Cogging force and its estimation using a neural network based on 2D field model of PMLSM

This paper analyzes the cogging force of a permanent magnet linear synchronous motors (PMLSM) using finite element method (FEM) based on 2D field model. A neural estimator of cogging force basing on the slot shape optimization is presented. To investigate the effects of cogging force, the force ripples of different slot shapes, skew effect with different skew angles, diverse close slot, varied air gap and fractional slot are compared respectively. Force ripple coefficient is defined to describe the cogging force. The motion of 2D field model has been assumed and the nonlinearity of magnetic saturation of the PMLSM has been taken into consideration while modeling. Simulations evaluations show that fractional slot is most efficient and convenient to depress the cogging force of PMLSM. Suggested estimator is put up as a neural network based on BP algorithm. With the training sets of FEM calculations, the neuron estimator will evaluate the slot shape and air gap for optimization of the cogging force.