| 引用本文: | 曹文祺,李少远.具有可参数化不确定性系统的对偶自适应模型预测控制[J].控制理论与应用,2019,36(8):1197~1206.[点击复制] |
| CAO Wen-qi,LI Shao-yuan.Enhanced parameterizable uncertainty to dual adaptive model predictive control[J].Control Theory and Technology,2019,36(8):1197~1206.[点击复制] |
|
| 具有可参数化不确定性系统的对偶自适应模型预测控制 |
| Enhanced parameterizable uncertainty to dual adaptive model predictive control |
| 摘要点击 2266 全文点击 1428 投稿时间：2018-05-02 修订日期：2018-11-07 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2018.80324 |
| 2019,36(8):1197-1206 |
| 中文关键词 模型预测控制 自适应控制 对偶控制 不确定性 动态系统 |
| 英文关键词 model predictive control adaptive control dual control uncertainty dynamic systems |
| 基金项目 国家自然科学基金重点项目 |
|
| 中文摘要 |
| 控制系统中存在的不确定性为其性能优化带来诸多问题. 自适应控制和鲁棒控制是针对不确定性采取的不同设计策略; 前者没有充分考虑系统的未建模动态, 而后者往往是针对不确定的最大界而设计, 具有较强保守性. 本文试图将自适应控制和鲁棒控制的策略结合, 提出一种在模型预测控制中利用未来不确定信息的对偶自适应模型预测控制策略. 该策略将系统中由未建模动态引起的不确定性参数化表达, 并为其设定边界约束, 作为优化问题中新的约束, 在优化控制目标的同时减小系统不确定性对控制的影响. 仿真结果表明, 本文提出的算法较传统自适应模型预测控制算法, 对于系统存在的不确定性由于在迭代过程中采用参数化描述, 得到了更好的系统性能, 且具有更好的收敛性. |
| 英文摘要 |
| Control performance always deteriorates because of the uncertainty in control systems. Adaptive control and robust control are two different strategies against system uncertainties. Adaptive control decreases uncertainty through updating model parameters without fully considering unmodeled dynamics, while the latter maintains control performance under strong conservation by setting the upper bound of the uncertainty in systems. This work tries to combine robust control and adaptive control, and presents a novel dual adaptive model predictive control with enhanced parameterizable uncertainty of future. The uncertainty causing from unmodeled dynamics is parameterized, and the control performance is enhanced by adding a constraint of parameterized uncertainty in the optimization problem with designing its upper bound. This new approach optimizes model predictive control performance and decreases the effect of uncertainty at the same time. Simulation result shows that compared with traditional adaptive model predictive control, our approach costs less control energy to achieve the same performance and is of better convergency. |
|
|
|
|
|