Application of Discrete Wavelet Transform (DWT) in Combustion Failure Detection of IC Engines

Fault detection of internal combustion engines has been started to develop more rapidly in recent years. Few works have been published on time-frequency analysis of IC engine vibration signals. These types of analysis are fully adapted to transient nature of IC engine signals and do not possess the drawbacks of traditional methods based on Fourier analysis. In this paper we used the vibration signals taken experimentally from a typical 4-cylinder IC engine. Ignition failures were introduced and their effects were investigated. Since the signals were noisy the time synchronous averaging (TSA) was used to extract the dominant transient behavior of the signals. After TSA, discrete wavelet transform (DWT) was employed to transfer the signal in time domain to the time-frequency domain. DWT holds significant properties which make it suitable for fault detection and diagnosis. The results showed that DWT decomposition in 4^th approximation level is appropriate for fault detection. For this level and the higher levels, the trends of the signal variations were studied in different operating conditions of the engine, namely the healthy and faulty ones. The results are promising and reveal the effectiveness of the method. Comparison between our previous works and this new approach shows a simpler interpretation of DWT results over other time-frequency methods such as STFT and CWT. These facts demonstrate the possibility of using DWT for monitoring combustion failure of IC engines in addition to traditional applications such as diagnosis of rotating machineries.