Inverse Optimal Control of Electric Vehicle Cabin Heating Loop via Real Time Data-Supported Physics-Based Model

The key contribution of this study lies in the first application of Inverse Optimal Control (IOC) to the cabin heating loop and overall thermal management of electric vehicles (EVs), utilizing a real-time data-supported physics-based model that accounts for varying ambient conditions. IOC is implemented as a model-based optimal control strategy that seeks the most general objective function while ensuring the dynamic system's asymptotic stability. When the nonlinear system exhibits an input-affine structure, the objective function can be formulated as a Control Lyapunov Function (CLF), reducing the problem to the identification of a symmetric positive definite matrix. We demonstrate that the cabin heating loop model approximates an input-affine structure, with one non-affine term addressed via a Taylor series expansion. Experimental data collected from a prototype EV at three different ambient temperatures are used for model development and parameter identification. IOC is applied to each temperature-specific model by optimizing a positive definite matrix using the Big Bang–Big Crunch (BB-BC) global optimization algorithm, with the goal of minimizing tracking error. The performance of the proposed controller is compared against a Linear Quadratic Regulator (LQR) and a Linear Model Predictive Controller (LMPC) across all temperature conditions. Results indicate that IOC outperforms both controllers in terms of disturbance rejection, reference tracking and transient performance, while maintaining comparable energy consumption. Notably, IOC achieves faster settling times and reduced overshoot, demonstrating superior overall control performance under diverse environmental conditions.