一类非线性系统预测切换控制方法研究

Model predictive control is by now the most influential modern control technology in the industrial world.The control move is updated by means of a receding horizon strategy which requires to slove a constrained optimazation problem on line, which limited its application for the "fast process" systems with fast sampling rate and frequent dynamic changes. This project will consider the approximate explicit model predictive control for a class of nonlinear switching systems in presence of both input and state constraints. The aim is to simplify the current existing model predictive control algorithm, and make it also apply to "fast process" systems with small sampling time and not allowing to slove the optimization problem on line. A switching scheme will be designed to maximize the domain of the feasible set, and guarantee stability, constraints satisfaction, and even certain performance requirements. This project will also employ the closed-loop constrained robust optimal control methodology and formulate the approximate explicit robust model predictive control algorithm for a class of nonlinear switching systems with model uncertainties and external disturbances. The experimental simulation will also be carried out to illustrate the effectiveness of the results, and the dynamic and static performance of the closed-loop systems will also be analyzed.

模型预测控制是迄今为止对工业界最有影响的先进控制策略，但是因为其应用的"滚动优化"的思想在线求解优化问题，所以，在很大程度上限制了其在采样时间断、变化频繁的"快过程"中的应用。本项目将针对一类同时具有输入和状态约束的非线性切换控制系统，采用近似显式模型预测的方法设计切换控制器，并力求对现有的模型预测控制算法进行简化，使其可以适用于采样时间短以至于不允许在线实时求解优化问题的"快过程"，并且设计切换控制机制使整个闭环系统满足稳定性，状态和输入的约束条件以及一定得性能要求，并使其可行域的范围增大。本项目还将利用闭环鲁棒约束优化问题的研究方法，提出同时具有模型不确定性和外界扰动的非线性切换控制系统近似模型预测方法。此外，本项目还将对所提出的理论方法进行实验仿真和验证，分析闭环系统的动、静态性能。

本项目分别针对同时具有输入和状态约束的线性系统和非线性大规模控制系统进行了分布式预测控制方法的研究。采用可分解预测控制方法设计分布式预测控制器，并力求对现有的模型预测控制算法进行快速求解，使其可以适用于采样时间短的在线实时求解优化问题的“快过程”，并且设计自驱动预测控制使整个闭环系统满足稳定性，状态和输入的约束条件以及一定得性能要求，并使其计算速度加快。本项目还将利用闭环鲁棒约束优化问题的研究方法，提出同时具有模型不确定性和外界扰动的非线性鲁棒预测方法。此外，本项目还将对所提出的理论方法进行实验仿真和验证，分析闭环系统的动、静态性能。