Continuous wavelet transform and neural network for condition monitoring of rotodynamic machinery

This paper describes a novel method of rotodynamic machine condition monitoring using a wavelet transform and a neural network. A continuous wavelet transform is applied to the signals collected from accelerometer. The transformed images are then extracted as unique characteristic features relating to the various types of machine conditions. In the experiment, four types of machine operating conditions have been investigated: a balanced shaft; an unbalanced shaft, a misaligned shaft and a defective bearing. The back propagation neural network (BPNN) is used as a tool to evaluate the performance of the proposed method. The experimental results result in a recognition rate of 90 percent.