Blind Source Separation and Identification of Engine Radiation Noise Based on Independent Component Analysis and Wavelet Transform
Kang Shao · Transactions of Csice · 2012
Independent component analysis(ICA)and wavelet transform technology were used to study the blind source separation and noise source identification of engine radiation noise.Based on independent component analysis theory,the FastICA algorithm to get maximum negative entropy was adopted to separate noise signals of the four cylinder diesel engine,and a series of independent components were obtained.Fast Fourier transform and wavelet transform were used to identify the corresponding relationship between different engine noise sources and independent components.Combining the time-frequency analysis results and engine noise sources frequency spectrums,the corresponding relationship was determined.Results show that these independent components correspond to diesel engine combustion noise,piston knocking noise,timing gear noise and exhaust noise respectively.