Research On Identification Method Of Vibration Features When Drill Bit Approaching Casing

As cluster wells and infill adjustment wells are being widely used in the exploration and development of offshore oil and gas resources in China,the well tracks in shallow formations have become more intensive,which increases the risk that drill-bit collides with adjacent casing during drilling operations.Therefore,there is an urgent need for effective anti-collision monitoring and early warning of the drilling process.The bit induces a series of vibrations during the drilling objects are crushed.It’s great significance for real-time warning of drill-bit collision and scratching of adjacent casing to study the feature extraction method of vibration signal when drilling different objects,and then to establish the identification method for drilling different materials(formation rock,cement sheath,casing)during drill bit approaching casing.This paper investigates the research status of drilling conditions monitoring by using vibration signals at home and abroad.In this paper,dynamic characteristics in process of drilling,the generation,and propagation mechanism of bit vibration waves are summarized.In combination with engineering practice,a wellbore simplified model consisting of rock,cement sheath and casing was constructed.Based on the model,the XY-200 Y drilling rig was used to carry out the laboratory experiment in which the drill-bit approached the casing.A variety of vibration signal samples in different drilling conditions were collected by orthogonally matching the three types of variable of weight on bit,rotate speed and drilling objects.The time-frequency domain analysis,wavelet analysis and Hilbert-Huang transform are applied to analyze the characteristics of vibration signals.So,the characteristic parameters that can reflect the drilling conditions can be selected,and the feature vectors of signals in each drilling condition are constructed.Finally,SVM is used to realize drilling conditions identification and classification.The research results show that the wavelet transform based on generalized threshold can reduce the noises of the original vibration signal,such as white noise and resonance noise.Based on the EMD decomposition,the first 6 orders IMF’s deviation,root mean square,and kurtosis are used as time domain characteristic parameters,and its energy,normalized energy,peculiar frequency band,center of gravity frequency and center of gravity amplitude are used as frequency domain characteristic parameters.Then feature vectors for each drilling condition are constructed.Based on these constructed feature vectors,SVM can meet engineering requirements for the recognition results of different working conditions when bit is approaching the casing.SVM’s overall recognition rate for six kinds of WOB and rotate speed reached 96.1%.The recognition rates of yellow sandstone and red sandstone which have same lithology are 82.2% and 86.7%,of the higher hardness limestone and mudstone are all about 95%.The recognition rate of each condition exceeds 90% when bit approaches the casing form the side and the front.In particular,the recognition rate reaches 100% on condition with significant feature differences that the bit drills casing on the front.The research results can provide effective technical support for anti-collision monitoring and early warning during drilling.