The current generation of Polymer Electrolyte Membrane (PEM) fuel cells suffers from decreased durability characteristics compared to conventional internal combustion engines. If PEM fuel cells in automotive applications are to be competitive, it is important to minimise losses in performance (due to degradation of components) over extended periods of time. One way to achieve this is to develop accurate Fault Detection and Isolation (FDI) techniques to ensure effective fault tolerant control and maintenance strategies, which in turn will ensure stable and prolonged operation of the system. This ongoing research project is focused on model-based FDI. The main challenge with this approach is that fuel cell models are complex and the internal structure of a fuel cell is poorly instrumented making it impossible to measure some of the key parameters. Hence a qualitative diagnosis using Bond Graphs (BG) has been chosen here in order to investigate some major faults occurring in fuel cell systems. BG is a graphical modelling approach which represents interactions between system components as exchanges of energy. Not only does this enable the simulation and study of the dynamic performance of a system, it also enables FDI analysis by deriving Analytical Redundancy Relations (ARR) and calculating residuals. In this work a bond graph model of a PEM fuel cell is developed using Modelica modelling language and validated against a custom built fuel cell test system. A set of faults and degradation mechanisms within PEM fuel cells will be detected using BG’s. Future stages of research will use quantitative analysis of the current state of health and degradation rates of the system to extrapolate this information for prognostic studies.
展开▼
