Modelling and design of a modified neuro-fuzzy control-based IM drive via feedback linearization

Abstract

This research paper presents a design of a simplified structured adaptive neuro-fuzzy controller (NFC) technique with an intuitive feedback linearization controlled induction motor (IM) model for extensive torque and speed ripple minimization with better performance enhancement of IM drive. The non-linear dynamics of IM is modeled and simulated based on state space linearization technique in the stationary reference frame. The proposed simplified adaptive NFC is the fusion approach of fuzzy logic and neural network with one input as torque error unlike conventional two-input NFC as torque error and change in torque error. It also improves the computational efficiency by making the structure very simple and robust as compared to the conventional NFC, thereby easy to apply in a realistic environment. The effectiveness and execution of the proposed control technique based linearized IM drive is investigated in MATLAB environment in various operating conditions and is contrasted with the conventional two-input NFC as well as PI-controller to analyze the superior performance of IM drive. The system is also implemented in real-time system using DSP 2812 to validate the different control strategies.