Health-Aware Bi-Level Optimization of Component Sizing and Real-Time Energy Management in Fuel Cell Hybrid Electric Trucks
Jinyuan Zhang, Qian Xun, Marco Liserre, Hengzhao Yang
Abstract
Component sizing and energy management are two critical issues that affect the fuel economy of fuel cell hybrid electric trucks. Solving these two strongly coupled problems separately may lead to compromised performance at the system level. To address this concern, this paper proposes a bi-level optimization framework in which the upper level optimizes the fuel cell and lithium-ion battery sizes while the lower level optimizes the power allocation between them. Based on the sizing variables at the upper level, the lower level utilizes convex programming and rolling optimization to generate an optimal energy management solution and returns the solution to the upper level. Then, the upper level runs a particle swarm optimization process to determine the optimal sizes of the fuel cell and battery to minimize the system total cost including the component costs and operating costs. This framework is health-aware in that it considers the fuel cell and battery degradation by incorporating the corresponding degradation costs in the objective function. A case study using a modified CYC_WVUCITY drive cycle is conducted. Results demonstrate that the proposed framework can optimize the component sizes and the power allocation between them while reducing the component degradation. Moreover, this framework can complete the computation within the sampling interval and therefore can be implemented in real-time applications.