Research On Intelligent Diagnosis Method Of BHA Vibration Mode And Driving Mechanism

Severe vibration of drill string is one of the main causes for drill string failure,which seriously threatens drilling safety and greatly increases drilling cost.The vibration forms of the drill string include axial vibration,lateral vibration,and torsional vibration,among which lateral vibration mostly occurs on the bottom hole assembly(BHA),while torsional vibration may occur on the drill string,BHA,and bit.These two forms of vibration have strong destructive power to the BHA.In some cases,torsional vibration can be monitored at the surface,but lateral vibration is difficult to identify at the surface.Therefore,it is of great significance to study the vibration mode and driving mechanism of BHA and establish an intelligent identification method for BHA vibration mode and driving mechanism,which can provide reference for surface engineers to make targeted suppression measures for different vibration modes,and improve drilling efficiency,reduce downhole accident rate and drilling cost.Firstly,based on the generalized Hamilton virtual work principle and the assumption of Euler-Bernoulli beam,a BHA finite element dynamic model considering the geometric nonlinearity,contact nonlinearity and external damping of the drill string is established.A comprehensive bit force model was developed considering the bit-rock interaction,BHA eccentricity,and mud fluid dynamics.The Newton-Raphson method and Newmark-β method are used to solve the model respectively,and the nonlinear static analysis and dynamic analysis of BHA are realized.By simulating different vibration modes,the displacement,velocity,and acceleration characteristics of the two most severe forms of lateral vibration and torsional vibration,whirl,and stick-slip are studied,and the characteristic differences of acceleration in time domain and frequency domain under different vibration modes are analyzed.Secondly,after completing the numerical simulation of the BHA finite element dynamic model,a two-dimensional BHA kinematic model is established to simulate the high-frequency torsional vibration and whirl in the downhole.The mean value,peak value and root mean square value of the simulated acceleration data are calculated to obtain the time domain characteristics of the data.The fast Fourier transform(FFT)and short time Fourier transform(STFT)are used to extract the frequency domain and time frequency characteristics of the data,and the label-data library of the vibration acceleration data of the drill string under different vibration modes is established.Thirdly,relying on the above-mentioned label-data library,an automatic downhole drill string vibration pattern recognition method is established by using the naive Bayesian method and features of time domain,FFT spectrum,and STFT time frequency.The 300 sets of simulated data in the label-data library are split into a training set(250sets)and a test set(50 sets).After completing the model training using the training set,the accuracy of the model is verified by the test set.The results show that the recognition rates of stick-slip vibration,forward whirl and backward whirl are about 90%,90.2% and87.5% respectively for time domain features,and about 80%,85.7% and 90%respectively for frequency domain features.After sufficient research and summary,it is proposed to identify the driving mechanism of BHA vibration by grouping the excitation sources with the same characteristics into one category as the object of driving mechanism identification and by identifying the category corresponding to the physical excitation.Finally,the vibration pattern recognition analysis and driving mechanism diagnosis are carried out on the three-axis acceleration data of a vertical well,which verified the applicability of the proposed method.The automatic diagnosis method of downhole BHA vibration mode and driving mechanism described in this paper has certain guiding significance for the development of more accurate and intelligent downhole vibration diagnosis system.