基于线性判别分析与主成分分析的骨干网架生存性评估方法  

Survivability Evaluation of Backbone Network Based on Linear Discriminant Analysis and Principal Component Analysis

作　　者：赵一婕[1] 刘涤尘[1] 吴军[1] 董飞飞[1] 宋春丽[1] 潘旭东[1] 王浩磊[1] 

机构地区：[1]武汉大学电气工程学院,湖北省武汉市430072

出　　处：《电网技术》2014年第2期388-394,共7页Power System Technology

基　　金：国家863高技术基金项目(2011AA05A119);国家自然科学基金项目(51207114);国家电网公司大电网重大专项资助项目课题(SGCC-MPLG029-2012)~~

摘　　要：从电网网架的可抵抗性、可恢复性、安全性和连通性4方面出发,建立了差异化规划生存性评估指标体系,并进行指标效益成本属性分析和规范化处理;在此基础上,建立二级指标的线性判别分析评估模型和一级指标的主成分分析评估模型。首先求取二级指标决策矩阵的线性判别分析判决准则函数的最大值,以其所对应的最大投影方向确定三级指标权重,再对二级指标矩阵进行主成分分析,根据累积方差贡献率来确定二级指标权重,最终实现骨干网架综合生存性评估。IEEE118节点算例的仿真结果验证了该评估方法的有效性和可信性。From the perspective of resistibility, recoverability, security and connectivity of backbone network an indicator system for survivability evaluation of differentiated planning is established, and the analysis on benefit and cost attributes of survivability indicators and the standardization of them are performed. On this basis, a linear discriminant analysis （LDA） model for the second-class indicators and a principal component analysis （PCA） model for the first-class indicator are built. Firstly, the maximum value of LDA judgment criterion function for the second-class decision matrix is solved, and according to the largest projection direction corresponding to this maximum value the weights of the third-class indicators are determined, then the PCA of the second-class indicator matrix is performed, and then according to the contribution rates of cumulative variances the weights of the second-class indicators are determined, and finally the synthetic survivability evaluation of backbone network is implemented. The validity and creditability of the proposed evaluation method are verified by simulation results of IEEE 1 ! 8-bus system.

关 键 词：差异化规划 骨干网架 生存性 线性判别分析 主成分分析 

分 类 号：TM72[电气工程—电力系统及自动化]