【作者】 王娟；

【导师】 刘福才；

【作者基本信息】 燕山大学 ， 控制理论与控制工程， 2003， 硕士

【摘要】 预测控制通过在线求解一个有限时域开环最优控制问题来获得最优预测控制序列，并将当前时刻的控制作用于对象，在下一采样时刻基于新的状态或输出测量值重复上述过程。其主要优点是具有处理控制和状态硬约束的能力。线性模型预测控制已经广泛应用于石油化工等工业过程控制中，但它不适合于强非线性模型预测过程控制，而人们对生产能力及产品质量的要求大大刺激了非线性模型预测控制的发展。因此，研究非线性模型预测控制具有重要的理论意义及应用价值。本文在阅读大量参考文献的基础上，着重研究了几种非线性模型预测控制算法。论文主要分为两大部分，即基于滑模的非线性模型预测控制方法研究和混沌系统的预测控制与同步研究。执行机构引起的非线性，包括饱和、死区、回滞等，这些特性在通常的控制设计中被当作未建模动态而不予考虑，但这样处理将导致对控制器鲁棒性要求的增加。因此我们针对死区非线性特性，采用非线性逆函数的全局I/O线性化方法，提出了具有非线性补偿的广义预测控制算法。将广义预测控制和滑模控制结合起来，提出一种新的广义预测控制方法。该方法具有预测控制在线处理约束及滑模控制对于干扰的不变性的优点。分析了零终端滑模约束的广义预测控制的稳定性。将MPC方法延伸至混沌同步的研究，分别针对连续及离散混沌系统设计控制器，实现两个相同或不同混沌系统之间的控制与同步。用带有终端滑模等式约束的模型预测控制方法，对Hénon系统和Logistic方程的追踪控制和同步进行研究。提高了受控系统抑制参数摄动和随机扰动的能力，改善了控制系统的鲁棒性。在连续时间混沌系统的控制与同步问题统一处理的基础上，给出一种可实现两个相同或不同连续时间混沌系统的控制与同步的预测变结构控制方法。其控制器由针对对象标称非线性模型设计的连续预测控制器和针对对象不确定性设计的滑模变结构控制器组成。所提出的方案具有变结构控制鲁棒性强的优点。更多还原

【Abstract】 Predictive control is a form of control in which the current control action is obtained by on-line solving open-loop optimal control problem, and the above procedure is repeated based on new state and out measurements at the next time. An important advantage of predictive control is its ability to cope with hard constraints on controls and states. Linear MPC has been widely applied in petrochemical and related industries, however, it is inadequate for highly nonlinear predictive processes control. Increasingly stringent demands on throughput and product quality has spurred the development of nonlinear model predictive control (NLMPC). As a result, to study on NLMPC is of great significance theoretically and practically.Based on various references related to MPC, the dissertation develops several NLMPC algorithms, and can be divided into tow main parts, i.e. study on NLMPC based on sliding mode, and predictive control and synchronization of chaotic system. The main contents are as follows:The nonlinear caused by actuators are saturation, dead area, margin loop and so on. During the designing of ordinary controller, these characters are regarded as non-modeling dynamic. If dealing with it this way, it will result in adding the requirement of controller’s robustness. So with the wide I/O linearization method of nonlinear reverse function, the generalized prediction control algorithm with dead area nonlinear compensation is suggested.A new generalized predictive control (GPC) scheme is proposed, which combines the generalized predictive control and the sliding control (SMC). The proposed scheme, which have advantages of GPC and SMC, can deal with system constraints on-line and has strong robustness on the sliding surface. By constraining terminal sliding mode to be zero, the stability of GPC system is analyzed.MPC method is extended to chaotic synchronization research in the paper. We design the controller for continuous and discrete chaotic system respectively which realized the control and synchronization between two<WP=6>homology and different chaotic system.This paper studies the tracking control and synchronization of Hénon system and Logic equation using the model predictive control method with terminal sliding mode equation restriction. We advanced the ability of restraining the parameter perturbation and stochastic distribution and improved the robust performance of the controlled system. We presents a realized predictive variable structure control method to the control and synchronization of two homology or different series-time chaotic systems based on a unified frame for both the control and the synchronization of series-time chaotic system. The controller is composed of predictive controller for nominal nonlinear model and sliding model and variable structure controller for system perturbation. The proposed method has the merit of robustness of variable structure control.更多还原