Machine fault feature extraction based on wavelets and recurrence quantilification analysis

This paper presents a simple and efficient machine fault feature extraction approach based on the wavelet transform and recurrence quantification analysis (RQA). This approach first decomposes the signals into several layers using discrete wavelet transform (DWT), then features are extracted from each decomposition based on RQA. The features contain the informative attributes of the signals. Then, machine faults are diagnosed based on these feature vectors using a probabilistic neural network. In the experimental process, features are extracted by 3 ways, DWT, RQA, and DWT combined with RQA. The experimental results from the DWT combined with RQA on bearing fault diagnosis have shown that the proposed approach is able to effectively extract important intrinsic information content of the test signals and increase the overall fault diagnostic accuracy as compared to conventional methods.