Bearingless synchronous reluctance motor integrates torque system and suspension system. The key performance of both systems is affected by many rotor structure parameters

and the traditional multi-objective optimization strategy will face the problem of inefficient optimization caused by high dimensional design variables. To address the above problems

this paper proposes a sensitive classification parameter optimization strategy. The optimization strategy firstly carries out global sensitivity analysis through Sobol method to select the design variables with higher sensitivity. Then

according to the different influence laws presented by the sensitive parameters on the key performance

the coupled sensitive parameters and single-system sensitive parameters are defined

so as to realize the classification and dimensionality reduction of the sensitive parameters. Further

according to the different degree of sensitivity

the classified sensitive parameters are integrated into corresponding agent models

and combined with the second-generation non-dominated genetic algorithm (NSGA-Ⅱ) to carry out the multi-objective hierarchical optimization design. A fuzzy aggregation strategy is then employed to select the solution with the best overall performance from the resulting Pareto front. Finally

the key performances of the torque system and suspension system of the motor before and after optimization are simulated and compared

and an experimental platform is built to test the performance of the optimized prototype. The results verify the feasibility of the optimization strategy.