电力系统可靠性评估逆问题的理论与方法研究

Component reliability parameters are the foundation and key part in power system reliability evaluation. It may lead to wrong evaluation conclusions if they are incorrect or missing, making the evaluation result meaningless. Reliability parameters are mainly obtained from the statistical analysis of historical blackout records. Due to the influences of complex and dynamic statistical work and human factor, data errors and omissions are inevitable, which currently have become problematic issues affecting the engineering application of reliability evaluation..Starting with the inverse process of reliability evaluation, this project explores the method to obtain/verify component reliability parameters based on known reliability indices. It first proposes the concept of inverse problem of power system reliability evaluation. Based on scenario analysis, this project combines series-parallel equivalent, network equivalent and other technologies to establish analytical model of equivalent inverse problem. Based on the monotonic variation rule of reliability parameters and reliability indices, it proposes the method to estimate, compress and segment the feasible interval of inverse problem’s initial solution. Based on reliability tracing and sensitivity analysis, it discretizes the unknown parameters and the inverse problem is thus converted to equivalent combination optimization model for solution..The research project is expected to establish a complete theoretical framework for the inverse problem of power system reliability evaluation, and to provide new analysis methods and quantitative decision-making basis for planning, operation, reliability management, equipment transformation and bidding.

元件可靠性参数是电力系统可靠性评估的基础和关键，一旦有误或缺失，就可能导致错误的评估结论，使得评估结果毫无意义。可靠性参数常通过历史停电记录的统计分析得到。但是，由于统计工作的复杂性、动态性及人为因素的影响，数据错误和遗漏不可避免。当前，其已成为影响可靠性评估工程应用的棘手问题。.本项目从可靠性评估的逆向过程入手，研究如何从已知可靠性指标出发求取/校核元件可靠性参数。首次提出电力系统可靠性评估逆问题的概念；基于场景分析，结合串并联等效、网络等值等技术，建立等值的逆问题解析模型；基于可靠性参数与可靠性指标间的单调变化规律，提出逆问题初始解可行区间的估计和压缩分割方法；基于可靠性跟踪或灵敏度分析离散化待求参数，将逆问题转化为等值的组合优化模型求解。.通过项目研究，希冀建立电力系统可靠性评估逆问题的完整理论框架体系，为规划、运行、可靠性管理及设备改造和招投标等提供新的分析方法和量化决策依据。

由于统计的复杂性、动态性及人为因素等影响，元件可靠性参数数据统计错误不可避免，已成为影响可靠性评估工程应用的棘手问题。幸运的是，随着PMU和智能电表等大量安装，从用户角度统计的系统/节点可靠性指标准确度较高，如果能从已知的可靠性指标入手校验或求取元件可靠性参数，则不失为一种好的方法，这正是本项目要研究的可靠性评估逆问题。本项目围绕可靠性评估逆问题的定义、建模及工程应用等开展研究，主要内容及成果如下：.基于状态枚举法，推导了可靠性指标关于元件可靠性参数的解析计算模型；基于全概率公式等，提出了基于时序Monte Carlo模拟的可靠性指标解析计算模型。解析模型一旦建立，当元件可靠性参数发生变化后，可以快速解析求取系统可靠性指标。. 结合工程实际中存在的可靠性“逆向”评估需求，首次提出可靠性评估逆问题的概念，基于前述解析模型构造可靠性评估逆问题的解析非线性方程组模型，提出基于区间对分和Krawczyk-Hansen算子的精确数值解法。所提模型和求解方法无论是针对单个待求故障率、修复率，还是二者混合组成的多个元件待求参数，均可将解压缩在一个较小的取值区间内。基于所提可靠性评估逆问题理论，进行了两方面的工程应用：. 1）错误元件可靠性参数辨识。首先，根据灵敏度将所有元件参数聚类，基于逆问题模型求取敏感参数的初步估值；然后，提出非线性参数估计问题的迭代算法求取所有参数的估值，辨识出部分错误参数；最后采用区间检验对参数进行二次估计，辨识出其余错误参数并校正。该研究从元件参数和系统指标适应性入手，提出了一种新的错误元件可靠性参数辨识方法。. 2）面向可靠性提升的系统加固规划。基于可靠性评估逆问题建模过程中推导的可靠性指标解析模型，结合传统系统加固规划模型，构造面向可靠性提升的系统加固规划全解析模型。该全解析模型从一定程度上克服了传统基于“正向”可靠性评估进行系统加固规划的海量计算问题。