Incremental Learning Strategy-Assisted Multi-Objective Optimization for an Oil–Water Mixed Cooling Motor

Abstract

Abstract As the core component of electric vehicles (EVs), the performance of motors affects the use of EVs. Motors are sensitive to temperature, and overheated operating temperature may cause the deterioration of the magnetic properties and the reduction of efficiency. To effectively improve the heat dissipation of the motor, this work presents an incremental learning strategy-assisted multi-objective optimization method for an oil–water mixed cooling induction motor (IM). The key parameters of the motor are modeled parametrically, and the design of the experiment is carried out by the Latin hypercube method. The incremental learning strategy is used to improve the low accuracy of the surrogate model. Four multi-objective optimization algorithms are used to drive the optimization process, and the optimal cooling system parameters are obtained. The reliability of the proposed method is verified by motor bench experiments. The optimization results suggest that the maximum temperature of the motor is reduced by 5 K after optimization, and the heat dissipation of the motor is improved effectively, which provides a theoretical basis for further promotion and improvement of the induction motor.