智能配电网风险综合评估研究

发布时间：2018-01-19 07:17

本文关键词： 智能配电网风险评估最优变权风险值多时段　出处：《山东大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来,针对智能电网的研究和建设成为世界各国的关注焦点,随着电网智能化进程的快速推进,智能电网的发展在取得显著成效的同时,也面临着极大的挑战。由于智能配电网直接面向电力用户,且积极支持分布式电源等新能源电力设施的并入,这不但加深了配电网调度与运行管理的复杂程度,而且给配电网的施工和运行维护带来了更多的限制因素,因此导致配电网面临的风险类型更为繁杂、受到的风险影响更为突出。为了可靠地保证智能配电网的正常运营,须实时掌握其面临的风险类型及严重程度,故应对网络的风险状态进行实时评估。首先,本文在剖析智能配电网特征的基础上,选取目前影响网络正常运行的关键风险因素,按照风险评估的基本准则和一般流程,建立多层级网络风险评估体系;根据实际工程的需要,将网络风险状态划分为5个等级,并量化为具体的风险值区间,为直观地评判网络的运营情况提供了极大的便利。然后,本文提出了一套针对网络风险评估体系的最优变权分配方案。考虑到传统的主观赋权方法在分配权重时常常忽略专家主观意识对处理过程的影响,本文基于相似度聚类分析定义了专家权威关联系数这一新概念,将其与基于模糊层次分析法确定的权重相融合,以使确定的主观权重更为有效;为了充分计及评估指标参数的时序特征,本文利用熵值改进传统CRITIC赋权法,根据各微观指标在网络发展的历史阶段、当前阶段和未来阶段的评估值,确定指标的客观权重;将两类权重系数整合为指标的最优变权,将传统的静态评估转化为结合趋势分析的动态评估,为全面掌握网络的运行状况提供了更为可靠的理论依据。其次,本文具体阐述了各微观指标风险值的计算过程。针对定性评估指标,本文基于集值统计专家估价法对其进行合理的评判,考虑到评估过程中存在各种模糊性、随机性和专家心理波动等不确定因素,引入专家评估信任因子度量评判结果的信赖度;针对定量评估指标,本文建立了岭形模糊隶属度函数作为其评估函数,并通过算例证明了本文构建的评估函数相比于传统评估函数在性能上具备明显的优越性。最后,本文以某10KV智能配电网示范工程作为研究对象,详细地阐述了网络整体风险值的求取过程。通过将计算结果和网络实际运行状态作对比可以证明本文提出的智能配电网风险评估方法能够有效地识别网络发展过程中面临的主要风险类型,并可对网络面临的风险严重程度作出较为准确的量化,对于全面推进智能配电网的趋优发展具有重要的指导意义。
[Abstract]:In recent years, the research and construction of smart grid has become the focus of attention all over the world. With the rapid progress of intelligent power grid, the development of smart grid has achieved remarkable results at the same time. Because smart distribution network is directly oriented to power users and actively supports the integration of new energy power facilities such as distributed generation, this not only deepens the complexity of distribution network dispatching and operation management. Moreover, it brings more restrictive factors to the construction and operation and maintenance of distribution network, which leads to more complicated risk types faced by distribution network. In order to ensure the normal operation of smart distribution network, it is necessary to grasp the risk types and severity in real time, so the risk state of the network should be evaluated in real time. On the basis of analyzing the characteristics of smart distribution network, this paper selects the key risk factors that affect the normal operation of the network, and establishes a multi-level network risk assessment system according to the basic criteria and general process of risk assessment. According to the need of the actual project, the network risk state is divided into five grades, and quantified into the specific risk value interval, which provides a great convenience for the intuitive evaluation of the network operation. Then. In this paper, an optimal variable weight allocation scheme for network risk assessment system is proposed, considering that the traditional subjective weighting method often ignores the influence of the subjective consciousness of experts on the processing process. In this paper, a new concept of expert authoritative correlation coefficient is defined based on similarity clustering analysis, which is combined with the weight determined by fuzzy analytic hierarchy process (FAHP) to make the subjective weight more effective. In order to fully take into account the time series characteristics of the evaluation index parameters, this paper uses entropy to improve the traditional CRITIC weighting method, according to the microcosmic indicators in the historical stage of network development, the current stage and the future stage of the evaluation value. To determine the objective weight of the index; Integrating the two kinds of weight coefficients as the optimal variable weight of the index and transforming the traditional static evaluation into the dynamic evaluation combined with trend analysis provides a more reliable theoretical basis for the overall operation of the network. Secondly. In this paper, the calculation process of the risk value of each micro index is elaborated. Aiming at the qualitative evaluation index, this paper makes a reasonable evaluation on it based on the set value statistical expert valuation method, considering the various fuzziness in the process of evaluation. The uncertainty factors such as randomness and psychological fluctuation of experts are introduced to evaluate the reliability of the evaluation results of the trust factor. For the quantitative evaluation index, this paper establishes the ridge fuzzy membership function as its evaluation function. An example shows that the evaluation function constructed in this paper has obvious superiority compared with the traditional evaluation function. Finally, this paper takes a 10KV smart distribution network demonstration project as the research object. The calculation process of the whole network risk value is expounded in detail. By comparing the calculation result with the actual network running state, it can be proved that the intelligent distribution network risk assessment method proposed in this paper can effectively identify the developed network. The main types of risk in the process. Furthermore, it can accurately quantify the severity of the risks faced by the network, which has an important guiding significance for the overall development of intelligent distribution network.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM76

【参考文献】