and (surface cracks ¼ slight) and (time measurements ¼ medium) then (addition of flour ¼ none) and (roll gap ¼ 0 mm)
6.1. Fuzzy system- variation 1 In FV 1, the dough thickness and surface cracks are selected as input parameters. This variation provides us two output parameters: addition of flour and roll gap. In FV 1, examples of the rules set for the dough with different crack surfaces are shown below. When the dough thickness is classified as thick (refer Table 4) and there are no surface cracks then it implies a good quality dough, hence there is no necessity of adding any flour and next rolling step can be drastically reduced to 15 mm. This can be represented by the following rule:
Thus the second variant of fuzzy system helps to achieve a good quality dough surface by defining a set of fuzzy logic rules by taking in an extra input parameter as show in the examples above. 7. Results and discussions In the presented research the test setup from the inline measurement technique (Mahadevappa et al., 2015) is used in conjunction with the fuzzy system developed using MATLAB (refer Fig. 2). 7.1. Analysis and assessment of the fuzzy control system
If (dough thickness ¼ thick) and (surface cracks ¼ none) then (addition of flour ¼ none) and (Roll gap ¼ 15 mm) When the dough thickness is classified as thick (refer Table 4) but there are varying degrees of surface cracks observed (slight, high) then it implies that the dough needs more processing. Hence there may be a necessity to add flour and the next rolling step is set accordingly. This can be represented by the following rules: If (dough thickness ¼ thick) and (surface cracks ¼ high) then (addition of flour ¼ medium) and (roll gap ¼ 11 mm) If (dough thickness ¼ thick) and (surface cracks ¼ slight) then (addition of flour ¼ none) and (roll gap ¼ 14 mm)
The two variations of fuzzy control system (FV 1 & FV 2) are tested on two types of dough: dough prepared using Keks flour, tap water without yeast (puff pastry) and dough prepared using Brema flour, tap water and yeast (ciabatta). According to the rheological measurements these two dough exhibited mid-range viscosity values. Two basic program cycles are considered as a reference for the fuzzy system (Table 3). These two programs provide the basic minimum number of steps required to achieve a good quality dough. Program 1 (P1) consisting of eight steps can be envisioned for a dough with very bad structure whereas, program 2 (P2) consisting of five steps can be considered for a dough having
98
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
medium to good quality structure. The job of the fuzzy control system is to optimize the sheeting program by regulating between these two programs. Thus giving a new optimized and efficient program (this program can consist a maximum of eight steps to a minimum of four steps depending on the dough quality). Both types of dough were rolled using FV 1 and FV 2. The sheeting profile of the tested dough types is shown in Fig. 4. In case of puff pastry both the fuzzy variations FV1 and FV2 exhibits similar sheeting profile (program). The dough sheeting process starts with the thickness of 45 mm followed by 30 mm, 15 mm and finally ends with 5 mm (four step program). In case of ciabatta, both the variations of fuzzy system showed different sheeting profiles with equal number of rolling steps (six step program). This infers that the fuzzy system selects the best optimum suited program based on the input quality of the dough. Adopting similar system in the bakeries can help the bakers to achieve optimum processing time. To understand the effects of FV1 and FV2, ciabatta dough results are considered as it showed two different sheeting profiles (programs). Here the difference between FV 1 and FV 2 is seen mainly in the values of the crack ratio obtained (Fig. 5). FV 1 processes the ciabatta dough in six rolling steps indicating good dough quality with the final crack ratio of 0.12. FV 2 also processes the ciabatta dough in six rolling steps again indicating a very good dough quality with the final crack ratio of 0.09. FV 2 displays a better crack ratio when compared to FV 1. This is mainly attributed to the rheological parameter that is taken into account by FV 2 during the dough sheeting process. From the above results obtained, it can be inferred that a fuzzy control system is capable of deciding the best possible program with least number of rolling steps depending on the dough quality. FV 1 and FV 2 are found to be working as designed to roll any type of dough, finally resulting in a smooth surface exhibiting least crack ratio. The fuzzy systems FV 1 and FV 2 were tested on two commonly used types of dough (puff pastry and ciabatta). However, it can be extended to test various other types of dough combinations with the limitation being that the dough consistency should be such that it is able to pass through the first rolling step. A tabulated summary on the effects of the four programs P1, P2, FV 1 and FV 2 on both the dough types with respect to the number
Fig. 5. Difference in the crack ratio between FV1 and FV2 on the ciabatta dough surface.
of rolling steps, time and crack ratio is presented in Tables 7a and 7b. The above results (Tables 7a and 7b) emphasize: FV 1 and FV 2 requires less time for rolling compared to P1 and P2. FV 1 and FV2 help in achieving a better or equal crack ratio compared to P1 and P2 but, with minimal steps. FV 2 displays a better crack ratio due to the fact that rheological measurement is taken into consideration. This design of fuzzy logic system can be extended to SMEs since, it provides adequate advantage over the existing manual processes. Implementation of such a fuzzy system results in achieving a good and preferred dough surface during dough processing stage. The advantages mainly being having an optimized method to roll the dough thus saving time and in turn saving energy and costs when compared to a manual setup. Generally in a manual setup the dough displaying high surface cracks is considered unfit for processing and is discarded. On the contrary, a fuzzy control system will analyze the dough surface and help to reduce the surface cracks and increase the quality by the addition of an accurate amount of dusting flour. In this way the dough that would have been discarded is now modified and its usability is increased. This fuzzy system design provides an optimized, time saving and efficient control on the sheeting process compared to manual intervention which can lead to wastage and is also prone to manual errors such as decisions on the dough quality based on manual senses (vision and touch).
