旋转导向系统稳定平台自适应动态面控制  

Adaptive dynamic surface control of stabilized platform in rotary steerable drilling system

作　　者：万敏[1] 宋佳儒 黄山山 陈苗苗[1] WAN Min;SONG Jiaru;HUANG Shanshan;CHEN Miaomiao(School of Mechatronic Engineering,Southwest Petroleum University,Chengdu 610500,China)

机构地区：[1]西南石油大学机电工程学院,四川成都610500

出　　处：《中国石油大学学报（自然科学版）》2024年第5期190-199,共10页Journal of China University of Petroleum(Edition of Natural Science)

基　　金：四川省自然科学基金创新研究群体(2023NSFSC1980)。

摘　　要：井下的多种干扰因素为旋转导向系统稳定平台的控制器设计增加了复杂性。为了应对未知摩擦力矩和建模误差对稳定平台的不良影响,提出一种自适应神经网络动态面控制方法,该方法使用RBF神经网络逼近摩擦及干扰力矩,设置状态观测器获取由于相关参数不确定导致的建模误差,并引入动态面方法避免传统反步控制带来的“微分爆炸”,最后使用李雅普诺夫法证明系统的稳定性。结果表明,该控制方法在稳定平台模型存在摩擦力矩、未知干扰和建模误差的情况下,仍能使工具面角准确、快速地跟踪输入指令信号,具有较好的自适应性和鲁棒性。A variety of interference factors underground increase the complexity of the controller design of the stabilized platform of the rotary steerable drilling system.In order to deal with the adverse effects of unknown friction torque and modeling error on the stabilized platform,an adaptive neural network dynamic surface control method was proposed.The RBF neural network was used to approximate the friction and disturbance torque,and the state observer was set to obtain the modeling error caused by uncertainty of the relevant parameters.The dynamic surface method was introduced to avoid the“differential explosion”caused by the traditional backstepping control.Finally,the Lyapunov method was used to prove the stability of the system.The results show that the controller can make the toolface angle track the input command signal accurately and quickly under the condition of friction torque,unknown interference and modeling error in the stabilized platform model,and has good adaptability and robustness.

关 键 词：旋转导向系统 稳定平台 动态面控制 RBF神经网络 状态观测器 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]