An abnormal behavior detection based on deep learning

The electric power industry big data contains a large amount of event information for power companies, mining the abnormal electricity data is of great significance, the efficiency of abnormal electric current detection methods is not high, looking for an accurate and efficient method has become an urgent problem in electric power industry. This paper presents the abnormal electrical detection of deep learning model based on the convolution network algorithm supervised multilayer hidden layer, the model is trained on a large number of user data has been marked, can be used for automatic detection of user consumption is normal, screening out the abnormal users, provide guidance for on-site verification of power companies, improve the the hit rate of efficiency and investigate the abnormal situation of electricity found.