This paper presents an experimental study dealing with a non‐intrusive signal‐based method for a fault diagnosis of a proton exchange membrane fuel cell (PEMFC). The aim of this tool is to define a specific signature of different operating conditions. First, a healthy state has to be defined: an operating condition with fixed parameters values for current, temperature, pressure is used as a reference state. When the operating conditions divert from this state of functioning, a fault is considered. Thanks to the wavelet transform of different signals from the fuel cell system, a signature can be extracted and linked to the considered operating conditions. The energy contents of the wavelet transform is used as an indicator of the state‐of‐health (SoH) of the fuel cell system with the purpose to detect deviations from healthy state operating conditions and to detect the fault related to. The diagnosis tool presented in this paper is based on existing signals acquired on the fuel cell system, in order to minimize the number of actual sensors. The other goals are to minimize the number of extracted parameters from the fuel cell system and to be as less intrusive as possible in the fuel cell system.