电力营销海量客户交易行为异常检测系统设计  

Design of abnormal transaction behavior detection system for massive customers in power marketing

作　　者：左航 Zuo Hang(Yinchuan Power Supply Company of State Grid,Yinchuan Ningxia750011,China)

机构地区：[1]国网银川供电公司,宁夏银川750011

出　　处：《现代科学仪器》2023年第3期17-22,共6页Modern Scientific Instruments

基　　金：国网宁夏电力有限公司科技项目(5229XT20003T)。

摘　　要：为提高电力营销中异常数据的检测能力,设计一种电力营销海量客户交易行为异常检测系统。通过优化系统中央控制器以及数据存储器功能,并设定电源电路为优化后设备供能完成系统的硬件设计,引入聚类分析算法获取用户交易异常信息,并将其输出,实现系统基础功能,实现电力营销海量客户交易行为异常检测系统的软件设计。性能测试结果表明:本文系统的功能测试合格,满足使用功能的要求;本文系统对海量数据的承载力较高,异常数据检测误差小,行为异常区分正确率高。本文系统表现出较好的应用性能。In order to improve the ability of detecting abnormal data in power marketing,an abnormal detection system of massive customers'trading behavior in power marketing is designed.By optimizing the functions of the central controller and data memory of the system,and setting the power supply circuit to power the optimized equipment,the hardware design of the system is completed,and the clustering analysis algorithm is introduced to obtain the abnormal information of user transactions and output it,so as to realize the basic functions of the system and realize the software design of the abnormal trading behavior detection system for massive customers in power marketing.The performance test results show that the function test of this system is qualified and meets the requirements of using functions;In this paper,the system has high bearing capacity for massive data,small error in abnormal data detection and high accuracy in distinguishing abnormal behaviors.This system shows good application performance.

关 键 词：电力营销 聚类分析 行为异常 交易异常 检测 大数据分析 卷积神经网络 数据库 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]