Research On SOFC Parameter Identification Technology Based On Intelligent Optimization Algorith

With the rapid development of the global economy,the environmental pollution caused by massive consumption of fossil energy is becoming increasingly serious.In recent years,various new energy and power generation technologies have attracted the attention of governments and researchers all over the world.Among them,hydrogen energy,as a green and clean secondary energy,has been positioned as a national strategic energy source,and fuel cells are a key way to achieve efficient utilization of hydrogen energy in the downstream industry of hydrogen energy.Solid oxide fuel cell(SOFC)has become one of the most promising green power generation technologies due to its high conversion efficiency,reliable operation,no pollutant emission,modularization and low noise.It is widely used in power systems and transportation.However,higher operating temperatures will limit the service life of SOFCs,so it is necessary to optimize their operating points,monitor their status,and diagnose their faults,which has promoted the rapid development of research related to fuel cell system modeling,parameter identification,etc.In addition,due to the characteristics of high nonlinearity,multivariability,and strong coupling among unknown parameters,it is difficult for traditional methods to effectively solve the parameter identification problem of fuel cell models.Therefore,aiming at the problem of parameter identification of two kinds of gray box models of SOFC(namely electrochemical model(ECM)and simplified electrochemical model(SECM)),this paper proposes two parameter identification frameworks based on improved heuristic algorithm and numerical optimization technology,respectively,and fully validates the proposed two parameter identification frameworks using the experimental data of three kinds of single cells(i.e.,a planar single cell developed by SOFC Research & Development Center of Huazhong University of Science and Technology,a planar single cell produced by Elcogen Fuel Cell Technologies,Finland,and a single cell from CEREL,Poland)and three kinds of stack cells(i.e.,a 1 k W planar SOFC stack developed by Huazhong University of Science and Technology,a 5 k W tubular SOFC stack of Montana State University,and a 100 k W SOFC system of Siemens).Finally,the design of SOFC parameter identification visualization platform and the development of related functions are realized.The main research work,contributions and achievements of this paper can be summarized as follows:(1)To improve the accuracy and stability of SOFC parameter identification,the AEOMRFO coordinating optimizer(EMCO)is proposed through in-depth analysis and integration for the optimization mechanism of artificial ecosystem-based optimization(AEO)algorithm and manta ray foraging optimization(MRFO)algorithm.EMCO is easy to implement because it does not introduce additional complex mechanisms.Meanwhile,particle swarm optimization(PSO)algorithm,equilibrium optimizer(EO),dandelion optimizer(DO),heap-based optimizer(HBO),white shark optimizer(WSO),AEO algorithm,MRFO algorithm,and EMCO were used to conduct a comprehensive parameter identification study on the above six experimental data,it is verified that EMCO has excellent performance for SOFC parameter identification task.(2)Although the proposed EMCO has significantly improved parameter identification performance compared to the other seven traditional heuristic algorithms,the inherent randomness of the heuristic algorithm can also make its parameter identification stability weak in some scenarios.Hence,this paper continues to develop a parameter identification framework based on Levenberg-Marquardt back propagation(BP)algorithm.Firstly,ECM and SECM are designed as fully equivalent special artificial neural network(ANN),and then the special ANN is trained using experimental data and LMBP algorithm based on gradient information.After the training,the unknown parameters of SOFC model can be extracted from the specific parameters of the network through simple algebraic operation,which can better meet the accuracy and stability requirements of parameter identification.(3)To realize the standardization of SOFC parameter identification process and the visualization of identification results,and improve the information processing speed and data management ability of the operation and maintenance personnel and relevant research and development personnel,a SOFC parameter identification software is developed based on MATLAB language,which integrates the above nine parameter identification technologies for users to choose,and can easily extract the unknown parameters of EMC and SECM.The research results show that WSO and PSO are the fastest and slowest among the eight heuristic algorithms,while LMBP algorithm is significantly slower than the eight heuristic algorithms in average running speed due to the need to perform the inverse operation of the matrix in the process of parameter optimization.In general,the two SOFC parameter identification techniques proposed in this paper have relatively optimal stability and accuracy.Moreover,whether accurate and reasonable model parameters can be identified depends not only on the performance of the algorithm but also on the completeness of the data used.At the same time,if the current exchange density of the cathode and anode of the fuel cell is similar,the slope of the Tafel curve between them will show a double quantitative relationship under the same data set due to the characteristics of the representation of the ECM and SECM.