控制理论与应用2026，Vol.43Issue(4)：905-914,10.DOI:10.7641/CTA.2025.40318
数据驱动的未知复杂系统在线滚动辨识与优化控制
Data-driven online rolling horizon identification and optimization control for unknown complex system
摘要
Abstract
This paper focuses on the method of identifying unknown complex systems and designing predictive control using only finite input-state data.The proposed method's core innovation lies in establishing a linear model for the nonlinear system based on the time series as the system operates,and continuously identifying the model parameters through rolling horizon techniques,thereby designing a model predictive controller.Firstly,the proposed identification method considers a finite number of noisy input-state data of the unknown system with disturbance,by combining the set operation and solving a series of optimization problems to obtain a linear state space model that can describe the data.Based on this model,a model predictive control(MPC)algorithm is used to compute the control input.By measuring the system state at the next moment,the input-state data set is updated and then the next round of system identification and control input computations starts.The experimental results of linear system and intelligent water supply system show that the proposed method can realize online rolling identification and optimal control of unknown complex systems.关键词
在线滚动辨识/模型预测控制/奇诺多面体/智能供水系统Key words
rolling horizon identification/MPC/zonotope/intelligent water supply system引用本文复制引用
杨少布道,傅安琪,乔俊飞..数据驱动的未知复杂系统在线滚动辨识与优化控制[J].控制理论与应用,2026,43(4):905-914,10.基金项目
科技创新2030-"新一代人工智能国家科技重大专项"重大项目(2021ZD0112301),北京市自然科学基金项目(L221005),国家自然科学基金项目(62 003009,62021003,61890930-5)资助. Supported by the National Key Research and Development Program of China(2021ZD0112301),the Beijing Natural Science Foundation(L221005)and the National Natural Science Foundation of China(NSFC)(62003009,62021003,61890930-5). （2021ZD0112301）
