文章摘要
王海燕,袁新平.基于态势感知的电网停电用户敏感度及投诉预测[J].电力需求侧管理,2023,25(2):107-111
基于态势感知的电网停电用户敏感度及投诉预测
User sensitivity and complaint prediction for power outage based on situational awareness
投稿时间:2022-12-15  修订日期:2023-01-18
DOI:10.3969/j.issn.1009-1831.2023.02.017
中文关键词: 电网停电  用户敏感度  投诉预测  态势感知  随机森林算法
英文关键词: power grid outage  user sensitivity  complaint prediction  situational awareness  random forest algorithm
基金项目:中国南方电网有限责任公司信息化项目(21059300HK42210018001)
作者单位
王海燕 云南电网有限责任公司 信息中心昆明 650214 
袁新平 云南电网有限责任公司 信息中心昆明 650214 
摘要点击次数: 627
全文下载次数: 182
中文摘要:
      在电网停电用户敏感度及投诉预测中,由于预测结果不准确影响了电网公司的精准化服务,因而设计一种基于态势感知的电网停电用户敏感度及投诉预测方法。通过SAS软件中的 Enterprise Miner workstation 模块和 Enterprise Guide模块采集电网停电用户敏感度及投诉相关数据,具体包括停电敏感用户标签数据、故障处理数据、停电事件数据、客户通话数据、95598 工单数据。对挖掘数据实施缺失数据处理、异常数据处理以及告警误报漏报数据处理等预处理。基于态势感知技术与随机森林算法构建电网停电用户敏感度及投诉预测模型,实现用户对于停电的敏感度及投诉预测。利用该方法对某地区电网实施用户关于停电的敏感度及投诉预测,测试该方法的预测性能。测试结果表明该方法有着高于90%的查准率、查全率,F 测度数据值较高,AUC 面积较大,数据灵敏度始终大于97%,说明设计方法有着优越的电网停电用户敏感度及投诉预测性能。
英文摘要:
      In the power grid outage user sensitivity and complaint prediction, the inaccurate prediction results affect the precise service of the power grid company, so a situational awarenessbased grid outage user sensitivity and complaint prediction method is designed. Through the Enterprise Miner workstation module and the Enterprise Guide module in the SAS software, the grid power outage user sensitivity and complaint-related data are collected, including the power outage sensitive user tag data, fault handling data, power outage event data, customer call data, and 95598 work order data. Perform preprocessing on the mining data such as missing data processing, abnormal data processing, and alarm false positive and false negative data processing. Based on situational awareness technology and random forest algorithm, a grid outage user sensitivity and complaint prediction model is constructed to realize user sensitivity to outage and complaint prediction. The method is used to predict the sensitivity and complaints of users about power outages in a power grid in a certain area, and the prediction performance of the method is tested. The test results show that the method has a precision and recall rate higher than 90%, the F-measure data value is high, the AUC area is large, and the data sensitivity is always greater than 97%, indicating that the design method has superior grid outage user sensitivity and complaint prediction performance.
查看全文   查看/发表评论  下载PDF阅读器
关闭