Performance Improvement of a Split Winding Bearingless Induction Machine Based on A Neural Network Flux Observer

This work describes the performance improvement of a 4 poles 1,1 kW split winding bearingless induction machine based on a neural rotor flux observer. The system control executes simultaneously the vector speed control, the radial positioning control and the stator winding current control. The neural flux observer compensates for parameter variations due to temperature changes or due to the rotor magnetic saturation. A program developed in the ANSI C language executes all the system control. The used DSP resources are the Analog/Digital converters, the PWM outputs and the parallel and RS-232 serial interfaces, which are responsible, respectively, for the DSP programming and the data capture through the supervisory system. Comparisons with a fixed parameter observer validate the proposed method.