Research Of Flooding Process And Design Of Fault Diagnosis System For Pem Fuel Cells

The fault diagnosis system is the key issue for stable operation of protonexchange membrane fuel cells. In order to develop the fault diagnosis system forPEM fuel cells, amount of researches of basic theories, software and hardwareare carried out in this study.The relevant researches of basic theories, mechanisms, affecting factorsand diagnosis methods on water management were investigated and evaluated.The life definitions and estimation methods of fuel cells and the functional goalsof fuel cell fault diagnosis system were summarized and analysed.The research on flooding prediction technology of fuel cells: Flooding canbe affirmed by comparing the actual hydrogen pressure to the ideal value, whilethe exact determine boundary is not known. Flooding experiments and analysisof fuel cells were carried out, and the step-level characteristic of hydrogenpressure drop during flooding process was discovered. The numerical value ofthe standard level coincided perfectly with the theoretical value, and the voltagewould drop rapidly after the step level without any treatment measures. So theflooding process of fuel cells was divided into four periods, which wereanhydrous, wet, transition and flooding. The growth rate of the step level wasdiscussed, the effects of current and temperature were little while that ofpressure and hydrogen stoichiometric were obvious. It could alleviate floodingtrend to adjust stack temperature or air humidification temperature properly inwet period, while even the purge failed to make fuel cells out of flooding periodcompletely. In this study, the growth rate of the step level was about20%,which provided the control range for closed-loop control of water management.The application development of the fault diagnosis system: The safetywarning and fault diagnosis based on neural networks of sensor signals weredeveloped, the fails of components were monitored and the reliability of systemwas achieved. The remaining life estimation method was designed in thesoftware, on one hand, the operation condition frequency could be recorded andthe decline rules of various conditions could be gained through the long-termprocess; on the other hand, the dynamic environment factor could be updated through the recent performance declines and operation conditions. After all, theremaining life of fuel cells could be calculated by equationsThe whole embedded fault diagnosis system was developed which includedsignal monitoring and control functions, multichannel voltage monitor module,and the remaining life estimation. The functions were verified by experimentsand data analysis on a self-developed test playform. The prediction and recoverof poor water status, and the safety monitoring and fault diagnosis of sensorsignal were realized.The system could easily be introduced to various PEM fuelcell applications.