Design induction machines' controller in soft-start based on neural networks

Oriented to soft starter driven induction machines, the values of open-circuit voltage, stator current and the harmonics' line spectrum were accquired by means of circuit calculations and Fourier analysis. The neural network of voltage current-back electromotive force(BEMF) speed were trained by simulation data of voltage slope soft-start, and the identified values of BEMF and speed were obtained by exciting the network with soft-start mode's data. A new control pricinples containing speed's information were provided to realize motor's soft start. Simulation results indicate it can not only save more energy, but also reduce suppress output torque's viberations at the beginning of start.