Modeling And Performance Research Of Proton Exchange Membrane Fuel Cell System

Proton exchange membrane fuel cell(PEMFC)has the advantages of high power density and zero pollution,but its commercialization still needs to solve the application bottleneck of life and cost.At present,there are many research and development difficulties in fuel cell industry,such as the lack of efficient test data analysis methods,the lack of advanced mathematical model modeling and verification methods,and the lack of maturity of optimization technology.This leads to the fact that the experience design method used in the whole vehicle factory cannot evaluate the performance of the stack in advance,and then the system integration and debugging rely heavily on the high-cost stack test bench.In order to solve this problem,it is necessary to establish a forward research and development system based on system engineering.As an important tool and technical means of system design,mathematical model and parameter calibration are essential to realize high precision control of fuel cell system.However,since the previous lumped parameter model is difficult to reflect the spatial distribution characteristics of physical parameters and cannot consider the generation and accumulation process of liquid water,it has been difficult to meet the requirements of the development of fuel cell control strategy.Therefore one-dimensional model is used as a tool,and evolutionary algorithm is used for parameter identification based on experimental test to explore the mechanism of hydrothermal transmission and establish a high-precision fuel cell model.The model structure under different scales is studied,and the multi-scale system-level model is constructed by combining the model and lumped parameter model.The fuel cell system analysis and control strategy optimization are realized,which provides certain reference value for the ’ digital twin ’ in this field.The main research work is as follows:(1)A control-oriented one-dimensional non-isothermal two-phase flow fuel cell model is proposed.Combining the advantages of lumped parameter model and distributed parameter model,considering the gas transient effect in the channel and the water phase transition in the porous region,the reactor mechanism model is established and its effectiveness is systematically verified.The results show that the established model has good real-time performance and structural feasibility,which can reflect the internal gas concentration and hydrothermal distribution characteristics,predict the influence of environmental conditions and model parameters on the output performance,and improve the limitation of underestimation of membrane water content in the lumped parameter model.(2)A relatively perfect model calibration verification process system is constructed.The element effect method is used to study the parameter sensitivity,and the influence of current density on the sensitivity results is explored.Then,the uncertain parameters are screened based on the graphical method according to the order of dominance.Finally,the parameter identification is carried out based on the adaptive genetic algorithm combined with the steady-state test data,and the transient performance of the model under non-calibration conditions is verified.The results show that the calibration verification process system has high parameter identification efficiency and model prediction accuracy.(3)Model structure optimization analysis.The steady-state and transient characteristics of fuel cell models under different modeling methods are studied by comparing lumped parameter model,aggregate model and machine learning model from the perspectives of catalytic layer components and stack system.The influence of model structure on output performance is discussed.The results show that the one-dimensional model has good applicability and stability,which is suitable for fuel cell system integration and control strategy design.(4)Fuel cell system control and performance analysis.Based on the high-precision fuel cell model and lumped parameter model,a multi-scale system-level model is established to evaluate the dynamic performance of components and reactors.Four feedback control algorithms from traditional control to intelligent control are compared to realize the rapid optimization design of air system control strategy.The results show that the system model has good integrity and can reflect the hydrothermal coupling relationship between variables,which provides strong technical support for further system optimization design and control strategy development.