Novel physics informed neural network based power-thermal integrated control for extended-range electric tractor

Power and thermal systems are critical components of hybrid tractors, where effective energy and thermal management can substantially enhance both fuel economy and thermal performance. However, the asynchronous control timescales of energy and thermal management may lead to performance degradation under prolonged high-load operations. Thus, this article presents a power-thermal integrated control strategy based on a physics-informed neural network (PINN). First, a bidirectional power-thermal coupling model is established to clarify the mutual influence relationship between power and thermal systems. Then, with the model predictive control strategy of multi-structured physics-informed neural network (MPINN-MPC), a power-thermal integrated control strategy is developed to synchronize power and thermal management within a unified time domain, which can optimize the energy flow distribution, improve the fuel economy, and maintain the critical components within the best thermal states. Finally, the effectiveness as well as performance of the proposed integrated control strategy is verified by using comparative hardware-in-the-loop tests.