基于多群体多目标灰狼优化器的电动汽车热管理系统多目标优化

Abstract The optimization of thermal management system design parameters is critical for enhancing overall energy efficiency and dynamic performance of electric vehicles. In this study, a high-fidelity vehicle simulation model incorporating a detailed thermal management system was constructed, and the orthogonal experimental design was employed to systematically explore design space. On this basis, a high-accuracy surrogate model was developed using extreme gradient boosting, and the many-population many-objective grey wolf optimizer was proposed to navigate the complex multi-dimensional pareto front. The results indicate that extreme gradient boosting model exhibits robust generalization capability with an average prediction error below 3% across all optimization objectives. The many-population many-objective grey wolf optimizer demonstrates superior performance compared to conventional benchmarks, achieving rapid convergence and a well-distributed solution set, as evidenced by optimal hypervolume and spacing metrics. The final optimized solution leads to a 27.08% reduction in compressor power consumption, a 43.62% increase in coefficient of performance and a 5.23% extension in driving range, with validation errors maintained below 2% relative to high-fidelity simulations. The performance analysis of final solution reveals that optimizing compressor displacement, enlarging refrigerant circuit diameters, and fine-tuning heat exchanger dimensions significantly improve energy efficiency. These findings provide in-depth insights into the energy flow regulation mechanisms within thermal management system architectures.