Research On Intelligent Diagnosis Methods For Bearing Based On Hilbert-Huang Transform

Rolling bearing is one of the most commonly used and liable to be damaged components in the rotating machinery. Many faults of rotating machinery are related to rolling bearings and the running state of them directly influences the performance of the whole machine. Therefore, it is of practical significance to study fault diagnosis technology for rolling bearings.In this paper, a new time-frequency analysis method, Hilbert-Huang Transform (HHT) and Adaptive Network-based Fuzzy Inference System (ANFIS) were combined and applied in the fault diagnosis of the rolling bearing, so that we could carry out intelligent diagnosis for rolling bearing. The details were studied as follows: building up experimental equipment, designing schemes to collect vibration signals of rolling bearings, analyzing the signals effectively and extracting feature to identify the different pattern of bearings, and then combining neural network with fuzzy logic to classify the different fault pattern of rolling bearings.HHT method consists of two successive parts, i.e., the Empirical Mode Decomposition (EMD) and the Hilbert transform. EMD is important which decomposes the complicated signal into a number of Intrinsic Mode Functions (IMF), then the Hilbert transform is performed on each IMF, and the Hilbert spectra of all IMFs are grouped to get the Hilbert spectrum of the original signal, which possesses high time-frequency resolution. Through analyzing simulated signal and real vibration signal of rolling bearing by HHT method and other time-frequency analysis methods, the conclusion was deduced that HHT method is valid and superior in the field of processing non-stationary signals.During the practical application, an improved sifting stop guide was applied to enhance the speed and precision of EMD. Meanwhile, in order to get accurate IMFs, the EMD method based on envelop and demodulation technology was proposed to decrease the effect caused by noise and highfrequency intrinsic vibration of fault vibration signal. Then we could extract characteristic information effectively.Based on the EMD diagnosis, the fuzzy neural network technique was applied to carry out intelligent diagnosis. We chose ANFIS and BP network respectively to classify the rolling bearing fault patterns. And the EMD-based characteristic parameters and the wavelet analysis-based characteristic parameters were put into the networks for comparison. It is shown that the EMD-based feature parameters can present fault feature of rolling bearings more correctly and effectively. When ANFIS is combined with the EMD-based feature parameters, the network can identify the mode of bearing accurately.