摘要
为了提高对电力营销数据异常的识别能力,基于孤立森林算法设计了一种电力营销数据异常识别方法。首先构建电力营销数据异常特征检测和大数据存储模型,结合分布式随机信息采样方法对电力营销数据异常特征展开传感跟踪,并采用压缩感知方法提取电力营销数据异常特征。在分析其谱结构的基础上,通过随机解耦性特征分解方法分解数据异常谱特征,然后建立信息融合模型,从中提取异常电力营销数据的关联特征量,再通过关联规则调度和粗糙集特征匹配方法检测和识别异常数据的特征。在这一过程中,采用孤立森林算法对识别结果进行寻优控制,提高识别过程的收敛性。仿真结果表明该方法对电力营销数据异常的识别精度较高,表明该方法识别性能较好。
In order to improve the ability to recognize the anomalies of electric power marketing data,a method for identifying the anomalies is designed based on the isolated forest algorithm.First of all,a model for the anomaly characteristics of electric power marketing data is constructed for testing and large data storage.Combining distributed random information sampling method,the anomaly characteristics of the electric power marketing data can be tracked through sensors,and the compression perception method is used to extract the anomaly characteristics.On the basis of the analysis of the spectral structure,by decoupling decomposition characteristics and randomly analyzing data anomalies spectrum characteristics,an information fusion model is set up to extract abnormal correlation characteristics of electric power marketing data quantity Through association rules and rough set feature matching method the characteristics of the abnormal data can be detected and identified.In this process,the isolated forest algorithm is used to optimize the recognition results and improve the convergence of the recognition process.The simulation results show that this method has a high accuracy in the identification of power marketing data anomalies and a good performance in the identification.
作者
陈婷
许睿
孟维丽娅
刘畅
许蕾
胡文彦
CHEN Ting;XU Rui;MENG Weiliya;LIU Chang;XU Lei;HU Wenyan(Power Customer Service Center, Yunnan Power Grid Co. Ltd., Kunming 650000, China)
出处
《微型电脑应用》
2022年第6期75-78,共4页
Microcomputer Applications
关键词
孤立森林算法
电力营销数据
异常识别
谱特征提取
信息融合
粗糙集特征匹配
isolated forest algorithm
power marketing data
abnormality identification
spectral feature extraction
information fusion
rough set feature matching
