A theoretical model of a humidifier of proton exchange membrane （PEM） fuel cell systems is developed and analyzed first in this paper. The model shows that there exists a strong nonlinearity in the system. Then, the...A theoretical model of a humidifier of proton exchange membrane （PEM） fuel cell systems is developed and analyzed first in this paper. The model shows that there exists a strong nonlinearity in the system. Then, the system is identified using a wavelet networks method. To avoid the curse-of-dimensionality problem, a class of wavelet networks proposed by Billings is employed. The experimental data acquired from the test bench are used for identification. The one-step-ahead predictions and the five-step-ahead predictions are compared with the real measurements, respectively. It shows that the identified model can effectively describe the real system.展开更多
High-frequency resistance(HFR)is a critical quantity strongly related to a fuel cell system’s performance.It is beneficial to estimate the fuel cell system’s HFR from the measurable operating conditions without reso...High-frequency resistance(HFR)is a critical quantity strongly related to a fuel cell system’s performance.It is beneficial to estimate the fuel cell system’s HFR from the measurable operating conditions without resorting to costly HFR measurement devices.In this study,we propose a data-driven approach for a real-time prediction of HFR.Specifically,we use a long short-term memory(LSTM)based machine learning model that takes into account both the current and past states of the fuel cell,as characterized through a set of sensors.These sensor signals form the input to the LSTM.The data is experimentally collected from a vehicle lab that operates a 100 kW automotive fuel cell stack running on an automotive-scale test station.Our current results indicate that our prediction model achieves high accuracy HFR predictions and outperforms other frequently used regression models.We also study the effect of the extracted features generated by our LSTM model.Our study finds that only very few dimensions of the extracted feature are influential in HFR prediction.The study highlights the potential to monitor HFR condition accurately and timely on a car.展开更多
基金supported by the Young Scholars Developing Fund of Tangshan Teacher’s College (No.06D06)the Doctoral Fund of the Schoolthe Outstanding Overseas Chinese Scholars Fund of Chinese Academy of Sciences (No.2003-1-10)
文摘A theoretical model of a humidifier of proton exchange membrane （PEM） fuel cell systems is developed and analyzed first in this paper. The model shows that there exists a strong nonlinearity in the system. Then, the system is identified using a wavelet networks method. To avoid the curse-of-dimensionality problem, a class of wavelet networks proposed by Billings is employed. The experimental data acquired from the test bench are used for identification. The one-step-ahead predictions and the five-step-ahead predictions are compared with the real measurements, respectively. It shows that the identified model can effectively describe the real system.
文摘High-frequency resistance(HFR)is a critical quantity strongly related to a fuel cell system’s performance.It is beneficial to estimate the fuel cell system’s HFR from the measurable operating conditions without resorting to costly HFR measurement devices.In this study,we propose a data-driven approach for a real-time prediction of HFR.Specifically,we use a long short-term memory(LSTM)based machine learning model that takes into account both the current and past states of the fuel cell,as characterized through a set of sensors.These sensor signals form the input to the LSTM.The data is experimentally collected from a vehicle lab that operates a 100 kW automotive fuel cell stack running on an automotive-scale test station.Our current results indicate that our prediction model achieves high accuracy HFR predictions and outperforms other frequently used regression models.We also study the effect of the extracted features generated by our LSTM model.Our study finds that only very few dimensions of the extracted feature are influential in HFR prediction.The study highlights the potential to monitor HFR condition accurately and timely on a car.
