分项计量平台数据质量故障诊断与修复

Data Quality Fault Detection,Diagnosis and Recovery in Sub-metering System

为满足能耗监管的目的,有大量电表支撑的分项计量平台在中国商业建筑中得到推广和使用,电耗数据可用来进行能耗分析。然而,由于环境电磁干扰以及其他电表故障,电耗数据的质量受到质疑。提出了一种基于相关系数和数据光滑度的方法,来分别诊断等比例偏离故障和环境干扰故障。对上海某建筑的测试发现,当故障电表个数较少时,该方法可获得较高的准确率。考虑到多个电表同时发生故障的可能性较小,该方法可应用于工程实际。

Due to the increasing power metering devices in commercial buildings in China for energy monitoring, significant amount of electricity consumption data are now available for analysis. However, because of environmental electromagnetic disturbance and various faults that could occur with electricity meters, quality of the collected data is a question. A methodology based on correlation coefficient and data smoothness is proposed to detect and recover the proportional deviation faults and environmental disturbance faults, respectively. Testing this methodology with data collected from a real building in Shanghai, it is found that a high accuracy can be achieved given a small number of faulty sub-meters. Considering the low possibility of simultaneous multiple faults, this method is practical in real situations.