Fault Diagnosis Based On RBF Neural Network

The producing process of petroleum drilling is a continuous process, if it goes wrong, the economic loss will be very seriously. The drilling equipment is one kind of large equipments, has many transmission mechanisms, and its rolling bearings easily go wrong. This paper studied the fault diagnosis and forecast of the drilling rolling bearings, and prevented faults from occurring by this method.The fault feature information can not be obtained availably by regular methods to extract the features of nonstationary vibration signals produced by fault rolling bearings. The paper used the principle of wavelet to analyze and process the vibration signal data of the rolling bearings, used RBF neural network to diagnose the faults, and the result of the fault diagnosis was very well.Because there was lots of background noise in the vibration signals of the rolling bearings, the real fault feature information could not be obtained availably if the features were exacted directly. The paper used the principle of wavelet denoise to preprocess the vibration signals, and several methods of selecting the thresholds of wavelet denoise were also introduced. The effect of anti-ground noise was perfect with the methods of wavelet denoise, and improved the signal-to-noise rate. According to the characteristics of the vibration signals of the rolling bearings, a method named wavelet entropy was researched to extract the features, and was used to analyze the three types of vibration signals of the rolling bearings. The results showed that the real fault features of the vibration signals could be extracted very well by this method.On the basis of feature extraction, the paper used RBF neural network with k-means algorithm to classify the feature vectors, which were extracted by the wavelet entropy method and put into the RBF neural network,...