智能制造技术下异常数据信息化分析方法
Information Analysis Method of Abnormal Data Under Intelligent Manufacturing Technology
马春生1
作者信息
- 1. 南京软核科技有限公司,江苏 南京 210012
- 折叠
摘要
为了解决现有技术中电能异常数据信息化分析方法计算效率低、检测精度差等问题,设计了一种新型的电能异常数据分析系统,该系统基于局部矩阵重构(Local Matrix Reconstruction,LMR)实现异常电能数据检测,通过使用 5 个日负荷特征代替高维日负荷曲线来表征功耗模式.结合主成分分析(Principal Component Analysis,PCA)计算局部范围内的加权重构误差,最后将每个样本的重构误差与其相邻样本进行比较,以计算局部异常值分数.实验结果表明,该系统处理异常电能数据精度高,2 TB测试数据时误差率仅有 4.2%.该研究的方法大大提高了异常数据信息化分析能力.
Abstract
In order to solve the problems of low calculation efficiency and insufficient detection accuracy of information analysis method of electricity data anomaly in the prior art,a new type of abnormal electricity data analysis system is designed.The system relies on local matrix reconstruction(LMR)to realize abnormal electric energy data detection,and characterizes the power consumption mode by using 5 daily load characteristics instead of high-dimensional daily load curves.Principal component analysis(PCA)is combined to calculate the weighted reconstruction error in the local range,and finally the reconstruction error of each sample is compared with its neighboring samples to calculate the local outlier score.The experimental results show that the system has high accuracy in processing abnormal electricity data with the error rate of only 4.2%upon applying 2 TB test data.The proposed method greatly improves the ability of information analysis of abnormal data.
关键词
智能制造业/异常电能数据/局部矩阵重构/主成分分析/智能电能表Key words
intelligent manufacturing/abnormal electricity data/local matrix reconstruction/principal component analysis/intelligent electric energy meter引用本文复制引用
出版年
2024
NETL
NSTL
万方数据