Modelling and flow rate control methods for anode tail gas circulation intake system at SOFC

In the solid oxide fuel cell (SOFC) power generation system, the anode tail gas circulation intake system plays an important role in the entire system for efficient and stable operation. Currently, there are relatively few studies on the operating characteristics of claw pump, which works as the core component of the SOFC anode exhaust gas circulation intake system. Based on the prototype of 20 kW SOFC power generation system, this paper builds the emulation testbed for the anode exhaust gas circulation intake system and analyzes the working characteristics of the claw pump in the testbed firstly. The support vector regression (SVR) and Gauss process regression (GPR) methods are then used to establish the characteristics model of the claw pump based on the operation data. According to the verification results, GPR method owns higher prediction accuracy than SVR method in our case. Therefore, the GPR method is adopted to build the flow rate prediction model of the claw pump. Finally, the flow rate control method based on feedforward and model prediction (FMP) with GPR flow rate prediction model is proposed for the built anode tail gas circulation intake system. In experiments, the response time of the FMP method is within 22 s and the overshoot is less than 3.2%, whose performance is better than the traditional flow rate control method, such as PID or model predictive control.