Multifaceted Velocity Prediction-Based Bipartite Integration Optimization Strategy of Cabin and Battery Thermal Management for EVs

In high-temperature environments, integrating dynamic traffic information with an efficient thermal management optimization strategy has proven effective in rapidly reducing battery temperature, keeping it within the optimal operational range. This approach ensures electric vehicles (EVs) maintain optimal power output and range performance. However, due to the prolonged battery heat accumulation, effective optimization requires a long predictive horizon. The real-time implementation of centralized model predictive control (MPC) faces challenges in computational complexity. To address this, the proposed bipartite integration optimization strategy combines two levels for cabin and battery thermal management, enhancing computational efficiency through a segmented, hierarchical optimization approach. A comprehensive thermal model for the cooling system is developed, employing a combination of liquid cooling and active air cooling. In the upper level, vehicle-to-cloud (V2C) communication is utilized to obtain the average vehicle velocity over a long horizon, enabling more accurate forecasts of power demand and thermal load, thereby optimizing the thermal trajectory. In the lower level, the extreme learning machine (ELM) method is used to predict future vehicle velocity over a short horizon, facilitating precise temperature tracking and minimizing variations for optimal control. Simulation results under real driving conditions indicate that the proposed strategy reduces energy consumption by 18.90%, improves computational efficiency by 83.53%, and results in a 0.055% improvement in battery state of health (SOH) over extended cycles, without compromising the cooling requirements of the passenger cabin.