Drill Pipe Fault Diagnosis Based On Wavelet Transform And SVM

With the rapid economic development in China,drilling rigs have played an increasingly important role in infrastructure construction.During the piling process,the problems of the drill pipe falling caused by the drill pipe damage or the bolt loosening have occurred frequently,and the issue is more and more remarkable.In this paper,drill pipe damage and bolt loosening were studied.First,the research platform used to simulate the working principle of the rotary drilling rig was set up for research on fault diagnosis,and experimental research was carried out based on the research platform.Through the simulation of the state of the equipment,the vibration signals were collected under three states of normal,drill pipe damage and bolt loosening.By processing the collected signals,which is based on the wavelet packet energy decomposition algorithm,the fault features were extracted and the fault feature vectors were constructed.To classify the drill pipe's vibration signals,the SVM models were trained by the constructed feature vectors.Besides,the method of wavelet-fractal was used to classification the state of the drill pipe,and in the process of denoising the collected signals,an improved threshold function was introduced.The result shows that denoising with the improved threshold function can achieve good results.The trained SVM model and the wavelet-fractal method both can achieve good classification results,and the latter is better.