Rule Extraction in Transient Stability Study Using Linear Decision Trees  

作　　者：SUN Hongbin WANG Kang ZHANG Boming ZHAO Feng 

机构地区：[1]State Key Lab of Control and Simulation of Power Systems and Generation Equipments （Dept. of Electrical Engineering, Tsinghua University）, Haidian District, Beijing 100084, China

出　　处：《中国电机工程学报》2011年第34期I0006-I0006,8,共1页Proceedings of the CSEE

基　　金：国家杰出青年科学基金（51025725）; 国家自然科学基金项目（50877043）

摘　　要：为了提高精细规则的性能,提出一种基于支持样本的线性决策树的规则提取方法。该方法筛选临近稳定边界的支持样本,作为决策树的输入样本,精简了样本数目;提出基于线性分类器的决策树方法,以得到基于组合属性的安全稳定运行规则。在单机无穷大系统和IEEE 39节点系统中的对比研究表明：由于考虑了支持样本的特殊性,用线性组合规则代替单属性规则,减少了计算时间,提高了泛化能力,丰富了规则的物理含义,得到的灵敏度信息可用于辅助决策,在安全稳定精细规则提取中具有应用潜力。Traditional operation rules depend on human experience, which are relatively fixed and difficult to fulfill the new demand of the modern power grid. In order to formulate suitable and quickly refreshed operation rules, a method of linear decision tree based on support samples is proposed for rule extraction in this paper. The operation rules extracted by this method have advantages of refinement and intelligence, which helps the dispatching center meet the requirement of smart grid construction.

关 键 词：英文摘要 内容介绍 编辑工作 期刊 

分 类 号：G255.2[文化科学—图书馆学]