Dynamic identification and control of IM soft-start using ANN

Soft-start and effectively running of thyristor controlled induction motor (IM) system with light-loads has been proved to save energy and improved the efficiency. The complex characteristics of IM often bring difficulties to electrical driven system. Due to its nonlinearity, strong coupling, high orders and uncertainties, only are studied the input and output characteristics in conventional applications. From the system point of view, during the course of soft-start, shaftpsilas pulsations make the dynamic performance worse. Traditional system analysis and modeling still completely depend on steady circuit calculation, the result of which leads to weakness of dynamic performances. This paper has listed open loop and close loop modes to control IM, has proposed a method of combining artificial neural network (ANN) identification with RBF network and dynamic information acquisition, and has present a modified algorithm to suppress stator current pulsation. Comparison of simulation results, experiment data and actual waveform, the validity of the design has been verified. It has provided theoretic support to industrial applications.