Table 7a Result summary for puff pastry. Puff Pastry
P1
P2
FV 1
FV 2
Rolling steps Rolling time Crack ratio
8 4min 20s 0.10
5 2min 36s 0.12
4 1min 40s 0.11
4 1min 50s 0.10
Table 7b Result summary for ciabatta.
Fig. 4. Representation of the sheeting profile of wheat dough using fuzzy control system.
Ciabatta
P1
P2
FV 1
FV 2
Rolling steps Rolling time Crack ratio
8 4min 15s 0.17
5 2min 35s 0.17
6 2min 30s 0.12
6 2min 40s 0.09
J. Mahadevappa et al. / Journal of Food Engineering 199 (2017) 93e99
8. Conclusions A novel method envisaged to improve the dough sheeting process in SMEs using a fuzzy control system is presented in this paper. The fuzzy system is essentially implemented as a closed loop control system where, the entire process is repeated until an acceptable value of the crack ratio is obtained. Two formats of fuzzy systems were developed (FV 1 & FV 2) differentiated by the input parameter sets. Considering the rheological properties of the dough as an input parameter gave an added advantage to comprehensively assess the dough. Developing a fuzzy system gives an edge in the sheeting process by effective time management and minimizing losses leading to an efficient dough processing method. The target group which would be benefited by implementation of this research are the small and medium scale baking industries. This approach would help them to save time, cost, minimize dough wastage and reduce manual errors. The market studies shows that more and more baking enterprises are moving towards adapting fuzzy control system. Although there are many successful fuzzy logic applications in baking enterprises, there is still a need to transfer and adapt it in the sheeting process of the wheat dough. This pioneering approach also caters to manufacturers of sheeting machine in enhancing their machine design. Acknowledgements The authors wish to thank the German Ministry of Economics and Technology (via AiF) and the FEI (Forschungskreis der Ernahrungsindustrie e.V., Bonn) for the financial support (AiF 16755 N) and the members of Bremerhaven Institute of Food Technology and Bioprocess Engineering (BILB) for their cooperation. The acknowledgment would be incomplete without thanking Nicola Cebulla €lfert for their assistance in accomplishing this work. and Kathrin Wo References Bloksma, A.H., 1990. Dough structure, dough rheology, and baking quality. Cereal Foods World 35, 237e244.
99
€ m, J.S., Lindskog, E., Sridhar, T., 2010. Computational Chakrabati-Bell, S., Bergstro modeling of dough sheeting and physical interpretation of the non-linear rheological behavior of wheat flour dough. J. Food Eng. 100, 278e288. Davidson, V.J., Ryks, J., Chu, T., 2001. Fuzzy models to predict consumer ratings for biscuits based on digital features. IEEE Trans. Fuzzy Syst. 9, 62e67. Dobraszczyk, B.J., 1997. The rheological basis of dough stickiness. J. Texture Stud. 28, 139e162. Dobraszczyk, B.J., Morgenstern, M.P., 2003. Rheology and the breadmaking process. J. Cereal Sci. 38, 229e245. Faridi, H., Faubion, J.M., 1990. Dough Rheology and Baked Product Texture. AVI Van Nostrand Reinhold, New York. Finaev, V.I., Sinajvskay, E.D., Pushnina, I.V., 2015. Fuzzy control model of temperature at the bread baking chamber. Izestiya SFedU. Eng. Sci. 150e157. Gangadharappa, G.H., Prabhasankar, P., 2011. Spreadsheet aided fuzzy model for prediction of chapati making quality. J. food Sci. Technol. 48 (3), 344e348. Ioannou, I., Perrot, N., Mauris, G., Trystram, G., 2004. Development of a control system using the fuzzy set theory applied to a browning processeetowards a control system of the browning process combining a diagnosis model and a decision modeleepart II. J. Food Eng. 64, 507e514. Kim, S., Cho, I., 1997. Neural network modeling and fuzzy control simulation for bread-baking process. Trans. ASAE 40, 671e676. Linko, S., Linko, P., 1998. Developments in monitoring and control of food processes. Food Bioprod. Process. 76, 127e137. Mahadevappa, J., Groß, F., Benning, R., Delgado, A., 2015. Development of an inline measurement technique to assess the quality of wheat dough during the sheeting process. J. Cereal Sci. 64, 183e188. Mittel, G.S., 1997. Computerized Control Systems in the Food Industry. Marcel Dekker, Inc, USA (New York, NY). Perrot, N., Trystram, G., Le Guennecb, D., Guely, F., 1996. Sensor fusion for real time quality evaluation of biscuit during baking. Comparison between Bayesian and fuzzy approaches. J. Food Eng. 29, 301e315. Perrot, N., Trystram, G., Chevrie, F., Guely, F., Schoesetters, N., Dugre, E., 2000. Feedback quality control in the baking industry using fuzzy sets. J. Food Process Eng. 23, 249e279. Quail, K.J., McMaster, G.J., Tomlinson, J.D., 1990. Effect of baking temperature/time conditions and dough thickness on arabic bread quality. J. Sci. Food Agric. 53, 527e540. Sharma, N., Hanna, M.A., Chen, Y.R., 1993. Flow behavior of wheat flour-water dough using a capillary rheometer. I. Effect of capillary geometry. Cereal Chem. 70, 59e63. Singh, K., Mishra, A., Mishra, H., 2012. Fuzzy analysis of sensory attributes of bread prepared from millet-based composite flours. LWT - Food Sci. Technol. 48, 276e282. Young, L.S., 2007. Application of baking knowledge in software systems. In: Cauvain, S., Young, L.S. (Eds.), Technology of Breadmaking. Springer, New York, pp. 207e222. Zhang, Q., Litchfield, J., 1993. Fuzzy logic control for a continuous crossflow grain dryer. J. Food Process Eng. 16, 59e77.