Comparison of Instantaneous Properties Extraction Methods and Execution of Optimal Empirical Mode Decomposition on Faulty Ball Bearing Experimental Data

Empirical mode decomposition (EMD) is one of the new methods for decomposing a signal into its constituent components. The existence of multiple error sources has led to activities to eliminate or mitigate their effects. In this research, one of the major problems of EMD for the separation of noise-polluted signals, namely, mode mixing problem has been studied. To solve this problem, bandwidth EMD has been used, which enhances the EMD method and processes speed and greatly prevents mode mixing problem. Also, among the available methods to extract the instantaneous properties, the proper pair of instantaneous properties identification and signal normalization method is presented by an example. To investigate the efficiency of the bandwidth EMD method, using the optimal method of extracting the instantaneous properties, the experimental data of a faulty bearing have been studied and the instantaneous properties of both EMD method and the bandwidth EMD method have been extracted. Using the coefficient of variation criterion, it is shown that the bandwidth EMD method has a higher resolution and better results than EMD method. Finally, using information of decomposed white noise by EMD, the noise isolation quality of the original data is examined, which indicates a better decomposition of the results of the bandwidth EMD method.