OPPORTUNITIES FOR USING FUZZY SYSTEMS AND NEURAL NETWORKS TO OPTIMIZE QUALITY OF DRIED MATERIALS WITH COMPLEX RHEOLOGY

摘要

The paper deals with application of neural networks for forecasting of the quality ofrnmaterials with complex rheology. Fuzzy sets are used here to estimate a discrepancyrnimposed when modeling processes of drying and thermal treatment. The examples of arnfuzzy neural network used to predict quality of reinforcing cord made from syntheticrnfibres are given. The possibility for using a similar approach to analyze dispersedrnsystems such as particulate bio-products is pointed out. The advantages and drawbacksrnof three different approaches (the physical- mathematical modeling and therncorresponding approximation of quality indices, polynomial approximation, and neurofuzzyrnmethods) are compared.