Algorithm of lifting wavelet package in real-time fault diagnosis system

Wavelet package transform splits finely the high frequency band gradually. So it has more powerful frequency resolution than wavelet transform and is better for machinery fault diagnosis such as hydro-generator with higher frequency vibration signal. But its algorithm is too complex to be used in real-time diagnosis system which an greatly increase the reliability of machinery facility running. In order to simplify the algorithm of wavelet package and reduce the need of memory in real-time system during signal decomposition, one novel lifting scheme of wavelet package is put forward which can quickly decompose signal and extract fault information while collecting signal. With optimization of the program structure of the algorithm on DSP the scheme can realize on-line monitoring and real-time diagnosing of machinery fault. Experiment shows that this algorithm of lifting wavelet package runs well in a real-time fault diagnosis system of hydro-generator.