ANALYSIS OF NEURAL SYSTEMS WITH FUZZY LOGIC

The article examines the efficiency of fuzzy neural networks (FNN) compared to conventional systems based on crisp and fuzzy logic. The study focuses on the practical feasibility of implementing hybrid neuro-fuzzy models in industrial process control. A computer simulation of an automatic control system was used, incorporating crisp, fuzzy, and neuro-fuzzy control methods. The ANFIS (Adaptive-Network-Based Fuzzy Inference System) architecture was analyzed as a key model, demonstrating adaptive fuzzy inference principles. The results indicate that neuro-fuzzy systems provide comparable control accuracy to traditional methods while exhibiting lower error rates and slower adaptation speeds. The practical significance of the research lies in justifying the choice of neuro-fuzzy systems for automating complex technological processes.