基于主成分分析的多变量决策树在接线方式识别建模工程化实用化应用探讨

Based on principal component analysis of multivariate decision tree in the connection mode identification modeling engineering practical application

本文提出一种用于电网接线方式识别的工程化实用化技术,该技术基于主成分分析的多变量决策树,利用机器监管学习的决策树原理,建立电力系统接线方式智能识别软件结构。采用决策树的思想训练接线方式特征模型样本,生成分类器,实现了未知情况下的接线方式识别,且能够不断进行接线方式特征模型学习,对新出现接线方式进行精准的分类识别。

This paper presents an engineering practical for grid connection mode recognition technology, the technology is based on principal component analysis of multivariate decision tree,using supervised learning machine principle of decision tree based power system connection mode structure of intelligent recognition software. Connection mode characteristic model by adopting the idea of decision tree training samples,the generated classifier,realized the connection mode identification of unknown circumstances, and able to ongoing connection mode feature model to study,to identify the precise classification of emerging connection mode.