Proton exchange membrane fuel cell(PEMFC)is regarded as one of the key energy conversion devices due to its high efficiency,low carbon,no emissions,and renewable fuels,especially under the background of global warming and carbon emission control.At present,the large-scale commercialization of PEMFCs still faces the problems such as low power density,high cost,and short lifetime.It is necessary to study the dynamic characteristics of PEMFCs because of their important impact on improving performance and reducing degradation.In this study,a multi-condition,multi-phase,and multi-physical PEMFC simulation system from the cell level to stack level has been constructed,including a quasi-two-dimensional dynamic model,a data-driven control optimization model,a dynamic stack model coupled with uneven fluid distribution,a Platinum particle degradation model based on the electrochemical reaction mechanism,a PEMFC stack performance degradation model based on a hybrid method of semi-empirical equations and machine learning.This paper comprehensively investigates the effort of electrochemical active surface area(ECSA)on cell performance under low humidity conditions,the optimization of cold-start process control conditions,the effect of uneven distributions on stack performance,the effort of heating assisted cold start strategies on stack performance from ultra-low temperatures,the influence of platinum particle degradation in catalyst layer(CL)and the prediction of long-term degradation of stack performance.The main research work of this paper is as follows:(1)A universal and reliable empirical equation of ECSA in the CL has been proposed.The ECSA is measured by cyclic voltammetry(CV)experiments and then an empirical equation of ECSA changing with humidity and temperature is determined,which makes up for the lack of considering the change of ECSA in the traditional PEMFC model.Its universality is verified by other literature data.Then a quasi-two-dimensional multi-physical transient PEMFC model with completely independent intellectual property rights is established,which fully considers the oxygen transport mechanism in cathode CL,the variation law of ECSA,the phase change mechanism above zero and below zero temperature,the crossover transport mechanism of reaction gas,and the local current density distribution mechanism.The model is verified by experimental data under various transient and steady-state conditions and it is found that considering the effect of ECSA on cell performance under low humidity conditions can improve the predictive accuracy of this model from 70% to 90%.The distribution law of local current density inside the cell is investigated.It is found that compared with the co-flow arrangement,the local current density of the PEMFC with the counter-flow arrangement is more uniform in the range of operating current,inlet pressure,inlet stoichiometric ratio,and working temperature.(2)A data-driven PEMFC fast prediction model and a real-time adaptive control strategy(RACS)have been proposed.A data-driven fast prediction model based on the neural network coupled semi-recurrent sliding window method is established.The effects of initial boundary conditions such as temperatures and operating currents(constant current and constant current loading rate)on the cold start processes are investigated.It is found that a smaller constant current loading rate will result in a longer cold start duration time.A higher constant current loading rate will lead to the rapid increase of impedance and reduction of cell voltage,resulting in the early termination of a startup.The safe critical range of constant current loading rate is determined,and the specific relationship between cold start initial conditions,operating current,and cold start time are summarized.Then RACS is proposed to optimize the operational conditions of the cold start process.This strategy uses real-time predicted cell performance as feedback to optimize the current control considering the startup time and cold start capacity.After optimizing the constant current loading rate strategy,the cold start time of the PEMFC can be reduced by about 26.7% when the starting temperature is-20 °C.The optimized operational current by RACS is verified by the numerical model.Finally,the ability of the fast prediction technology to predict the ice formation and melting process during the cold start process of PEMFC is also verified.(3)The characteristics and optimizations of heterogeneous performance in the PEMFC stack have been investigated.A sub-model of fluid distribution in the PEMFC stack is established,which considers the heterogeneous fluid distribution caused by friction pressure loss,local pressure loss,and reaction gas consumption in the cells.The reliability of the sub-model is verified by the pressure difference distribution of the cells.Then,the submodel is coupled with the quasi-two-dimensional sub-model of PEMFC to establish a system-level PEMFC stack model,which is verified by the cell voltage distribution.The effects of operating conditions and design parameters on the stack performances,such as inlet mass flow rate,cell voltage,temperature,and oxygen concentration,are studied.It is found that the fluid distribution in the U-type flow arrangement stack is more heterogeneous than that in the Z-type flow arrangement stack.The effects of inlet pressure,relative humidity,and inlet temperature on the fluid distribution can be almost ignored.However,large operating current,stoichiometric ratio,cell number,or small manifold diameter will significantly result in the non-uniformity of mass flow at the inlet of single cells in the stack.The optimization of design parameters to improve the performance difference in the U-type stack is further investigated.It is found that increasing the cross-sectional area of the intake manifold can improve the performance of single-cell far from the stack inlet and improve the non-uniformity of performance distribution in the stack.The reduction of the cross-sectional area of the single-cell channel can comprehensively improve the performance of the stack due to the increase of the airflow rate in the channels,which can strengthen the oxygen mass transportation in the porous zones and enhance the oxygen concentration in the cathode CL.Increasing the coolant flow rate can improve the uneven temperature distribution along the channels and reduce the inconsistency of stack performance distribution.(4)The effect of heating-assisted cold start strategies on the PEMFC stack has been investigated.Based on the developed PEMFC stack model,the self-cold-start ability and heating-assisted cold start strategy(resistance wire heating method and circulating coolant heating method)of the stack are explored.The effects of operating voltage,current density,heating power,number of single cells,thermal conductivity,and coolant flow rate on cold start performance and physical distributions of the stack are detailed analyzed.The distributions of voltage,temperature,super-cooled water,and ice saturation in the stack are expounded.It is found that the PEMFC can start successfully from-20 °C or-30 °C,but it will fail from-40 °C.Then the research on the assisted heating cold start of the stack from-40 °C shows that the resistance wire heating method is limited by the thermal conductivity,which can only improve the cold start performance of edge fuel cell,but can not effectively heat the internal temperature of the stack.The circulating coolant heating method can make the stack temperature rise more evenly,but the successful start-up of-40 °C temperature only by the output power of the stack is not enough for the demand of heating power,so it is necessary to supplement additional heat source power.(5)The degradations of Platinum particle and stack performance have been investigated.A one-dimensional Platinum particle degradation model is established based on electrochemical reaction mechanisms,which consider the difference in reaction rates of different particle sizes and the influence of boundary conditions under different working conditions.The effect of platinum loading on the catalyst performance is studied,and it is found that a smaller platinum loading would lead to a faster decay rate of the ECSA.To investigate the degradation of stack performance,two different models are respectively established based on semi-empirical method and data-driven method.It is found that the deviation of the semi-empirical model in predicting the stack voltage is concentrated between 2%,but the predicted voltage has no local dynamic characteristics,which can only reflect the overall degradation trend of stack performance.The deviation of short-term voltage decline predicted by the data-driven model is concentrated in 1%,and the predicted voltage also has accurate local variation characteristics.However,for the long-term prediction,the error will accumulate with the iterations and the deviation of the predicted voltage begins to fluctuate gradually,and the deviation distribution increases to 8%.Based on the above characteristics of the two models,a hybrid prediction model is further developed.The prediction results of the semi-empirical model are used to modify the input of the data-driven model,which can effectively improve the oscillation of prediction results of the data-driven model during the long-term degradation.It is found that the hybrid model has good error distribution(within 2%)and local performance dynamic characteristics and can be used to predict the process of long-term stack performance degradation. |