广义预测控制算法在协调控制系统中的应用研究-控制科学与工程;检测技术与自动化装置专业论文.docxVIP

华北电力大学硕士学位论文 华北电力大学硕士学位论文 I I 摘 要 经过近些年的发展建设，大容量、高参数的单元机组已经成为承担电网负 荷的主力机组，而作为电厂自动控制系统中最为核心的组成，协调控制系统对 机组安全、稳定和经济运行起着越来越关键的作用，因而对于协调控制系统提 出了更高的要求。随着现代控制理论的不断发展以及计算机技术的提高，越来 越多的先进控制策略开始应用到协调控制系统中。 本文对广义预测控制算法在协调控制系统中的应用进行了研究。首先，详 细阐述了单元机组协调控制系统的一些基本概念，确定了本课题使用的被控对 象模型；其次，对单变量广义预测控制进行研究，针对其基本算法在大迟延过 程控制中的不足，提出一种改进型算法（AGPC），并且推导了了带有可测外部 扰动的 GPC 算法（FF-GPC），通过仿真验证了 FF-GPC 算法的可行性以及 AGPC 算法优良的控制品质；然后，尝试不同的方法对多变量 GPC 进行解耦， 通过仿真对比，验证了基于分散优化思想的广义预测解耦控制算法的优越性； 最后，采用模糊加权的方法实现多模型控制，结果表明：多模型 GPC 的定值跟 踪能力、抗干扰能力都要好于单模型 GPC，多模型控制策略是解决非线性控制 问题的一种简单可行的办法，具有重要的意义。 关键词：协调控制系统；广义预测控制；解耦控制；多模型控制 II II Abstract By several years of development, units with large-capacity and high-power have become the main units of the grid. As the core of automatic control system of power plant, coordinated control system is becoming more and more important to operating safely, stably and economically. As a result, higher requirements have been put forward. With the development of control theory and computer technology, more and more advanced control strategies have been used in CCS. This paper studied the application of generalized predictive control algorithm in the coordinated control system. First, this paper expounded the basic concepts of coordinated control system. Then the model of controlled object was given. Secondly, the single-variable generalized predictive control was studied. In view of the shortcoming of GPC algorithm in large delay progress, an improved generalized predictive control algorithm (AGPC) was proposed. Then this paper deduced the generalized predictive control algorithm with measurable external disturbances (FF- GPC). Based on simulation, the feasibility of FF-GPC algorithm and the control quality of AGPC algorithm were verified. Thirdly, we tried different approaches to decouple the multivariable GPC. By comparing the simulation result, we found that the decoupling algorithm based on distributed optimization is better. Last, the multi- model control strategy was realized by weig