Bearing Fault Diagnosis Based on Empirical Mode Decomposition and Neural Network

Bearings are widely used in many equipments and its operating state directly concerns the performance of the whole machinery. In this paper, empirical mode decomposition method is firstly used to analyze the signals of different fault types of the bearing and extract the feature vectors. By comparing the performances of different BP networks using three different algorithms on the training data, then BP network using Levenberg-Marquardt algorithm is chosen to detect and diagnose the test data of the bearing. The result proves the effectiveness of the combined method for the bearing diagnosis.