Online estimating voltage source inverter nonlinearity for PMSM by Adaline neural network

Abstract

This paper investigates how to online estimate the voltage source inverter (VSI) nonlinearity by Adaline neural network (ANN) in a permanent magnet synchronous machine (PMSM) drive system. The proposed estimation includes the estimation of PMSM stator winding resistance, inductance and rotor flux linkage and can follow the VSI nonlinearity variation due to the variation of PMSM working condition. Compared with existing literatures, the proposed method is fit for id=0 control and does not need the nominal value of any PMSM parameter and will not suffer from the mismatching between actual parameter value and nominal parameter value. The estimated distorted voltage value due to VSI nonlinearity is used for compensating the drive system and the experimental result shows that it can improve the control performance more significantly compared with existing VSI nonlinearity compensation method.