Drill Pipe Fault Identification Method Of Rotary Drilling Rig Based On EMD Principle And BP Neural Network

As an important equipment in modern pile foundation engineering,rotating drill is widely used in many fields:such as oil and gas exploitation,mine excavation,cross-sea bridge and so on.However,the drill off accident often occurs during actual working process of the rotating drill.There are no fewer accidents due to the pipe failure.Therefore,it is of great significance to identify drill pipe fault to achieve the state detection and fault diagnosis of rotating drill.Based on above purpose,this thesis conducts the research on fault feature identification and extraction of rotary drill pipe under laboratory conditions.the main research contents and work are summarized as follows:1)In this thesis,by simulating the working principle of the drilling rig,the platform is designed.The detailed scheme design and calculation are made for the components of the drilling experimental bench.Two kinds of faults:the loose connection of drill pipe and the damage of the drill pipe are simulated.This thesis also uses the test bench to collect the signal samples of different speed and different conditions.2)The principle of EMD algorithm and the related spectral analysis methods of signal processing are studied.This thesis reduces the noise of actual drill pipe signal by wavelet threshold method.The EMD method is used to achieve adaptive separation of drill pipe signals effectively.The time spectrum,AR spectrum and marginal spectrum of drill pipe fault signal are analyzed.3)Aiming at the drill pipe fault identification,two methods:EMD-IMF energy,EMD-SVD are proposed to extract the fault feature of the drill pipe.The eigenvalues are extracted and analyzed for the signals of different working samples collected.The results show that both two methods can effectively identify the characteristics of the drill pipe failure.4)The principle of BP neural network is studied.Two feature extraction methods:EMD-IMF energy and EMD-SVD are used to construct the fault eigenvectors for signals of different speeds and conditions.The two methods are combined with BP neural network.Two kinds of failure are diagnosed and classified.The results show that two kinds of methods based on EMD can be used to extract the fault feature of drill pipe.EMD-IMF energy method has a higher recognition success rate for the loose connection of drill pipe.In practical engineering,this method can be used to identify and diagnose the loose connection of drill pipe.