Noise Prediction and Its Application to Engine Detect Events and Faults Diagnosis

Engine is the most common and important component in all the vehicles. An improved understanding of noise signal is required for the early detection of incipient engine component failure to achieve high reliability. In diesel engine, the combustion noise is so named to distinguish it from the noise of mechanical sources in the engine such as the valve train, fuel injection system and timing gears. On the other hand, piston slap is a combination of combustion and mechanical sources within the engine, and is not straightforward one. However, the aim of the present work is to separate the noise contributions of combustion and mechanical (motored) sources. Moreover, the noise spectra obtained when the engine was in fired condition were used to detect engine events and faults. The noise spectra were predicted based on the vibration velocity measurements, which were taken over the diesel engine surfaces (panels). An artificial fault event was made on the engine valve tapping, liner diametral, main bearing diametral and big end bearing diametral clearances. The results indicate that the prediction procedure presented in this paper has been shown to give good definition of small scale faults and errors; the nature of this procedure, which can summarize engine condition with small data files, lends itself well to record, keeping on a large number of engines.