Real-time Predictive Energy Management for Fuel Cell Electric Vehicles

This paper presents the design and application of a model predictive control-based energy management for a fuel cell hybrid electric vehicle. To estimate the upcoming vehicle speed within each receding horizon, a speed-forecast method is proposed using the layer recurrent neural network (LRNN). Then, the power-allocating decisions are derived via minimizing the multicriteria cost function by considering the predicted speed sequence. It has been verified that the LRNN predictor has a higher accuracy versus the benchmark methods. Software-in-the-Loop testing results have indicated that the proposed control strategy can improve fuel economy and fuel cell durability versus a rule-based benchmark, with an acceptable online computational burden.