Real-Time Optimal Control of Power Management in a Fuel Cell Hybrid Electric Vehicle: A Comparative Analysis

Power split in Fuel Cell Hybrid Electric Vehicles (FCHEVs) has been controlled using different strategies ranging from rule-based to optimal control. Dynamic Programming (DP) and Model Predictive Control (MPC) are two common optimal control strategies used in optimization of the power split in FCHEVs with a trade-off between global optimality of the solution and online implementation of the controller. This is due to the fact that DP that offers the global optimal solution requires the pre-known knowledge of the driving condition for the whole drive cycle, which makes the real-time implementation of the strategy more challenging. In this article, both control strategies are developed and tested on a FC/battery vehicle model, and the results are compared in terms of total energy consumption. In addition, the effects of the MPC prediction horizon length on the controller performance are studied. Results show that by using the DP strategy, up to 12% less total energy consumption is achieved compared to MPC for a charge sustaining mode in the Urban Dynamometer Driving Schedule (UDDS) drive cycle. However, increasing the prediction horizon length makes the MPC performance converge to that of the DP controller and they perform with the same total energy consumption for the prediction horizon lengths of 7 seconds or longer. Finally, the results of the DP and MPC strategies are compared to a rule-based power split strategy which commands battery to provide the requested power during some transient vehicle operations. The comparison results between the DP and the transient mode rule-based strategies show that around 5% less total energy is consumed when the DP strategy is applied.