基于深度学习技术的电表大数据检测系统
Big Data Detection System of Electric Meter Based on Deep Learning Technology

随着我国电厂不断发展，我国智能电表装机量不断扩大，日臻成熟，对智能电表的监测越来越重要。本文通过对电表数据的采集、清洗，完成数据格式化。运用皮尔森相关系数分析以及K折交叉验证等方法，进行数据分析。通过采用深度学习时序模型进行预测研究，最终达到检测电表运行状态的目的。通过利用智能电表大数据对电表运行状态的分析，可以判断电表运行是否正常，如果异常是属于故障还是有偷漏电发生，判断相关位置，以便进一步采取行动。该检测系统的研究与应用，可以避免智能电表的物理检测，可以达到延长正常电表的使用寿命，节省大量的资源的目的。
With the sustained development of power plant of China, the installed base of electric meter grows gradually. It is more and more important to monitor smart meters with their mature use. The data formatting is completed through the collecting and cleaning to the data from meters. The data analysis is done by the methods including Pearson correlation coefficient analysis and K fold cross validation etc. Eventually testing abnormity and failure recognition can be obtained through prediction research based on deep learning time series model. Based on electric meter running big data analysis, the electric meter situation including normal and abnormal (error or electricity stealing) and relevant position can be obtained so as to take further action. The research and application of the testing system can avoid physical testing to electric meter. The service time of normal meter can be prolonged by abnormal meters testing. This will lead to saving a lot of resources.