Energy management strategy with model prediction for fuel cell hybrid trucks considering vehicle mass and road slope

Abstract

Trucks frequently encounter significant fluctuations in transport loads and operate on roads with complex gradients. Traditional energy management strategies, which focus solely on vehicle speed, often fail to optimize energy utilization, resulting in high comprehensive operating costs, particularly for fuel cell hybrid trucks. To address these challenges, this paper proposes a model predictive control strategy that integrates mass and slope effects (MS-MPC) based on a comprehensive analysis of how speed, mass, and slope affect comprehensive operating costs. Firstly, mass variation factors and transient speed are employed as key indicators to develop a forgetting-factor recursive least squares (FFRLS) method, combined with extended Kalman filtering, to achieve effective decoupling and estimation of mass and slope. To enhance estimation accuracy, an adaptive mechanism is introduced to dynamically update the forgetting factor in FFRLS and reallocate the covariance matrix. Subsequently, using the estimated results and vehicle speed information, a pattern recognition method is employed to adapt operating conditions in the radial basis function neural network prediction model. Finally, dynamic programming is applied to optimize energy distribution based on the predicted information. Simulation results demonstrate that the proposed strategy significantly improves mass and slope estimation accuracy, reduces speed and slope prediction errors, and effectively lowers the comprehensive operating costs of the vehicle.