Multi-objective optimization of thermal management parameters for a fuel cell hybrid heavy-duty truck using machine learning and GT-SUITE simulation

This study presents a systematic optimization framework for Fuel Cell Hybrid Electric Vehicles (FCHEVs). The framework integrates experimental testing, high-fidelity simulation, machine learning, and multi-objective optimization. It addresses the complex nonlinear coupling between thermal management parameters and key vehicle performance indicators. Experimental work includes energy flow bench tests and chassis dynamometer tests on a heavy-duty FCHEV. These tests provide vehicle energy consumption characteristics. A full-vehicle simulation model is built in GT-SUITE. The model includes the powertrain, multi-loop thermal management, and a rule-based energy management strategy. Its accuracy is validated against experimental data. Design of Experiments (DOE) simulations cover 400 operating conditions. Three dominant factors are identified: battery coolant temperature (25–35 °C), cabin air conditioning temperature (18–24 °C), and fuel cell coolant temperature (70–85 °C). Two machine learning models are developed for performance prediction: Backpropagation Neural Network (BPNN) and Support Vector Regression (SVR). The SVR model shows superior accuracy. On the test set, it achieves R 2 values of 0.9977 for efficiency prediction and 0.9978 for hydrogen consumption prediction. Multi-objective optimization is performed using the Non-dominated sorting genetic algorithm II (NSGA-II). The population size is 100, and the number of generations is 200. The optimal parameter combination is identified as 85 °C for fuel cell coolant, 35 °C for battery coolant, and 23 °C for cabin air conditioning. This combination improves fuel cell stack system efficiency from 47.10% to 49.14%, a relative increase of 4.33%. It also reduces hydrogen consumption per 100 km from 9.04 kg to 8.37 kg, a reduction of 7.41%. The proposed framework provides a practical and systematic methodology for thermal management design and performance enhancement in FCHEVs.