Disordered Electricity Consumption Behavior Identification Method Based on Demand Response Load and Big Data

Traditional research on electricity consumption behavior is relatively backward in processing and analyzing large-scale data, and cannot make full use of big data technology to deal with complex electricity consumption behavior patterns, usually using the whole power grid load data, and lacking in-depth understanding of fine-grained electricity consumption behavior. This results in a lack of accurate analysis of individual user and device-level behavior, making it difficult to capture changes in usage patterns across time and place. Based on power system data of European countries and other data sets, this article designs a Convolutional Neural Network and Gated Recurrent Unit (CNN-GRU) to extract local and global features from time series data. Combined with the time series to predict the response load power consumption, the average accuracy rate of CNN-GRU algorithm model for load power consumption was 0.7845, the average recall rate was 0.97, and the average F1 Score was 0.8393 in five experiments. The proposed algorithm model can accurately predict the load consumption.