EEMD-Based Failure Characteristics Principal Component Analysis of Rolling Bearing and PNN Modelling

Using acoustic emission signal measured under different running state of rolling bearing, a fault feature extraction and diagnosis method for rolling bearing based on the ensemble empirical mode decomposition(EEMD) and the probabilistic neural network(PNN) was built. Firstly, the EEMD method was applied to adaptively decompose the signal in time-frequency domain, and the intrinsic mode functions(IMFs) in different frequency bands were analyzed. Then the energy of IMFs was computed, and a failure feature vector was formed by the principal components selected according to the energy contribution analysis. Using PNN to approximate the functional mapping between the feature vector and the fault mode, the failure diagnosis could be realized. Comparison between simulation results and experiment data proves that the proposed method is effective.