Fault Diagnosis for Rolling Bearing Based on the (SVM) Combined with (EMD) Instantaneous Power Spectral Entropy

摘要

The dynamic characteristics of the bearing will be complex and nonlinearity when the bearing failure and the fault signal will also exhibit non-stationary. This paper presents a kind of rolling bearing feature extraction method based on the empirical mode decomposition (EMD) and instantaneous power spectral entropy, since the EMD decomposition is adaptive and suitable for processing the nonlinearity and non-stationary signal. The steps of the method are as follow: first decompose the bearing signal into a finite number of IMF components, second process these components with power spectrum, third calculate the information entropy of the power spectrum, at last take the power spectrum entropy as the feature vector, and classify the failures into different type by support vector machine (SVM). Experiments show that 96.25% of the classification is correct which verify that the feature extraction method proposed in this paper is feasible and effective.