A Research Of Fault Tolerant Control Strategy Applied To PEMFC

Proton exchange membrane fuel cell (PEMFC) is a device to convert chemical energy of hydrogen and oxygen into electrical energy. Because of its high energy conversion efficiency, environmentally friendly, no noise, light weight, and low operating temperature, study of the PEMFC is very meaningful in the situation of worsening energy crisis and environmental pollution. During operation of the PEMFC, dehydration of proton exchange membrane or excess water of electrode is unavoidable. Once these faults occurred, the performance of the PEMFC systems would lead to different degrees of decline. PEMFC system carrying out fault tolerant control scheme is in favor of remaining stable output performance when it is under fault conditions. Therefore, fault diagnosis method and fault tolerant control technology for PEMFC appear in this paper. The main contents include:(1) Establish PEMFC’s dynamic mathematical model. According to principles of electrochemical reaction, the reaction polarization, electro-osmosis process water and poor concentration of both sides of the proton exchange membrane, and using MATLA-B/SIMULINK tools, PEMFC’s dynamic mathematical model on conditions of healthy, flooding and membrene drying are established. The simulation results show that the mathematical models can effectively reflect change characteristics of PEMFC’s output voltage and pressure drop between the anode and cathode in the conditions of normal, flooding and membrene drying.(2) Propose fault tolerant control scheme for PEMFC system based on feedback linearization control algorithm. The fault tolerant control scheme includes:fault diagnosis module, the switching mechanism and fault-tolerant controllers module. In fault diagnosis module, firstly, the establishment of identification in PEMFC’s heathy condition is based on BP neural network, which identifies PEMFC’s output voltage and the cathode’s pressure drop beween inlet and outlet in healthy state. The real-time output voltage and cathode’s pressure drop beween inlet and outlet are compared, PEMFC’s faults of membrane drying and flooding are diagosed online. According to the type of faults detected, the switching mechanism switchs to the corresponding controller. In fault-tolerant controllers, we use the feedback linearization algorithm to design PEMFC’s controllers’ of conditions in health, flooding and membrane drying. Simulation results show that, under different fault modes, fault tolerant control scheme can always make PEMFC’s voltage following the set value, while maintaining the pressure drop between anode and cathode equal by adjusting inlet gas pressure of the anode and cathode.(3) Propose fault tolerant control scheme for PEMFC system based on neural network predictive control algorithm. The neural network model predictive control algorithm includes a prediction sets and a rolling optimizer. In prediction sets, a recursive step predictive model based on BP neural network is proposed. The past and current voltage and pressure drop between the anode and cathode are used to build single-step prediction model, then according to the recursive principle, output of the previous step prediction model is also input of the next step prediction model. A plurality of single-step prediction model consist of recursive step predictive model. In rolling optimizer, the performance optimization is sum of error between the output value of the multi-step predictive model with the corresponding voltage and pressure drop between anode and cathode. A three-layer feed-forward neural network controller is designed, the error of the one-step prediction model’s output value and the reference value are used to obtain a control law. The fault tolerant control scheme can avoid the establish of complex PEMFC’s mechanism model, and the PEMFC’s output voltage is tracked effectively, while the pressure drop between anode and cathode maintaining steady.