Proton Exchange Membrane Fuel Cell is a complex multiphysical system exhibiting complex and coupled phenomenon from several domains of energies (e.g. electrochemical, electrical, thermal, and hydraulic). However, this promising energy converter is vulnerable to faults (such as drying of the membrane, default of the cooling system and flooding) that can cause permanent damage. Moreover, the performance of the fuel cell is continuously affected by its aging. Therefore, it is essential to implement techniques to detect and isolates those faults. This paper proposes the use of Bond Graph as an integrated tool for the modeling and the generation of fault indicator for this multi-physical system. A diagnostic procedure is also detailed based on the complex structural and dynamical properties in order to isolate a set of single fault.