首页|基于改进决策树算法的输电线路风偏跳闸故障识别研究

基于改进决策树算法的输电线路风偏跳闸故障识别研究

扫码查看
当前的输电线路风偏跳闸故障识别节点设定方式一般为独立结构,可识别范围较小,导致误识率增加,因此提出基于改进决策树算法的输电线路风偏跳闸故障识别方法.先提取基础故障特征,采用多阶的方式,扩大可识别范围,部署多目标故障识别节点,然后以此为基础,构建改进决策树测算风偏跳闸故障识别模型,采用随机锁定处理实现故障识别.测试结果表明,对比传统 GRU的特高压三端风偏跳闸故障识别测试组、传统长短期记忆网络和随机矩阵风偏跳闸故障识别测试组,改进决策树算法输电线路风偏跳闸故障识别测试组最终得出的误识率被较好地控制在20%以下,说明在改进决策树算法的辅助下,当前的跳闸故障识别速度更快,误差可控,具有实际的应用价值.
Research on Fault Identification of Transmission Line Windage Tripping Based on Improved Decision Tree Algorithm
Currently prevailing technologies of identifying wind-induced tripping generally adopt independent node structure and are confounded by small recognition range and increased false positive rate.Therefore,this paper proposes a wind-induced tripping fault identification method for transmission lines based on an improved decision tree algorithm.The method first extracts the basic fault features and adopts a multi-level approach to expand the identifiable range,and arran-ges multi-objective fault identification nodes.Second based on the arranged nodes it constructs a modified-decision-tree-based calculation model of wind-induced tripping faults and uses random locking processing to achieve fault identification.The proposed method has been preliminarily proved by test results capable of control the false positive rate below 20%which is superior to those obtained by conventional technologies including GRU-based three-terminal method,long-short-term memory network method,and the random matrix method.This indicated a potential applicative value of modified-de-cision-tree-assisted fault identification manifested by faster identification and well-controlled false positive rate.

modified decision tree algorithmtransmission linewind-induced trippingfault identificationremote con-trolled identification

曹亚华、庄杰

展开 >

国网山东省电力公司超高压公司,山东 济南 250000

改进决策树算法 输电线路 风偏跳闸 故障识别 远程控制识别

2024

电工技术
重庆西南信息有限公司(原科技部西南信息中心)

电工技术

影响因子:0.177
ISSN:1002-1388
年,卷(期):2024.(3)
  • 8