一种用于暂态稳定评估神经网络的输入特征离散化新方法.pdfVIP

25 卷 15 期 中 国 电 机 工 程 学 报 Vol.25 No.15 Aug. 2005 2005 年8 月 Proceedings of the CSEE 2005 Chin.Soc.for Elec.Eng. 文章编号：0258-8013 (2005) 15-0056-06 中图分类号：TM311 文献标识码：A 学科分类号：47040 用于暂态稳定评估的人工神经网络 输入特征离散化方法 艳，顾雪平，李 军 （华北电力大学电力系统保护与动态安全监控教育部重点实验室，河北省 保定市 071003 ） DISCRETIZATION IN ARTIFICIAL NEURAL NETWORKS USED FOR TRANSIENT STABILITY ASSESSMENT LIU Yan, GU Xue-ping, LI Jun (Key Laboratory of Power System Protection and Dynamic Security Monitoring and Control under Ministry of Education, North China Electric Power University, Baoding 071003, Hebei Province, China) ABSTRACT: As the kernel of data pretreatment, discretization 关键词 电力系统；暂态稳定；神经网络；数据离散化；信 of continuous attributes is investigated for transient stability 息熵；粗糙集理论 assessment based on artificial neural networks. A new 1 引言 discretization scheme combined entropy with rough sets theory is proposed. The possible cut points for each conditional 电力系统暂态稳定评估（TSA）是关系到电力 [1-4] attribute are found out through clustering analysis of the input 系统安全稳定运行的重要问题 。随着人工智能和 space. The initial cut point sets for each conditional attribute 模式识别技术在电力系统应用领域的拓展，自1989 based on the entropy are established. The optimal combinations 年以来，基于人工神经网络（ANN ）的暂态稳定评 [5-10] of initial cut point sets for different conditional attributes can be 估（ANNTSA ）已获得各国学者的广泛关注 。 o