Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Real-Time Optimal Energy Management of Hybrid Energy Systems for Fuel Cell Electric Vehicles
Each fuel cell electric vehicle (FCEV) relies on an energy management strategy (EMS) to allocate its demand power within its hybrid energy system efficiently and safely. Based on a literature review, the research gaps in the existing EMSs for FCEVs were identified, which motivated this research to develop three novel real-time optimal EMSs to improve the fuel economy and mitigate the FC degradation of the FCEVs using the state-of-the-art EMSs. Firstly, this dissertation developed an adaptive model predictive control (AMPC)-based real-time optimal EMS. The AMPC-based EMS employed a linear, parameter-varying (LPV) prediction model developed for the FC-battery energy system of the FCEV. Compared with the prediction models used in the existing EMSs, the LPV prediction model provided higher accuracy by considering the variation of the system parameters while offering better real-time implementation capability. Secondly, this dissertation developed a deep reinforcement learning (RL) approach to solve the optimal energy management of the FCEVs powered by an FC-battery-supercapacitor energy system. In the approach, the EMS was represented by a deep neural network. A deep deterministic policy gradient (DDPG) algorithm and a twin delayed DDPG (TD3) algorithm were designed to train the deep neural network, respectively, to obtain an optimal EMS. Compared with the existing RL-based EMSs, the DDPG-based and TD3-based EMSs overcame their disadvantage of using tabular Q-learning, which suffers from the curse of dimensionality issue, and their limitation of learning the optimal EMS only in the discrete action. Finally, this dissertation developed two hardware-in-the-loop (HIL) test setups for the FCEVs studied. The first setup was developed to validate the AMPC-based EMS for the FCEVs powered by an FC-battery energy system. The second setup was developed to validate the DDPG-based and TD3-based EMSs for the FCEVs powered by an FC-battery-supercapacitor energy system. HIL testing-based comparative studies verified that compared with the state-of-the-art EMSs, the proposed EMSs improved the fuel economy and mitigate the FC degradation of the FCEVs.
Jia, Chao, "Real-Time Optimal Energy Management of Hybrid Energy Systems for Fuel Cell Electric Vehicles" (2022). ETD collection for University of Nebraska - Lincoln. AAI29165822.