CASIA OpenIR  > 毕业生  > 博士学位论文
基于混合逻辑动态的混杂系统建模和控制方法研究
其他题名Studies on Modeling and Control of Hybrid System Based on Mixed Logic Dynamic
李秀改
学位类型工学博士
导师高东杰
2003-04-01
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业控制理论与控制工程
关键词混杂系统 混合逻辑动态 模型预测控制 状态估计 分支定界法 遗传算法 Hybrid System Mixed Logic Dynamic Model Predictive Control Branch And bAnd Genetic Algorithm
摘要本文的主要研究内容和贡献表现在以下几个方面: (1)分析了混杂系统基于混合逻辑动态的建模方法,得到了混合逻辑动态的 系统化建模机理。研究了命题逻辑转化和混合整数不等式之间的转化关 系,提供了关于逻辑变量化简的一些策略。对混合逻辑动态模型和其它模 型的等价转化关系和相应的转化条件进行了分析。 (2)分析了预测控制的关键技术问题,分别采用调节器和跟踪器型两种性能指 标研究了混杂系统基于混合逻辑动态的预测控制。在分析权重矩阵对系统 性能影响的基础上,提出混杂系统基于MLD模型的变权重预测控制方法。 设计权重系数α、β可使控制目标在系统实时性和稳定性之间进行权衡。 (3)研究了性能指标为线性的模型预测控制,在此基础上提出基于1—1范数 的MLD预测控制,并对线性预测控制器的稳定性进行了分析。通过适当 地选取终端加权矩阵,可保证控制系统的稳定。设计控制算法并对进行了 仿真验证,并对线性规划和二次规划在计算性能方面进行了比较。 (4)在研究了混杂系统基于混合逻辑动态的开环预测控制基础上,进一步考虑 系统存在过程干扰和量测噪声的情况,提出在预测控制算法中加入系统状 态最优估计模块构成基于状态反馈的闭环预测控制策略,并对滚动时域状 态估计器的稳定性进行了证明。 (5)对QXLPC—III过程控制装置的三容系统进行了基于MLD的预测控制研究, 建立了该系统的混合逻辑动态模型,设计了基于跟踪器型的预测控制算 法,并进行了仿真验证。设计了VC++与Matlab混合编程的预测仿真工具, 为算法工程化实施打下了基础。 (6)研究了分支定界法求解混合整数规划的算法和步骤;提出采用改进遗传算 法MGA求解该问题的实施步骤。
其他摘要Main research contributions and achievements of this paper are as following: (1)Principle modeling approach are deeply analyzed and acquired based on Mixed Logic Dynamic (MLD). The transformation methods from prepositional logic to mixed integer inequality are studied and some strategies for reducing the number of integer variables are proposed. The relationship between other modeling approaches for hybrid system and MLD is analyzed which set a basis for making research other characters for hybrid based on the most suitable model frameworks. (2)The key techniques for model predictive control are studied. Predictive control for hybrid system with the adjustable performance and tracking performance based on mixed logic dynamic are researched. By analyzing the effects of weight matrix on controller's performance, a changeable weigh matrix predictive control based on MLD is presented. Through adjusting the parameters of α,β, the desired control result can be achieved by making tradeoff between rapidness and stability. Simulation results show the rationality and validity of the method. (3)Model predictive for MLD based on linear objection is studied. Compared with other linear objection, the 1-1 norm predictive control performance for MLD is presented. The controller's stability is also guaranteed by choosing the suitable terminal weight matrix. Simulation results show the validity of this algorithm. Controller's performance comparison with mixed integer linear programming and mixed integer quadratic programming is also analyzed. (4)Taking into account of the process disturbance and measure noise, closed loop model predictive control for MLD based on Moving Horizon State Estimation (MHE) is proposed. The algorithm provides an effective tool to estimate the system state and disturbance with constraints. Stability for MHE is also studied. (5)Aiming at the three tanks system of the QXLPC-III process control equipment, MLD principle model and predictive control algorithm is designed to keep each tank remain the desired level respectively. Application to the QXLPC-Ⅲ process control equipment is simulated successfully. The simulation tool for predictive control based on mixed programming technique between VC++ and Matlab is presented. (6)One of the Mixed Integer Programming's (MIP) solvers B&B is studied. Algorithm for solving the MIP based on Modified Genetic Algorithm (MGA) is proposed. Compared with B&B, MGA is intelligent global search strategy using the binary coding and float coding to represent each individual which is proved to be effective for the performances' analyzing or optimization and control of the hybrid system with lower real-time demand.
馆藏号XWLW727
其他标识符727
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/5743
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
李秀改. 基于混合逻辑动态的混杂系统建模和控制方法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2003.
条目包含的文件
条目无相关文件。
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[李秀改]的文章
百度学术
百度学术中相似的文章
[李秀改]的文章
必应学术
必应学术中相似的文章
[李秀改]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。