Electronic Science and Technology ›› 2022, Vol. 35 ›› Issue (7): 46-51.doi: 10.16180/j.cnki.issn1007-7820.2022.07.008

Previous Articles     Next Articles

Drilling Tool Acceleration Denoising Based on GRNN Network Adaptive Filtering

TONG Xiaosen,YANG Jinxian   

  1. School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454000,China
  • Received:2021-01-24 Online:2022-07-15 Published:2022-08-16
  • Supported by:
    National Natural Science Foundation of China(41672363);National Natural Science Foundation of China(U1404510);National Natural Science Foundation of China(61440007);Cultivation Program for Young Backbone Teachers in Colleges and Universities of Henan(2018GGJS061);Innovative Science and Technology Talents Team Construction Project of Henan(CXTD2016054);Young Backbone Teachers of Henan Polytechnic University(2017XQG-07)


In view of the problem of serious distortion of gravitational acceleration information caused by bottom-hole vibration and noise during measurement while drilling, a method that combines generalized regression neural network and improved recursive least squares is proposed to remove noise. Using the principle of mutual cancellation of adaptive noise, the accelerometer measurement signal and drilling tool vibration signal are used as the main noise model, and the drilling tool vibration signal processed by the generalized regression neural network is used as the auxiliary noise model. Recursive least squares filter processing is used to improve the accuracy of drilling tool accelerometer measurement and calculate real-time drilling attitude information. The simulation results show that the proposed algorithm can effectively remove the vibration and noise of the drilling tool accelerometer and improve the attitude measurement accuracy. After denoising, the calculated deviation angle error is within 1.45°, and the tool face angle error is within 1.65°.

Key words: measurement while drilling, adaptive noise cancellation, gravity extraction, generalized regression neural network, recursive least squares algorithm, accelerometer denoising, adaptive filtering, attitude calculation

CLC Number: 

  • TN713