Research on the Fault Warning Method Based on Dual-tree Complex Wavelet Transform and BP Neural Network

A fault warning method based on dual-tree complex wavelet transform and BP neural network was proposed for the efficient and timely detection of diesel engine operation and the accurate fault warning. Firstly, the suspected fault section was found by comparing the mutation of the frequency effective value in the continuous time period which was based on the interval division of the power spectral density of the input signal. Secondly, the dual-tree complex wavelet transform was used to analyze the suspected fault segment, and the transform scale was confirmed. Then, fault feature vector was extracted from the frequency range according to the suspected fault segment. At last, the fault feature was trained by BP neural network., and then it realized the diesel engine fault early warning. The experimental results showed that this method can accurately identify the running state of diesel engine and carry out fault warning, so it has certain application value.