一种基于KMeans与Random Forest的异常温升捕捉方法  

A Method for Capturing Abnormal Temperature Rise Based on KMeans and Random Forest

作　　者：汪海良[1] WANG Hailiang(Shanghai Institute of Architectural Design&Research Co.,Ltd.,Shanghai 200041,China)

机构地区：[1]上海建筑设计研究院有限公司,上海200041

出　　处：《现代建筑电气》2024年第6期21-26,49,共7页Modern Architecture Electric

摘　　要：针对线路老化、线路过载的火灾频发问题,分析了线路老化、线路过载与异常温升之间的关联性,以电流值、线缆温度作为输入,利用KMeans聚类算法划分可能存在异常温升的区间,通过Random Forest算法识别线路过载问题,可以提前通知用户整改线路,预防火灾的发生。In response to the frequent occurrence of fire caused by line aging and overload,this article analyzes the correlation between line aging,line overload,and abnormal temperature rise.Using current values and cable temperature as inputs,KMeans clustering algorithm is used to divide the possible intervals of abnormal temperature rise.Random Forest algorithm is used to identify line overload problems,and users can be notified in advance to rectify the line and prevent the occurrence of fire.

关 键 词：线路过载 异常温升 Random Forest KMeans 

分 类 号：TU855[建筑科学]