Generation of Statistical Descriptors for Running PEM Fuel Cell Stack by Means of Long-Time Electrochemical Noise Measurements

Electrochemical noise (EN) measurements have been provided with PEM fuel cell stack of 600 W nominal power. These measurements demonstrate that EN can be used as a tool for diagnostics of possible faults in long-term running of electrochemical sources of energy. The noise measurements were performed in well controlled operational conditions during 100 h. Robust and stable statistical noise descriptors of the stack have been obtained both in frequency and time domains. The spectral noise signature of the stack based has been obtained in the frequency range 0.1 < f < 10(3) Hz and involves two ``pink noise'' parts with different slopes and a pronounced pick at the characteristic frequency f = 1.6 Hz. The noise signature in time domain is based on calculations of different order moments (intensity, skewness, kurtosis) for original and detrended signals. It was demonstrated the sensibility of the statistical noise descriptors to relaxation phenomena arising in stack after interruption of its running.