【作者】 何志勤；

【导师】 张哲； 尹项根；

【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】 继电保护是保障电网安全的第一道防线,当其快速、可靠动作时,可有效遏制电网状态恶化；反之,则可能扩大事故范围,甚至引发全网大面积停电事故。传统后备保护利用被保护元件单侧信息识别故障,存在整定计算繁琐、对电网复杂运行方式适应性差、在大范围潮流转移时可能连锁动作等突出问题。因此,加强继电保护新原理与新技术的研究,构建性能优良、适应复杂电网运行环境的继电保护系统,对保障电力系统运行安全,防范大范围停电事故具有重要的理论和现实意义。近年来,随着广域同步测量与通信技术的快速发展,利用电网多点量测信息识别故障元件,并以此构建基于故障元件识别的新型广域后备保护的研究得到了广泛关注,并已成为智能电网保护技术的重要发展方向。本文在国家自然科学基金重点项目“集中决策与分布实现相协调的大电网后备保护系统研究”(50837002)资助下,针对新型广域后备保护系统的关键技术问题进行了深入分析与研究。实际电网规模庞大、结构复杂,如何将其划分为若干个合理的保护分区,是广域后备保护需解决的关键问题之一。论文以广域通信可靠性、保护系统投资经济性和分区通信量均衡性为指标,并以保护子站-中心站间的通信时延为约束,构建了广域后备保护的分区性能评价体系。同时,提出了一种基于双层聚类的自适应分区算法,以克服分区数目难以预先确定的问题。该算法根据各变电站的关联支路数,将电网初分为多个候选分区,继而根据候选分区数目确定进化群体中的基因链长度。之后,根据各基因链中基因值的变化,将未激活的候选分区逐一划分至相应的分区中,实现对应分区数目下的区域划分操作。仿真结果表明,所述分区优化算法可对不同规模电网进行合理划分,分区结果的性能指标和通信时延均满足应用要求。利用广域信息准确识别故障元件、通过故障区域搜索降低信息交互量是广域后备保护的研究重点和核心,论文从不同角度对该问题进行了分析与研究。根据电网中各节点故障分量电压的分布特性,提出了基于故障电压比较的故障元件识别算法和故障区域搜索算法。其中,故障元件识别算法利用线路一侧的故障电压、电流测量值推算另一侧电压故障分量和补偿电压分量。并利用两者与实测值的比值为动作参数,构建综合比较判据。以保证算法在线路阻抗与两侧系统阻抗相对大小不同时,均能正确识别故障元件。故障区域搜索算法通过检测节点故障电压与支路故障电流的变化幅度确定故障所在区域,进而减少广域通信量。并通过母线序电压排序搜索故障关联线路,减小故障元件识别计算量。仿真结果表明,故障区域搜索算法可通过子站启动检测和故障关联线路搜索,有效降低子站上传信息量和提高故障元件识别效率。故障元件识别算法在多种复杂条件下均能做出正确判断,具有较高的灵敏性与良好的安全性。论文根据电网中各支路故障分量电流的分布特性,提出了基于故障电流分布的故障元件识别算法和故障区域搜索算法。其中,故障元件识别算法根据最大故障电流出现在实际故障线路一侧、故障线路两侧的故障电流一般存在较大差异等特性分别构建识别判据。继而将2种判据综合运用,利用两者间的互补性进一步提高故障元件识别算法的整体性能。故障区域搜索算法通过对子站关联支路进行故障电流比幅,选取准故障疑似线路。之后通过关联矩阵计算,在上传信息的准故障疑似线路中搜索故障疑似线路,进一步缩小故障区域。仿真结果表明,所述故障区域搜索算法可有效降低子站信息传输量和中心站故障元件识别计算量。故障元件识别算法可正确识别各种内部故障,并在非故障扰动发生时可靠不误动,具有良好的安全性。广域后备保护需采集电网内多点量测信息识别故障元件,如何保证其在信息缺失和错误情况下作出正确决策,是广域后备保护工程应用中亟待解决的技术难题。论文利用D-S证据理论,提出了基于多源信息融合的智能决策模型与算法。其中,决策算法利用Manhattan距离衡量证据多属性间的冲突程度,进而以证据的可依赖度为权重对证据源与合成规则进行修正,克服了经典合成规则在高冲突证据融合时易产生悖论的问题。同时,以广域后备保护故障识别元件、就地保护元件的动作信息为证据,建立了广域后备保护智能决策模型。利用支路故障电流选线对距离Ⅱ、Ⅲ段元件的信任度进行辅助计算。并将决策结果的信任区间细分为非故障、不确定与故障等三部分,进一步防止算法在证据间冲突过大时出现误判。仿真结果表明,所述智能决策算法在多位证据错误和缺失时具有良好的防拒动与防误动性能。论文最后对所述研究成果进行了总结,并对后续研究工作进行了展望。更多还原

