【作者】 高军伟；

【导师】 贾利民；

【作者基本信息】 铁道部科学研究院 ， 交通信息工程及控制， 2003， 博士

【摘要】 本文重点研究了一类重要的混杂系统—切换系统的建模与控制理论，并将研究结果应用到焦炉加热自动控制系统中，实现了对复杂工业过程系统的控制。 论文提出了智能切换控制的概念，智能切换控制将智能控制理论与切换系统理论有机结合起来，以切换控制为主体，将智能控制的理论与方法应用到切换系统的建模、分析与控制中。智能切换控制是切换系统发展的高级阶段，是解决复杂系统控制问题的重要方法。 文章提出了切换系统中被控对象动态区间软划分的方法，降低了系统分析与建模的复杂性。动态区间软划分后得到的每一个动态子区间都是一个既包含连续变量和离散变量又具有两种变量相互作用的复杂集合体，同时每一个动态子区间又都是具有其动态特性和变化规律的独立整体。分析了动态子区间的区间重叠、区间漂移和区间缩放特性，其中区间重叠是子区间的外部特征，而区间漂移和区间缩放则是子区间的内部特征。基于T-S模型的模糊神经网络被用作子区间的参数辩识工具。 在复杂系统动态区间软划分的基础上提出了柔性切换控制。柔性切换是针对复杂系统控制问题的一种控制策略，柔性切换可以同时包含硬切换和软切换，其切换策略是既受连续变量驱动，又受离散事件驱动，而且切换具有多样性和层次性。针对时变滞后大惯性系统提出了半闭环控制，基于有界控制的半闭环控制的控制目标首先是要满足输出的有界稳定，其次才是满足输出的无差性，是对人类控制器控制复杂系统的有效模拟。而由于采样周期过大产生的半闭环控制则是要尽量避免的。 提出了柔性切换系统稳定性监控的概念，实现系统稳定性的监督与控制。稳定性监控器存在两个功能：首先是监督，即监测与判断系统的稳定或不稳定状态与趋势；其次是控制，使系统保持或恢复稳定状态，从而保证系统的全局有界稳定，稳定性监控器是控制器的控制器。 焦炉是具有大时滞、强非线性、多变量耦合、变参数的复杂对象，直行温度受多种因素的影响，焦炉生产过程既受连续时间信号驱动，又受离散事件信号驱动，采用常规的控制方法难以将直行温度控制到要求的精度范围内。本文结合焦炉加热控制这样一个复杂的工业过程控制系统，首次分析了焦炉生产过程中的混杂特性，并将智能切换系统理论应用到焦炉加热自动控制中。控制系统能够适应焦炉运行过程中工况的频繁变化，保证了直行温度的稳定性和合理的燃烧状况，提高了焦炭质量，同时对降低能耗、减少环境污染和延长炉体使用寿命都具有重要意义。更多还原

【Abstract】 In this dissertation, the theories of modeling and control for switching systems was researched, that switching systems is an important kind of hybrid systems. Switching systems is used in the coke oven heating automatic control system, and it realizes the control for complex industry process system.Intelligent switching control (ISC) was proposed in this dissertation, which combined intelligent control theory and switching systems (SS). Switching systems is the framework of intelligent switching control. The theories and methods of intelligent control are used in the modeling, analysis and control of ISC. ISC is the advanced level of the SS during its development, and ISC is the one of the important methods that will solve the control of complex system.Dynamic region soft partitioning method was also proposed which can reduce the complexity of analysis and modeling of complex system. Every dynamic sub-region is one aggregation which including continuous variables and discrete events, and their mutual actions. At the same time, every dynamic sub-region is one unit, which has its dynamic characters and developing rules. Region overlapping, region floating and region zooming of the dynamic sub-region was studied, that region overlapping is the exterior character of the sub-region, region floating and region zooming are inner characters of the sub-region. Fuzzy neural networks based on T-S models are used as the tools for parameters identification of dynamic sub-region.Based on dynamic region soft partitioning of complex system, flexible switching control was put forwarded. Flexible switching (FS) is one strategy for complex system control, FS would be including of hard switching and soft switching, FS can be driven by both continuous variables and discrete events, at the same time, FS has the characters of diversity and hiberarchy. Half close-loop control (HCLC) for time-delay and large inertia system with time-variant was proposed. The chief control object of HCLC based on bounding control is to satisfy the bounding stability of the out variables, and the second control object of HCLC is to satisfy zero error of the out variables. HCLC for complex system just like the control process of human controllers. On the other hand, HCLC caused by large sampling period should be avoided.Stability supervision control for flexible switching systems was proposed, that used to supervise and control the stability of the control system. There are two mainly functions of stability supervision controller. The first function is to monitor and judge the stable or unstable states and trends of the system. The second function is the control function, whichwill adjust the rules and parameters of the sub-controllers to remain or resume stable states of the control system, so stability supervision controller is also as the controller’s controller.Coke oven is a complex plant with the characters of large time-delay, strong non-linear, multivariable coupling and changeable parameters. The dynamic process of the coke oven is driven by both continuous variables and discrete events. The mean flue temperature is affected by many reasons and it is difficult to control the temperature to required precision by the normal control methods. In this dissertation, the hybrid systems characters of the coke oven and its operation were discussed, and intelligent switching systems theories were be used in the coke oven heating automatic control for the first time. The control system can be adaptive of the frequently change of the working states during the coke oven producing process, maintain the stability of the mean flue temperature and good combustion status, it also can improve the quality of the coke. At the same time, the control system has great signification to reduce energy consumption, lessen air pollution and prolong the service life of the oven body.更多还原