Vibration Signal Analysis And Fault Diagnosis On Rolling Bearing Of Drilling Pump

Rolling bearing is one of the most familiar components in rotaries; the running states will affect the whole machine’s functions directly. Drilling pump is the key equipment in Oil Field Equipment; failure occurred mainly in the rolling bearings of the power side, and condition monitoring and fault diagnosis for them can be achieved with the rational use and predictive maintenance of the drilling pump to ensure production in safety, which is of important and practical significance.In this paper, the vibration mechanism, failure forms and vibration types of the rolling bearing are analyzed, and the traditional vibration signal analysis methods in time domain and frequency domain are studied. On the background of fault diagnosis of rolling bearing of the drilling pump, vibration signals of the bearing are collected in three operating conditions, the normal operating case, inner race fault case and ball bearing fault case. They are considered as objects of study, which are brought to test the feasibility of the use of vibration signals to monitor the operational status of bearing.The fundamental theory about wavelet and wavelet packet analysis is studied. Since the vibration signal of the rolling bearing is the non-stationary signal, and the traditional Fourier analysis to signal is entirely in the frequency domain, which can not give the changing situation of the signal at some point in time, the wavelet and wavelet packet analysis are very applicable to fault diagnosis of the rolling bearing. The wavelet and wavelet packet analysis to signal are in the time domain and frequency domain at the same time with multi-resolution analysis.According to time-varying characteristics and characteristics of difficult to extract of bearing vibration signal, the wavelet threshold de-noising algorithm is used in the de-noising pre-processing to original vibration signal of the bearing in this paper. The wavelet packet is used in the feature extraction to the bearing vibration signal, and the "energy" is used as the element to structure feature vectors of the bearing vibration signal, so as to provide a approach for the feature extraction to the bearing vibration signal and the follow-up fault diagnosis. The basic theory about fuzzy mathematics is introduced. Finally, the close-degree principle in the fuzzy theory is used in the fuzzy pattern classification to different states of the bearing. Studies have shown that, by means of the wavelet de-noising, the peak and mutation segments of the vibration signal are well retained and the noise disturbance is effectively eliminated. The bearing fault state can be accurately identified by using the wavelet packet method and the pattern recognition method based on fuzzy close-degree. The respective advantages of two methods are brought into play and the diagnostic accuracy of recognition is improved.