| Solid Oxide Fuel Cell(SOFC)is a new technology with development potential,which has many advantages such as high power generation efficiency,no pollution,wide range of fuel sources,etc.,and has been favored by many researchers around the world.SOFC works in a high temperature environment,the value of waste heat utilization is higher,if SOFC cogeneration or with other power generation system cogeneration will greatly improve the energy efficiency,so the simulation analysis of SOFC under a variety of conditions is particularly important.Therefore,this paper takes the stack composed of 30 single batteries as the research object,collects the operating data of SOFC stack under a variety of working conditions,and conducts dynamic modeling and control strategy research.The research contents of this paper include:(1)The SOFC battery stack studied uses pure hydrogen as fuel,works in a high temperature environment of 650-750℃,and the range of load current is 0-30A.The experimental data of SOFC cell stack under various working conditions were collected and analyzed to understand the performance variation trend of SOFC cell stack under different working conditions.(2)according to electrochemical conservation,conservation of mass and conservation of thermodynamics structures,SOFC cell stack mechanism model,the simulation under different temperatures,fuel efficiency SOFC I-V curve,at the same time to carry on the dynamic simulation of the model and analysis of steady and transient characteristics.The results show that this model can better response operation characteristics of SOFC cell stack.(3)Aiming at the complex process of mechanism model establishment and the problem that it is not applicable to the control system,this paper uses BP neural network,support vector machine and random forest three algorithms to establish SOFC battery stack model.Three algorithm models are used to fit and predict the experimental data.The results show that the three algorithms can fit well,but the prediction performance of random forest is poor.The performance of BP neural network is better than that of SVM and SVM is better than that of random forest.(4)Based on MATLAB/Simulink software,BP neural network algorithm model as the research object to establish SOFC battery stack control system.Because external load changes will cause SOFC output voltage changes,it is necessary to design a controller to maintain the output voltage stability.In the control system established in this paper,the load current is the interference input,the hydrogen flow is the control input,and the control purpose is to maintain the stability of the output voltage.The controller adjusts the hydrogen flow to cope with the output voltage changes caused by the load current disturbance.In this paper,two control methods,PID and fuzzy PI,are used to simulate the battery stack system.The simulation results show that both of the two control methods can stabilize the output voltage at the set value.Under the PID control method,the variation range of output voltage caused by load current disturbance is smaller than that of the fuzzy PI control method.Under the fuzzy PI control method,the response time of output voltage is better than that of the PID control method.In general,the performance of fuzzy PI controller is better than that of PID controller. |
