Vibration De-noising Method For Inertial Measurement With Drilling

Drilling with measurement is the main and most commonly used technique for underground excavation and exploration,which provides accurate navigation of downhole drilling direction based on the attitude information solved in real time with drilling during the drilling process.During the drilling process,vibration of drilling tools is an unavoidable,random and complex phenomenon,which mainly consists of three forms: longitudinal vibration,lateral vibration and torsional vibration.The vibration of the drilling tool makes the signals collected and outputted by the inertial follow-drill measurement devices installed near the drill head contain a large amount of interference from the vibration signals,resulting in poor accuracy of the follow-drill attitude solution.In order to obtain accurate data from the follow-drill measurement signals containing a large number of vibration signals,the follow-drill vibration signals need to be denoised.With the research objective of extracting and identifying the vibration signal features in the context of follow-drill vibration,a method of follow-drill vibration identification based on impulse neural network is proposed,and then the lateral vibration,longitudinal vibration and torsional vibration signals are targeted for denoising according to the three identified follow-drill vibration signals.In summary,the main contents of this research paper are as follows:(1)The characteristics of the drilling vibration signals are briefly analyzed,and the different features of transverse vibration,longitudinal vibration and torsional vibration are used to identify the drilling vibration signals in combination with impulse neural networks.(2)Combining the identified longitudinal vibration signal characteristics,a longitudinal vibration signal measurement model based on quadratic transformation is constructed,and the longitudinal vibration signal is denoised by using an improved traceless Kalman iterative filtering algorithm.(3)Combine the identified transverse vibration signal characteristics,build a transverse vibration signal model,use the principle of mutual correlation between acceleration signal and magnetometer signal under transverse vibration,build a reference signal,combine Kalman filter to extract the mutual correlation of transverse vibration signal,and realize the denoising of transverse vibration.(4)Combining the identified torsional vibration signal features,a non-local block-based bottleneck structure filter with encoding-decoding is designed based on the combination of local block and convolution filter on the basis of the identified torsional vibration signal features,which can maximize the denoising ability of the torsional vibration signal with drilling.There are 24 figures,5 tables,and 73 references.