Fuzzy systems in permanent magnet motors description

Abstract

Several artificial intelligence techniques for fuzzy modelling of flux distribution, electromagnetic and disturbance (friction, ripples) torques in permanent magnet rotational motors are proposed. Based on a model taking into account several nonlinear phenomena such as non-sinusoidal flux linkage, saturation effects etc observer-based parameter identifiers approach is used to plan identification experiments and to obtain the data. Next fuzzy models are applied to approximate the experimental data. Several training algorithms tuning the fuzzy model parameters and to simplifying the model structure are compared.