多智能体系统协同控制方法及在分布式卫星应用研究

发布时间：2018-07-08 11:44

本文选题：多智能体系统 + 协同控制　；参考：《哈尔滨工业大学》2017年博士论文

【摘要】：多智能体系统是指由多个智能体单元构成的网络系统,如分布式卫星系统、无人机编队系统等,各智能体之间通过相对信息交互及协同工作来完成某些特定的任务。相比于传统的单智能体系统,多智能体系统具备容错能力强、工作效率高、成本低等优势。多智能体系统协同控制问题是当今控制领域的热点问题,也是多智能体系统的研究中最为重要的问题,因此有必要对其进行深入研究。本学位论文旨在对多智能体系统协同控制问题进行深入研究,并将相应的理论方法应用于分布式卫星相对轨道转移及姿态协同控制问题中。论文的主要研究内容如下:针对线性多智能体系统降阶问题,提出改进的线性矩阵不等式(Linear Matrix Inequality,LMI)高效降阶方法。在求解线性多智能体系统协同控制时,智能体的个数会直接影响LMI的阶数,大规模多智能体系统的LMI阶数将会非常高,从而带来巨大的计算花费。本文提出一种用于降低LMI阶数的改进降阶方法,相比较现有的降阶方法,计算时间大幅度降低。以分布式卫星相对轨道转移问题为仿真应用背景,在外干扰存在的情况,设计一种能耗约束下的H∞协同控制律,利用改进的LMI降阶方法,降阶效果明显,验证了提出方法的有效性。针对多智能体系统中存在的时滞问题,研究非一致时滞约束下的协同控制方法。各智能体在相互通讯的过程中,由于通讯距离较远且环境中存在干扰,将会导致智能体之间存在通讯时滞,且不同智能体之间的时滞大小不同,此时称多智能体系统中存在的时滞是非一致的(non-uniform)。本文将设计非一致时滞约束下的积分型与基于低通滤波器的动态滑模控制律,并以分布式卫星相对轨道转移问题为仿真应用背景,验证设计出的控制律的有效性以及相比于传统H∞控制律存在的优势。针对模糊理论在多智能体系统中的应用问题,提出基于T-S模糊理论的多智能体系统协同控制方法。在利用传统的非线性反馈方法求解非线性多智能体系统协同控制问题时,过于依赖系统的模型,若系统中含有不确定参数,将会对结果产生很大的影响。此外,对于复杂的非线性系统,非线性反馈控制方法将会在很大程度上增加执行机构的负担。本文将非线性多智能体系统构建为模糊多智能体系统,设计出一种分布式模糊控制律,从而将非线性系统线性化处理,并提出一种针对模糊多智能体系统的一般化稳定性分析方法。以分布式卫星姿态协同控制问题为仿真应用背景,验证了本文提出方法的有效性以及相比于传统非线性反馈方法存在的优势。针对非线性多智能体系统滑模控制问题,提出一种能够有效抑制抖振的模糊滑模控制方法。在应用任务中,外界环境干扰将会对协同控制系统性能产生影响,滑模控制是有效抑制干扰的重要方法之一。本文提出一种新的模糊模型转换方法,将非线性多智能体系统构建为模糊系统,针对构建出的模糊系统,分别设计出积分型与动态滑模控制器。以分布式卫星姿态协同跟踪为仿真应用背景,验证了本文提出方法的有效性。
[Abstract]:Multi agent system is a network system composed of multiple agent units, such as distributed satellite system, unmanned aerial vehicle formation system and so on. Each agent completes certain specific tasks through relative information interaction and cooperative work. Compared to the traditional single agent system, the multi-agent system has the ability of fault-tolerant and working efficiency. The problem of multi agent system cooperative control is a hot issue in the field of control, and it is also the most important problem in the research of multi agent system. Therefore, it is necessary to study it deeply. The method should be used in the problem of relative orbit transfer and attitude coordination control of distributed satellites. The main research contents of this paper are as follows: in order to reduce the order of linear multi-agent systems, an improved linear matrix inequality (Linear Matrix Inequality, LMI) efficient reduction method is proposed. The number of a number will directly affect the order of LMI, and the number of LMI in the large-scale multi-agent system will be very high, which brings huge calculation cost. In this paper, an improved reduced order method for reducing the number of LMI is proposed, which is compared with the existing order reduction method and the calculation time is greatly reduced. The problem of the distributed satellite relative orbit transfer is simulated. The H infinity cooperative control law under the constraints of energy consumption is designed with the background and the presence of external interference. The effectiveness of the proposed method is verified by using the improved LMI order reduction method. The effectiveness of the proposed method is verified. The cooperative control method under the non uniform delay constraint is studied for the time delay problem in the multi-agent system. In the process of communication, due to the distant communication distance and the existence of interference in the environment, the delay between the agents will be caused, and the time delay between the different agents is different. At this time, the delay in the multi agent system is not consistent (non-uniform). This paper will establish the integral type and the low pass based on the non uniform delay constraint. The dynamic sliding mode control law of the filter is used to verify the effectiveness of the designed control law and the advantages of the traditional H infinity control law. Based on the application of fuzzy theory in multi-agent system, a multi-agent system based on T-S fuzzy theory is proposed. The same control method. When using the traditional nonlinear feedback method to solve the cooperative control problem of nonlinear multi-agent system, it is too dependent on the model of the system. If the system contains uncertain parameters, it will have a great influence on the result. In addition, for the complex nonlinear system, the nonlinear feedback control method will be to a great extent. In this paper, the nonlinear multi-agent system is constructed as a fuzzy multi agent system, and a distributed fuzzy control law is designed to linearize the nonlinear system, and a general stability analysis method for fuzzy multi-agent system is proposed. The distributed satellite attitude coordination control problem is presented. For the simulation application background, the validity of the proposed method and the advantages of the traditional nonlinear feedback method are verified. A fuzzy sliding mode control method which can effectively suppress the chattering is proposed for the sliding mode control problem of nonlinear multi-agent system. In the application task, the external environment interference will be on the cooperative control system. The sliding mode control is one of the most important methods to effectively suppress interference. In this paper, a new method of fuzzy model conversion is proposed. The nonlinear multi-agent system is constructed into a fuzzy system. An integral and dynamic sliding mode controller is designed for the fuzzy system constructed. The real application background verifies the effectiveness of the proposed method.
【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：V448.2

【相似文献】