多智能体系统的一致性数据驱动最优迭代学习控制  

Data-driven optimal iterative learning control for the consensus of multi-agent systems

作　　者：耿燕 常杜辉 贺兴时[1] GENG Yan;CHANG Duhui;HE Xingshi(School of Science,Xi’an Polytechnic University,Xi’an 710048,China)

机构地区：[1]西安工程大学理学院,陕西西安710048

出　　处：《西安工程大学学报》2025年第2期118-126,共9页Journal of Xi’an Polytechnic University

基　　金：陕西省自然科学基础研究计划项目(2020JQ-831);陕西省教育厅专项科研计划项目(20JK0642)。

摘　　要：为提高多智能体系统的跟踪性能和放宽算法的收敛性条件,设计一种数据驱动的最优迭代学习控制策略。针对一类线性时不变的多智能体系统,通过最小化预测输出与实际输出的残差与相邻估计差值的和,构建参数估计算法来估计系统参数。以虚拟领导者来替代期望轨迹,在通讯拓扑的基础上,通过优化智能体一致跟踪误差与控制差值和的指标函数,并将估计的参数嵌入到学习律中,设计了最优迭代学习控制律。结果表明参数估计误差有界,系统的跟踪误差单调收敛。通过数值仿真验证了设计的控制策略的有效性。To improve the tracking performance of multi-agent systems and relax the convergence condition of the algorithm,a data-driven optimal iterative learning control strategy is proposed.For a class of linear time-invariant multi-agent systems,a parameter estimation algorithm is constructed to estimate the system parameters by minimizing the residual error between the predicted output and the actual output,as well as the difference between adjacent estimates.The virtual leader is used instead of expected trajectory,based on the communication topology,the optimal iterative learning control law is designed by optimizing the index function of the agent′s consistent tracking error and control input difference,meanwhile,incorporating estimated parameters into the learning process.The results show that the error of parameter estimation is bounded and the tracking error of the system converges monotonically.The numerical simulations validate the effectiveness of this designed control strategy.

关 键 词：迭代学习控制 多智能体系统 数据驱动 参数估计算法 最优控制 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]