引用本文:徐驰,赵希梅.永磁直线同步电动机智能递归非奇异终端滑模控制[J].控制理论与应用,2022,39(7):1242~1250.[点击复制]
XU Chi,ZHAO Xi-mei.Intelligent recursive nonsingular terminal sliding mode control of permanent magnet linear synchronous motor[J].Control Theory and Technology,2022,39(7):1242~1250.[点击复制]
永磁直线同步电动机智能递归非奇异终端滑模控制
Intelligent recursive nonsingular terminal sliding mode control of permanent magnet linear synchronous motor
摘要点击 1289  全文点击 463  投稿时间：2021-08-05  修订日期：2022-06-24
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2022.10711
  2022,39(7):1242-1250
中文关键词  永磁直线同步电动机  递归非奇异终端滑模控制  不确定性  有限时间  抖振  双隐层径向基函数神经网 络
英文关键词  permanent magnet linear synchronous motor  recursive nonsingular terminal sliding mode control  uncertainties  finite time  chattering  double-hidden-layer radial basis function neural network
基金项目  辽宁省自然科学基金重点项目(20170540677)资助
作者单位E-mail
徐驰 沈阳工业大学电气工程学院 xuchi@smail.sut.edu.cn 
赵希梅* 沈阳工业大学电气工程学院 zhaoxm_sut@163.com 
中文摘要
      为解决永磁直线同步电动机(PMLSM)在运行过程中易受参数变化、外部扰动、摩擦阻力等不确定性因素 影响的问题, 本文提出一种基于双隐层径向基函数神经网络(DRBFNN)的递归非奇异终端滑模控制(RNTSMC)方法 来提高PMLSM系统的控制性能. 首先, 分别构造非奇异终端滑模面和递归积分终端滑模面, 使得两滑模面依次连续 到达, 可在削弱抖振的同时保证跟踪误差在理论上的有限时间内收敛至零. 但由于系统不确定性的边界难以确定, 因此引入具有更高拟合精度和泛化能力的DRBFNN对不确定性进行逼近和补偿, 并通过在线自适应更新连接权重, 进一步提高神经网络的逼近能力. 最后, 系统实验结果表明, 该方法能够有效抑制不确定性对系统的影响, 提高了系 统的位置跟踪精度, 并使系统具有较强的鲁棒性.
英文摘要
      In order to solve the problem that permanent magnet linear synchronous motor (PMLSM) is easily affected by the parameter variations, external disturbances, friction and other uncertain factors during operation, a recursive nonsingular terminal sliding mode control (RNTSMC) method based on double-hidden-layer radial basis function neural network (DRBFNN) is proposed to improve the control performance of PMLSM system. Firstly, the nonsingular terminal sliding surface and the recursive integral terminal sliding surface are constructed respectively to make the two sliding surfaces arrive successively, which can weaken the chattering and ensure that the tracking error converges to zero in theoretical finite time. However, it is difficult to determine the boundary of the system uncertainties, so the DRBFNN with higher fitting accuracy and generalization ability is introduced to approximate and compensate the uncertainties, and the online adaptive updating is used to weight to further improve the approximation ability of neural network. Finally, the system experiment results show that the method can suppress the influence of uncertainties on the system, which effectively improve the position tracking accuracy of the system and make the system have strong robustness.