In this paper, a new structure of moving magnet type(MMT) permanent magnet synchronous linear motor is proposed. The thrust performance (trust and trust fluctuation) of permanent magnet synchronous linear motor(PMSLM) is improved by changing the structure parameters of the motor. Firstly, the topological structure of the MMT-PMLSM is designed, and the main structural parameters affecting the MMT-PMSLM are analyzed by using semi-analytical network.Then, the long and short term memory network(LSTM) regression modeling method is introduced, and the cuckoo search algorithm is used to adjust the super parameters of the regression modeling. Based on the mapping relationship between motor structure parameters and thrust performance,a regression model is builded.Secondly,The grey wolf optimization algorithm is used to iteratively optimize the regression model, the optimal structural parameters are obtained by taking thrust lifting and reducing thrust fluctuation as the objectives. Finally, the simulation results verify the effectiveness of the motor structure optimization design method to improve the thrust performance.

Deep Learning，optimization algorithm，permanent magnet linear synchronous motor