Research On Modeling And Control Of Fuel Cell System

Fuel cell is widely regarded as one of the most promising energy sources because of its high efficiency, low noise and zero emission. The research on modeling and control of fuel cell system is important to improve its efficiency, dynamic response performance and life. Currently, the research in this field is still in the initial state, many problems need to be solved.The fuel cell system includes four parts:the gas supply subsystem, the thermal management subsystem, the water management subsystem and the energy management subsystem. Because the dynamics response times of these subsystems are different, the thermal management subsystem and the water management subsystem are not considered. We put importance on air supply system and DC/DC converter system. In addition, the stack model used to calculate the output voltage is essential in control, so this part is also studied. The main works of this thesis are as follows:1. A semi-empirical steady-state voltage model is developed based on mechanism analysis of the proton exchange membrane fuel cell. The obtained model contains not only the process of electrochemical reaction but also takes into account the gas diffusion in gas diffusion layer (GDL). Through reasonable simplification, the analytical expression of gas diffusion in GDL is derived which is convenient for calculation. And a linear parameter equation is obtained through simplify and transformation. The model parameters don’t include operation conditions such as temperature, pressure and so on, so the model has a strong versatility. In addition, these parameters can be identified using the least square algorithm based on experiment data. The experiment results show the accuracy and generalization of the proposed model. For air supply subsystem, a fourth order dynamic model is put forward based on a third order model existing in earlier literature. The model comparisons prove the validity of the fourth order dynamical model.2. A model predictive control algorithm based on linear parameters varying model(LPV-MPC) is developed for air supply system considering its nonlinearity and constraints. To cope with the state unmeasured problem, the Kalman filter is added to the LPV-MPC algorithm. In the meantime, the measurement variables are determined through the observability analysis. The simulation comparisons with linear MPC and NMPC based on mechanism model demonstrate the validity and superiority of the proposed algorithm when the fuel cell’s load changes in a large range.3. An average model of Boost DC/DC converter is given for energy management subsystem. Then a PID controller is designed whose parameters are tuned though linear quadratic regulation(LQR) algorithm which not only avoid the complex process of parameters adjustment but also ensure the control optimality. At the same time, the control output with no steady-state error is achieved by adding the integral state of the output error. For the system combined the air supply component and DC/DC converter component, a decentralized control structure is put forward. As can be seen from the simulation results, the power-autonomous fuel cell system has been controlled well.