A PRACTICAL FUZZY LOGIC APPROACH IN THE STUDY OF FOOD PROCESSING PARAMETERS

Fuzzy Logic (FL) is a technique used in industry for predictive tasks due to its simplicity and ability to solve problems quickly and accurately. FL applies logical rules on fuzzy sets named with linguistic labels. In the study of food processing parameters, FL was used to propose predictive models by specific procedures for each process of interest. The aim of this work was to evaluate the predictive performance of a practical FL approach applicable to different food processing parameter studies. This FL approach was run in MATLAB R2023a software and applied to data from ten available experimental studies of food processing parameters that used factorial design and response surface methodology. The application of this approach required two steps: (1) fuzzification of variables, by creating fuzzy sets with their respective linguistic labels, where the fuzzy sets of the input variables were defined by the levels explored in the experimental design, while, those of output variables were delimited by segments of the observed range, (2) defuzzification with Mamdani inference system, where each fuzzy rule (if - then) was created according to each treatment explored in the experimental design. The predictive performance of this approach was evaluated by the coefficient of determination (R2) and the standard error of prediction (SEP (%)), which were determined by comparing experimental values and predicted values of the output variables. The results showed that this approach has high predictive performance (R2 > 90% and SEP < 10%) in the study of food processing parameters. Since this approach generates valid models complementary to those obtained by experimental designs, it becomes relevant for the intuitive understanding of processes through linguistic labels and fuzzy rules that imitate human reasoning. Thus, this approach represents a contribution to the study and control of processes in the food industry 4.0 environment.