Multi-resolution Correlation Entropy and Its Application on Rotating Machinery Vibration Signal Analysis

A new feature parameter of vibration signal used in rotating machinery fault diagnosis method is analyzed, and in this paper, it is named multi-resolution correlation entropy (MRCE). After extracting the rolling bearing vibration signal, the denoised signal should be transformed by wavelet packet into several sub-signals which are attached to different frequencies. Next, the wavelet packet correlation coefficient will be calculated. Combined with the information entropy theory, the MRCE is obtained. The signal classification and state identification are realized by support vector machine (SVM) intelligent algorithm, and the results reflect the truth that applying MRCE as the feature index to detect the working state of rotating machinery can get good diagnostic accuracy, so that MRCE can be used in rotating machinery fault diagnosis in the future.