【作者】 黎恒烜；

【导师】 文劲宇； 孙海顺；

【作者基本信息】 华中科技大学 ， 电力系统及其自动化， 2012， 博士

【摘要】 随着我国经济的飞速发展和工业化进程的不断加快,除了对传统互联电网的电力供应提出更高的要求外,对微电网(Microgrid)的需求也与日俱增。微电网包括各种小型独立电网和接入分布式电源(Distributed Generation, DG)的配电网。微电网与传统互联大电网在网络结构和供电方式上均存在较大的差异,因此,在研究故障重构问题时所考虑的因素和关注的侧重点也不尽相同。本文以舰船电力系统(Shipboard Power System, SPS)和接入分布式电源配电网为例,进行故障重构研究。论文在回顾舰船电力系统和含分布式电源配电网故障重构研究现状的基础上,对目前已有的研究方法和成果进行了概述和总结,归纳了目前各类方法中所存在的一些不足和局限性。针对舰船电力系统故障重构问题分别采用集中式和分布式等多种方法进行对比研究,并将分布式的多代理系统(Multi-Agent System, MAS)技术推广应用于含分布式电源配电网的故障重构研究。全文由以下几个部分组成：首先,建立了舰船电力系统故障重构的优化目标函数和约束条件体系,并采用二进制粒子群优化算法(Binary Particle Swarm Optimization, BPSO)进行优化求解。通过无向图描述舰船电力系统的拓扑结构,采用邻接矩阵存储网络拓扑信息。在进行优化求解时,直接对负载供电状态和少数关键电缆通断状态进行编码,有效地降低编码长度。采用深度优先搜索算法在系统发生故障后对舰船电力系统进行连通性分析,以确定粒子编码的取值范围,很大程度上缩小算法的搜索空间,提高收敛速度。其次,将分支定界法(Branch and Bound Method)应用于舰船电力系统的故障重构研究,建立舰船电力系统的故障重构的数学规划模型。在总结前人研究工作的基础上,对文献中已有的数学模型进行适当改进。结合实际情况,将舰船电力系统的主网络列入重构范围,可以提供更为灵活的供电结构,提高系统供电可靠性。仿真结果表明采用数学规划进行舰船电力系统的故障重构,能够保证解的稳定性和可靠性。再次,介绍了代理、多代理系统技术的发展历史以及相关的基本概念、体系结构等,综述了多代理系统技术在电力系统应用的研究现状和发展趋势。介绍了多代理系统开发软件平台JADE(Java Agent Development Framework),简述了基于JADE平台的多代理系统程序设计的一般过程。第四部分,提出一种分布式的MAS模型来实现舰船电力系统的故障重构,在保障高优先级负载供电的前提下尽可能地恢复系统供电。论文所提出的MAS模型中各代理都具备一定的推理能力,能够根据不同的环境做出合理的决策。采用Java程序设计语言和JADE平台设计MAS程序。最后通过典型的故障算例验证所提出MAS模型以及重构策略的可行性和有效性。最后,提出一种由母线代理(Bus Agent, BA)组成的完全分布式MAS,实现含分布式电源配电网的故障重构功能。论文所提出的MAS中,每一个BA都具备独立引导故障重构的能力,BA相互之间进行通讯、协作共同完成故障重构任务。为保障BA间通讯的有效进行,制定了深度优先通讯机制。该MAS模型能处理单一故障和极端情况下的连锁故障。采用Java程序设计语言和JADE平台设计MAS。通过典型的故障算例验证了该MAS的有效性。更多还原

【Abstract】 With the rapid development of economy and ceaseless acceleration in industrialized process, the request for microgrid (MG) is increasing rapidly apart from the higher requirements for the power supplied by the conventional interconnected power grid. MG consists of various kinds of small stand-alone power systems and the distribution networks with distributed generation (DG) integration. There is significant difference of structures and power supply modes between microgrid and interconnected power system. Therefore, when fault restoration problems are studied, the emphases in the two types systems are different. This thesis takes shipboard power system (SPS) and the distribution networks with DGs for instance to study the fault restoration problem.The current research methods and achieved results in the field of fault restoration in shipboard power system and distribution network with DGs are summarized at first. Then the insufficiency and the flaw of the existing methods are pointed out. Compared various methods for the fault restoration rearch of SPS, this thesis applied the multi-agent system (MAS) technology into the restoration problem of distribution network with DGs. The structure of the whole thesis is listed as follows:Firstly, the fault restoration optimization model of independent shipboard power system is proposed, and the Binary System Particle Swarm Optimization (BPSO) is used for the optimization problem solution. Undirected graph is used to describe the network topology of the shipboard power system, and the adjacency matrix is used to store the network topology information. The encoding method of this algorithm is based on the power state of all loads, the opening and closing state of important cables, which can reduce the encoding length effectively. Based on depth first search after fault, the network connectivity is analyzed in order to confirm the range of particle code value, which can shorten the searching space and improve the convergence speed of the algorithm.Secondly, the mathematic programming model of independent shipboard power system is proposed and the Branch and Bound Method is used for the optimization problem solution. Mathematical programming method proposed in the early literature has been improved in this thesis. According to the actual situation, considering the reconfiguration of major network can make the network structure more flexible, and increase reliability of the SPS. The simulation results indicate that using mathematical programming to solve the fault restoration problem can obtain stable optimal solution.Thirdly, the basic concepts, architectures, and development history of distributed artificial intelligence, agent, and multi-agent system technology are introduced. The current situation and the developing trend of the application of MAS technology in the electric power system are summarized. The Java Agent development framework (JADE) and its programming process are introduced in detail.Fourthly, a prototype of MAS is proposed to reconfigure the SPS in order to maximize the number of served loads with highest priority. Agents of the MAS have the ability of illation, and makes reasonable decisions according to different environments. The MAS is implemented using Java programming language and JADE. The simulation results of typical SPS fault restoration cases demonstrate the feasibility and the advantages of the proposed MAS.Finally, a fully decentralized MAS has been proposed for load restoration of power distribution network with DGs. The proposed multi-agent system consists of bus agents (BAs). Every BA in the MAS has the ability to lead the restoration work independently. The restoration work could be completed by the communication and cooperation among the BAs. In order to ensure smooth communication, a depth-first communication mechanism (DFCM), similar to the process of depth-first search (DFS), is proposed. The MAS is capable of dealing with a single fault or cascading failures. The proposed multi-agent system is implemented based on JADE. The simulation results of typical fault cases verifiy the feasibility of the proposed MAS.更多还原