摘要
以基于SCADA信息快速判别电网稳定薄弱环节为目标,提出一种新的启发式暂态稳定评估策略。应用留数分析和递归切割技术在可观测量构成的特征空间发现稳定模式,并以模式发现结果为样本,提出基于决策树的产生式稳定判别规则提取方法。通过在新英格兰10机系统的应用和测试表明,所获得的规则具有良好的稳定分类能力和适应性,可从潮流信息快速判别稳定水平偏低的扰动位置。从规则自身还可获得有关在线监测重点和电网稳定增强措施的参考信息。
A novel heuristic stability assessment scheme was proposed based on the objective of identifying power system stability weak-point according to SCADA data. Residual analysis and recursive partitioning algorithm were applied to discover the significant patterns in the operational data that compose feature space. With patterns as samples, a decision tree based rule extraction method was proposed to generate production rules for stability classification. According to application and tests of the proposed scheme in the New England 10-machine power system, the rules obtained show good classification and generalization ability and through fast rule reasoning, fault locations with low stability level can be identified. The obtained rules also provide reference for both the selection of monitoring emphases and stability enhancement strategies.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2007年第19期25-31,共7页
Proceedings of the CSEE
基金
国家自然科学基金项目(50407014)。~~
关键词
稳定评估
模式发现
决策树
规则提取
stability assessment
pattern discovery decision tree
rule extraction
