The combustion resonance is a focal point of the analysis of combustion and thermodynamic processes in diesel engines, such as detecting ‘knock’ and predicting combustion noise. Combustion resonant frequency is also significant for the estimation of in-cylinder bulk gas temperature and trapped mass. Normally, the resonant frequency information is contained in in-cylinder pressure signals. Therefore, the in-cylinder pressure signal processing is used for resonant frequency calculation. Conventional spectral analyses, such as FFT (Fast Fourier transform), are unsuitable for processing in-cylinder pressure signals because of its non-stationary characteristic. Other approaches to deal with non-stationary signals are Short-Time Fourier Transform (STFT) and Continue Wavelet Transform (CWT). However, the choice of size and shape of window for STFT and the selection of wavelet basis for CWT are totally empirical, which is the limit for precisely calculating the resonant frequency. In this study, an approach based on Empirical Wavelet Transform (EWT) and Hilbert Transform (HT) is proposed to process in-cylinder pressure signals and extract resonant frequencies. In order to decompose in-cylinder pressure spectrum precisely, the EWT are applied for separating the frequency band corresponding combustion resonance mode from other irrelevant modes adaptively. The signals containing combustion resonant mode is processed by HT, so that the instantaneous resonant frequency and amplitude can be extracted. Validation is performed by four in-cylinder pressure signals with different injection timing. And the effects of injection timing on resonant frequency are discussed.
Leakage Detection in Pipeline using Wavelet Transform Method
