Improved flux pattern by third harmonic injection for multiphase induction machines using neural network

This paper presents a modified V/f control strategy using neural network with an improved flux pattern using third harmonic injection for multiphase induction machines. The control objective is to generate a nearly rectangular air-gap flux, resulting in an improved machine power density for the required speed range. If just a proportional relation is used between the third harmonic and fundamental plane voltage magnitudes with zero phase shift, variable misalignment between fundamental and third air-gap flux components occurs with varying mechanical loading as a result of stator voltage drop. Due to this misalignment, saturation may take place which affects the total flux and increases machine iron losses. Neural network is used to obtain the required injected voltage phasors magnitudes and angles to ensure that the air-gap flux is near rectangular with a maximum value of 1 pu for all loading conditions. Simulations are carried out on an eleven-phase induction machine to validate the proposed controller using MATLAB/Simulink.