Abstract
This thesis aims to study three green power sources: proton exchange membrane fuel cells (PEMFCs), supercapacitors (SCs), and lithium-ion batteries (LIBs). The first research scope is to model their system dynamics for controller design and real-time simulators; the relationship among linear models and nonlinear models in the time domain and the frequency domain was analyzed. Next, a design procedure for optimal system integration and energy management of a hybrid electric power source was explored. It mainly focuses on the application to electric vehicles. For multi-disciplinary system modeling, a unified bond graph approach was employed. The highly-nonlinear model of PEMFCs using novel bond graph elements was firstly constructed. Then parameter identification and model linearization algorithms were solved. Comparison between the bond graph model and the electric circuit was completed to discuss the model accuracy and the physical interpretation. For the SCs, an equivalent linear model via AC impedance spectroscopy was built up. A bond graph derived from the equivalent model was further established to analyze the system dynamics in the time domain and in the frequency domain. To describe the nonlinear effects among model parameters, operation temperature, and the SC voltage, a three-layer artificial neural network was applied to form an online nonlinear SC model. For the LIBs, a new equivalent circuit according to the AC impedance approach was set up. The parameters were identified from the experimental data. Comparisons between the bond graph model and the equivalent circuit were then discussed. The second research target in this thesis was to develop a design procedure of optimal system integration and energy management for hybrid electric power sources. For hybrid system combination, two global optimization search methods: a time-independent and the other time dependent methods were proposed. By selecting evaluation indices, a cost function, and a multiple for-loop structure code, the optimal solution can be found. With similar procedures, the optimal energy management was evaluated in the form of multi-dimensional tables in the vehicle control unit to deal with the equipped dual power sources. Integrating the above two research issues, a procedure for optimal system designs/energy management was proposed. The absolute optimization can thus be achieved.