Working condition recognition of screw compressor using wavelets theory

A novel working condition recognition system of screw compressor using wavelets for engineering application is proposed. According to tested rotor and valve vibration signal energy changing characteristic under different working condition of the compressor, the fault position and fault reason can be obtained clearly. The paper firstly obtained rotor vibration signal power distribution on different frequency bands using wavelet packet transform and frequency bands divided power method of vibration signal. According to different wavelet power distribution on each band, the fault position of screw compressors can be obtained. Then, scale-wavelet power spectrum of valve flake vibration signal was got with continuous wavelet transform. According to different wavelet power distribution on scales, condition of valve flake angle can be found. The method builds different parts of screw compressor inherent relation. The experimental results show that the method of using wavelet transform and extracting each frequency-band power and scale-wavelet power spectrum as feature vectors can depict vibration signal changing rules along with working condition of compressor changing correctly. The methods provide a new effective tool for screw compressor conditions monitoring and fault diagnosis.