REACTIVE POWER MODEL REFERENCE ADAPTIVE SPEED-SENSORLESS SYSTEM WITH DIRECT TORQUE CONTROL TUNED WITH FUZZY NEURAL NETWORKS FOR IMPROVED SPEED CONTROL IN INDUCTION MOTOR DRIVES

Abstract

This article investigates a reactive power based Model Reference Adaptive System (Q-MRAS) with Fuzzy Neural Network (FNN) facilitated Direct Torque Control for improved speed control of an Induction Motor Drive. A key component of the conventional DTC control scheme is the use of PID controllers and its derivatives. PID controllers were found to be a source of ripples. To mitigate its effect, in this work, a novel idea of replacing all the PI controls in the conventional speed control DTC models with FNN based controllers is investigated. The goal is to reduce complexity, reduce ripples in speed and boost low speed operation suitable for industrial needs. The proposed FNN controllers-based model is implemented using Matlab/Simulink. Comparisons are made between the results obtained from the proposed model and that from conventional models (Switching Table based DTC and SVM-DTC). Results showed that in the proposed FNN based DTC model, low speed operation (100 rpm) had 0.13 % speed ripple compared to 0.94 % in the conventional Switching Table based DTC and 0.23 % in the conventional SVM-DTC. This represents a reduction in speed ripple by 86.17 % in the proposed scheme compared to the Switching-Table DTC and by 43.48 % in the proposed scheme compared to the SVM-DTC scheme.