Objective To address the issue of optimization difficulties arising from the numerous structural parameters of the stator permanent magnet PM doubly salient machine a multi-parameter and multi-objective optimization method based on the improved non-dominated sorting genetic algorithm Ⅱ is proposed. Methods First the parameter sensitivity was calculated according to multiple optimization objectives. Then the comprehensive sensitivity was computed through the set weight coefficients. Finally the structural parameters of the stator PM doubly salient machine were divided into three layers based on the comprehensive sensitivity. The structural parameters in the first and second layers were highly sensitive and they were optimized by fitting a Gaussian process regression model with a constant basis function and a quadratic rational kernel function. The structural parameters in the third layer were of lower sensitivity and were optimized using the single-parameter scanning method. Six schemes were established to compare the effects of different weight coefficients and thresholds on the system optimization objectives. Results The improved non-dominated sorting genetic algorithm Ⅱ outperforms the traditional algorithm. After optimization the electromagnetic torque of the stator PM doubly salient machine increased by 15. 06% the cogging torque decreased by 50. 9% and the torque ripple reduced from 20. 23% to 9. 45%. Conclusion The feasibility and effectiveness of the proposed multi-objective optimization method are verified by the finite element simulation results.