Electricity Price Default Detection Model Based on Long Short-Term Memory Network

Considering the long-term memory characteristics exhibited by user groups implementing the same electricity pricing strategy on time series data of electricity consumption, as well as the dynamic changes in user electricity consumption behavior, a long short-term memory network (LSTM)-based electricity price default user detection model was constructed. First, an autocorrelation analysis was conducted on the time series of electricity consumption for different electricity price categories to illustrate the long-term memory of users' electricity consumption patterns. Second, the time series data of electricity consumption was converted into a tensor form, and a classification model based on LSTM was constructed. At the same time, L1 regularization was applied to the model, and the L1 norm of the LSTM layer weight parameters was added as a regularization term in the loss function, making the model more focused on features that have a key impact on the prediction results. The experimental results showed that the model proposed in this paper could deeply analyze user electricity consumption data, accurately identify abnormal users in data sets with abnormal electricity price labels, and provide solid support for monitoring the implementation of electricity prices.