Soft-starter Controllers based on Neural Network for Induction Motor Drive System

Abstract

In industry, three-phase Induction Motors (IMs) are the most often utilized motors. The starting process of IMs can cause many unnecessary energy wastes because of the high starting supply current and huge accelerating torque. This research study proposes two Artificial Neural Network (ANN) based soft-starter control approaches to address IM starting issues. Current and torque control schemes are to keep IM starting current and accelerating torque constant at a specified level, respectively. This is achieved by properly selecting the thyristor firing angles within the soft starter. The intricacy of online firing angle determination is tackled using the ANN. The ANN models for the two control techniques are developed using the MATLAB Neural Network Toolbox. For a 3 kW three-phase IM, the results confirm the soundness and efficacy of control schemes. Moreover, the advantages of the proposed intelligent soft-starter controllers compared with several soft-start control techniques in literature are low computational burden for the control algorithm and high accuracy. Here, ANN is used to address the challenging problem of determining the optimal thyristor firing angles online. In addition, using the ANN results in an open-loop control configuration. As a result, the controllers' algorithms can be implemented industrially using a low-cost processor.