Diagnosis of MEA degradation for health management of polymer electrolyte fuel cells
2020-02-18T14:34:06Z (GMT) by
Diagnostics and health management are fundamental components in a strategy to improve durability and lifetime of polymer electrolyte fuel cells. Fuel cells require a range of operating conditions to be well managed for achieving performance or durability objectives. So far, water management issues and single parameter diagnostics for individual degradation modes have been the focus of research in the literature. However, there has been minimal research on the application of fuzzy inference systems for online, multiple parameter diagnosis of fuel cells. This research presents an advanced fuzzy inference system for diagnostics and health management of a membrane electrode assembly (MEA) for polymer electrolyte fuel cells. The fuzzy inference system facilitates simplified connections of the complex relationships between numerous operating conditions and subsequent degradation modes. The approach utilises the most important operating parameters for diagnosis of high priority degradation modes using multiple health sensors. The developed fuzzy inference system classifies the fuel cell input data into simple linguistic categories for example ‘cell voltage is very high’ or ‘stack temperature is low’ through a fuzzification process. Based on a set of antecedent-consequent (if-then) rules, an inference calculation is performed without necessity for complex mathematical models. This enables a fast diagnosis with fuel cell parameters classified on a scale of inclusion to the linguistic categories. The linguistic classification of a degradation mode is converted back into a numerical value through a defuzzification process. The output data can be used to inform the user on the fuel cell state of health. The investigation has focused on the diagnosis of MEA degradation as it has been identified as having critical impact on fuel cell performance and lifetime. A single cell with a 25cm2 active area was used for testing under numerous moderate to extreme operating conditions known to cause membrane and electro-catalyst degradation. A database of if-then rules was initially developed based on knowledge in the literature and refined with experimental testing. Results so far have supported validation of the fuzzy inference system membership functions and the rule base for diagnosing the consequential degradation modes based on fuel cell operating conditions. This diagnostic and health management approach facilitates proactive decision making for mitigation strategies to be employed according to performance or lifetime targets and can increase fuel cell availability and lifetime therefore improving the overall value of the system.