Artificial Neural Network Based Rotor Capacitive Reactance Control for Energy Efficient Wound Rotor Induction Motor

Abstract

Problem statement: The Rotor reactance control by inclusion of external capacitance in the rotor circuit has been in recent research for improving the performances of Wound Rotor Induction Motor (WRIM). The rotor capacitive reactance is adjusted such that for any desired load torque the efficiency of the WRIM is maximized. The rotor external capacitance can be controlled using a dynamic capacitor in which the duty ratio is varied for emulating the capacitance value. This study presents a novel technique for tracking maximum efficiency point in the entire operating range of WRIM using Artificial Neural Network (ANN). The data for ANN training were obtained on a three phase WRIM with dynamic capacitor control and rotor short circuit at different speed and load torque values. Approach: A novel neural network model based on the back-propagation algorithm has been developed and trained in determining the maximum efficiency of the motor with no prior knowledge of the machine parameters. The input variables to the ANN are stator current (I_s), Speed (N) and Torque (T_m) and the output variable is the duty ratio (D). Results: The target is pre-set and the accuracy of the ANN model is measured using Mean Square Error (MSE) and R² parameters. The result of R² value of the proposed ANN model is found to be 0.99980. Conclusion: The optimal duty ratio and corresponding optimal rotor capacitance for improving the performances of the motor are predicted for low, medium and full loads by using proposed ANN model.