Fault detection in engines through higher order spectral analysis of acoustic signatures

This paper describes new engine fault detection methods that involve a high er order spectral analysis method known as the parametric spectral coherenc e (PSC). Measurements of radiated acoustic and vibrational signatures were conducted on a one-cylinder engine with a continuously worsening loss of cy linder compression. Compared to vibrational signatures, the radiated acoust ic signatures provided the most dramatic indication of a loss of cylinder c ompression when PSC analysis was performed. It is demonstrated that the fau lt can be detected by simply monitoring the average spectral coherence or t he peak spectral coherence of the acoustic signature, whereas the more trad itional power spectral methods fail to detect the fault. (C) 1999 Acoustica l Society of America.