| 陈建勇.死区非线性输入系统的自适应迭代学习控制[J].高技术通讯(中文),2022,32(11):1134~1142 |
| 死区非线性输入系统的自适应迭代学习控制 |
| Adaptive iterative learning control of nonlinear systems with input dead-zone |
| |
| DOI：10.3772/j.issn.1002-0470.2022.11.004 |
| 中文关键词: 迭代学习控制（ILC）； 非对称死区输入； 非线性系统； 神经网络； 微分-差分学习律 |
| 英文关键词: iterative learning control (ILC), asymmetric dead-zone, nonlinear system, neural network, differential-difference learning law |
| 基金项目: |
| 作者 | 单位 | | 陈建勇 | (浙江工业大学信息工程学院杭州 310023) |
|
| 摘要点击次数: 675 |
| 全文下载次数: 514 |
| 中文摘要: |
| 针对一类含有非对称死区输入和任意初态的非线性系统，本文提出了一种实现有限作业区间跟踪控制的神经网络迭代学习控制（ILC）算法。构造新的修正函数形式设计校正参考轨迹，放宽了迭代学习控制初值一致要求。利用径向基函数(RBF)神经网络估计和补偿系统的不确定性及死区参数，从而设计迭代学习控制器。引入一级数收敛序列用于处理重构误差对系统跟踪性能的影响，并给出了未知参数的微分-差分学习律。理论分析表明，该控制器能够实现系统状态在预指定作业区间上对参考轨迹的零误差跟踪。最后的仿真结果验证了所提控制算法的有效性。 |
| 英文摘要: |
| For a class of nonlinear systems in the presence of asymmetric dead-zone input and arbitrary initial state, a neural network iterative learning controller is presented to realize the tracking control of finite operation interval. A rectified desired trajectory is designed by constructing a new correction function, which relaxes the consistent initial value of iterative learning control. The radial basis function (RBF) neural network is used to estimate and compensate the system uncertainties and dead-zone parameters; thus, the iterative learning controller is obtained. A series convergence sequence is introduced to deal with the influence of reconstruction error and the differential-difference learning law of unknown parameters is given. Theoretical analysis shows that the controller can achieve perfect tracking of the desired trajectory on the specified interval. Finally, simulation results demonstrate the effectiveness of the learning control schemes. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
