Enhancing Sealing Performance Predictions: A Comprehensive Study of XGBoost and Polynomial Regression Models with Advanced Optimization Techniques.

Motors, as the core carriers of pollution-free power, realize efficient electric energy conversion in clean energy systems such as electric vehicles and wind power generation, and are widely used in industrial automation, smart home appliances, and rail transit fields with their low-noise and zero-emission operating characteristics, significantly reducing the dependence on fossil energy. As the requirements of various application scenarios become increasingly complex, it becomes particularly important to accurately and quickly design the sealing structure of motors. However, traditional design methods show many limitations when facing such challenges. To solve this problem, this paper proposes hybrid models of machine learning that contain polynomial regression and optimization XGBOOST models to rapidly and accurately predict the sealing performance of motors. Then, the hybrid model is combined with the simulated annealing algorithm and multi-objective particle swarm optimization algorithm for optimization. The reliability of the results is verified by the mutual verification of the results of the simulated annealing algorithm and the particle swarm optimization algorithm. The prediction accuracy of the hybrid model for data outside the training set is within 2.881%. Regarding the prediction speed of this model, the computing time of ML is less than 1 s, while the computing time of FEA is approximately 9 h, with an efficiency improvement of 32,400 times. Through the cross-validation of single-objective optimization and multi-objective optimization algorithms, the optimal design scheme is a groove depth of 0.8-0.85 mm and a pre-tightening force of 80 N. The new method proposed in this paper solves the limitations in the design of motor sealing structures, and this method can be extended to other fields for application.