Abstract
A novel working angle recognition system of screw compressor valve using wavelets and Probabilistic Neural Networks for engineering application is proposed. According to tested valve vibration signal energy changing characteristic under different working angle of valve flake, the paper obtains vibration signal power distribution on different scales using continuous wavelet transform, that is, scale-wavelet power spectrum of vibration signal. Then, according to different wavelet power distribution on scales, a Probabilistic Neural Networks model is applied to classify corresponding working angle of valve flake using scale-wavelet power spectrum as feature parameters. The experimental results show that the methods of using continuous wavelets transform and extracting each scale-wavelet power spectrum as feature vectors can depict vibration signal changing rules along with valve flake angle of compressor changing correctly. The Probabilistic Neural Networks model applied can recognize different working angles and realize monitoring of valve working condition well. The methods provide a new effective tool for screw compressor valve states monitoring and fault diagnosis.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call