【Abstract】 Protective relaying is the first defense line of electric power grid. When it operates fast and reliably, the safety of power grid can be guaranteed. Otherwise, grid instability and blackout event may occur. Traditional backup protection identifies faults based on local information. It has complex setting principle, poor adaptability to grid operation change and is inclined to misoperate under heavily loaded flow transfer. Consequently, it is meaningful to exploit novel protection technology and design new-style protection system which has excellent property and is adaptive to the grid operation change.Accompany with the development of wide-area measurement and communication technique, applying wide-area information to identify fault element and studying novel wide-area backup protection (WABP) based on fault element identification (FEI) becomes a hot spot in modern power system research. Supported by the National Natural Science Foundation of China (No.50837002), this dissertation researches the kernel problems of wide-area backup protection based on fault element identification in depth.How to partition a large scale power grid into several reasonable subsystems is one of the critical issues of WABP research. There are three indexes including communication reliability of substation, investment economy of protection system and communication balance among subsystems have been adopted to establish a integrated hierarchy so as to evaluate the property of partition result. Meanwhile, a two-layer clustering partition algorithm is proposed. Firstly, the protected grid is decomposed into multiple candidate subsystems according to the compactness among substations, and then the length of chromosome can be determined. During two-layer clustering process, the number of subsystems is optimized when genes within chromosomes are changed, and the candidate subsystems can be further merged. Simulation result demonstrates that the presented algorithm could partition practical large grid into several reasonable subsystems, and the application requirements and latency constraints are satisfied.A novel FEI algorithm and a fault area search algorithm based on fault component voltage comparison are proposed, respectively. The fault component voltage and current at one terminal of line are applied to calculate the fault component voltage at the other terminal. Afterwards, a comprehensive identification criterion based on the ratio between measured and calculated values is established to identify fault element. On the other hand, fault area search algorithm could decrease wide-area communication traffic through substation pickup detection, and then accelerate FEI through fault incident line search. Simulation test proves the validity of presented algorithms under various complex conditions.The distribution characteristic of fault component currents is illustrated, and corresponding FEI algorithm and fault area search algorithm are also proposed. The FEI algorithm comprises branch fault component current comparison criterion and two-terminal fault component current comparison criterion. Then, these two criteria are synthetically applied to identify various internal faults. The fault area search algorithm searches quasi suspected faulty lines through branch current magnitude comparison. After that, the suspected faulty lines are searched through incident matrix computation. Simulation result shows that the uploaded information of substations and the calculation quantity of central station can be dramatically reduced through fault area search. Moreover, the FEI algorithm could identify different types of faults and has good security.How to safeguard the reliability of protection decision result under information error and loss is a formidable challenge. An intelligent decision-making model and a decision-making algorithm based on multi-source information fusion are proposed, respectively. The Manhattan distance is applied to describe conflict degree, and a weighted evidence method is used to decrease the conflict degree among decision evidences. Finally, source evidence and traditional composition rule are all improved. Besides that, the operation information from WABP and local relays are chosen as decision evidences. The fault currents of branches are applied to calculate the trust degree of Zone-2 and-3 elements. Furthermore, the trust interval is partitioned into three parts so as to further improve the performance of decision-making algorithm. Simulation result shows that the proposed algorithm could make correct decision under multiple information error or loss.更多还原