自动化学报2025，Vol.51Issue(10)：2313-2323,11.DOI:10.16383/j.aas.c250136
旋转导向钻井工具系统实时测量的智能粒子滤波方法
Intelligent Particle Filter for Real-time Measurement of Rotary Steerable Drilling Tool System
摘要
Abstract
To address the real-time measurement challenge of Toolface in rotary steerable drilling tool system,this paper proposes an intelligent particle filtering algorithm based on deep learning.Initially,the particle selection mechanism guided by conditional generative adversarial network(CGAN)is established to tackle the issues of particle impoverishment and degeneracy in particle filtering.In this mechanism,the generator network optimizes the sampling distribution through adversarial training,producing a high-quality set of particles;The discriminator evaluates the probability of the generated particles within the true posterior distribution,guiding the particle weight calculation.Subsequently,the covariance matrix estimator is designed based on a deep residual network(ResNet)to address the unknown but time-varying noise covariance matrices in complex downhole conditions.This module is integrated with the CGAN-guided particle filter in an end-to-end manner,forming a closed-loop optimiza-tion system.The ResNet module benefits from the model information in the particle filtering algorithm and provides the particle filter with estimates of the covariance matrices.Finally,experiments are conducted on the rotary steer-able drilling tool platform.The results demonstrate that the proposed algorithm effectively addresses the real-time measurement issue of Toolface and exhibits higher accuracy compared with existing algorithms.关键词
智能粒子滤波/旋转导向钻井工具系统/实时测量/深度学习算法/未知噪声协方差矩阵Key words
Intelligent particle filter/rotary steerable drilling tool system/real-time measurement/deep learning al-gorithm/unknown noise covariance matrix引用本文复制引用
盛立,刘一凡,高明,周东华..旋转导向钻井工具系统实时测量的智能粒子滤波方法[J].自动化学报,2025,51(10):2313-2323,11.基金项目
国家自然科学基金(62473379,62173343,62033008),山东省自然科学基金(ZR2024MF072,ZR2022ZD34,ZR2025ZD02),山东省泰山学者项目研究基金资助Supported by National Natural Science Foundation of China(62473379,62173343,62033008),Natural Science Foundation of Shandong Province(ZR2024MF072,ZR2022ZD34,ZR2025ZD02),and Research Fund for the Taishan Scholar Project of Shandong Province of China （62473379,62173343,62033008）
