Abstract
Thermal management in forced air-cooling asynchronous traction motor (ATM) is currently bottlenecked by the limitations of traditional axial ventilation architecture and independent cooling design for the stator and rotor, which restrict further escalations in cooling efficacy and lightweight. This paper proposes a collaborative dual-stage strategy to optimize stator and rotor axial ventilation ducts (AVDs) simultaneously. By spatially decoupling the design parameters of both components, the strategy facilitates an optimization transition from uniform cross-section (UC) to variable cross-section (VC) topologies from a global multi-physics perspective. This dimensionality reduction framework effectively alleviates computational overhead in computational fluid dynamics (CFD) simulations and reduces surrogate model training costs. At Stage 1, the circumferential parameters of the stator and rotor UC-AVDs are optimized via a multi-objective approach that combines stratified hybrid sampling, multiple surrogate models, the NSGA-II algorithm and TOPSIS decision-making to trade off hydraulic and thermal performance. Stage 2 subsequently leverages orthogonal design and ANOVA to maximize the performance evaluation criterion (PEC) by seeking optimal axial parameters of stator VC-AVDs. The results demonstrate that the proposed approach achieves a 14.1°C reduction in the stator winding hotspots alongside an 8.44% decrease in total pressure drop. These findings indicate a significant optimization of the internal flow field, offering a robust solution for cooling enhancement in high-power traction motors.