基于免疫模糊聚类算法的电网抗差状态估计  

Robust state estimation against error based on immune fuzzy clustering algorithm

作　　者：杨丽君[1] 卢志刚[1] 李爽[1] 陈毅强 

机构地区：[1]燕山大学电气工程学院,河北秦皇岛066004 [2]秦皇岛视听机械研究所,河北秦皇岛066000

出　　处：《燕山大学学报》2009年第4期347-351,共5页Journal of Yanshan University

基　　金：河北省自然科学基金资助项目(07M007)

摘　　要：提出了一种基于免疫进化模糊聚类算法的电网抗差状态估计方法。该方法首先计算出量测数据的标准残差和相邻采样时刻量测值之差,初步将量测数据划分为可疑数据和可靠数据。分别在0.5～1和0～1之间随机生成可靠数据和可疑数据的隶属度,形成初始分类矩阵,克服了整个分类矩阵在0～1之间随机生成的缺陷,并大大加快了算法的收敛速度。之后以和为特征值,应用免疫进化算法对分类矩阵进行模糊聚类,以获得各个量测量的良数据隶属度。依据各个量测量的良数据隶属度,进一步将量测数据划分入淘汰区、降权区、保权区进行状态估计。该方法能获得较高的抗差能力和状态估计精度,且数值稳定性较好。对IEEE14节点系统的算例仿真表明了该方法的有效性。A new robust estimation method against error in power system based on immune fuzzy clustering is presented. First, the metrical data can be classified as distrustful data or credible data roughly according to the metrical data＇s normalized residual rN and the difference △z between the sequential sampling metrical data. Then the fuzzy classified matrix is initialized based on the different membership degree to decrease the iteration step： the membership degree of the credible data is random set between 0.5-1, while that of the distrustful data＇s is between 0-1. Second, taking rN and Az as two characteristic values, the classified matrix is fuzzy clustered for the membership degree＇s better value by using artificial immune algorithm time and again. The algorithm then classify the metrical data further into elimination district, reducing right district and whole right district based on the new-found membership degree, and go on with the state estimation process. This method has better accuracy and numeric al stability. It can restrain the degenerating phenomenon during the evolutionary process of searching for the global optimum solution. Simulation resuits on IEEE 14-node system shows the method＇s efficiency.

关 键 词：模糊聚类 免疫算法 状态估计 

分 类 号：TM732[电气工程—电力系统及自动化] TM744