An Adaptive Soft Computing Model for Flux Estimation and Torque Control of Induction Motors

Abstract

Induction motors are widely utilized in various industries because of simplicity in service, durability, and low cost, leading to the displacement of DC motors. However, control system challenges have hindered their full potential for high performance. Many low-performance drives utilize scalar control, which adjusts only the stator magnitudes to uphold a persistent stator flux. Vector control emerged to address this limitation by enabling induction motor control analogous to DC motors. Subsequently, Direct Torque Controlled (DTC) induction motor drives were developed. Unlike vector-controlled drives where stator currents serve as control variables, DTC controls stator flux linkages. DTC employs a Reference Flux (RF) estimator to determine RF based on motor speed and a PI controller to estimate reference torque using speed error as input. The calculated stator flux angle determines the sector number for generating switching signals, traditionally done through a RF estimator. This estimator has been enhanced with fuzzy logic to ensure adaptability, and further improved with an ANFIS-based approach to combine fuzzy logic and artificial neural networks. Performance evaluation involves metrics utilizing speed and torque errors to gauge controller effectiveness. Comparing the fixed RF estimator to the ANFIS-tuned RF estimator with manually tuned speed PI, there is a performance improvement of 72.73%, 72.749%, and 46.636% for ISE, ITSE, and ITAE, respectively.

Keywords - Direct torque controller, reference flux, induction motor, ANFIS, vector controller.