Study on anomaly detection algorithm of electricity consumption data through time series analysis
Jiyong Guo, Guoli Lv, Ayun ga, Wenjiao Lu · 2025
The following research is focused on the identification of anomalies in power consumption data by applying the time series analysis algorithm to address a variety of issues, including malfunctioning of devices, unusual patterns of power consumption, and power load forecasting in the power grid. This paper considers an LSTM model that more intelligently detects anomalies through data preprocessing, model training, and reconstruction error calculation. Results indicate that the performance evaluation of the LSTM model is good in terms of accuracy, recall, and F1 score, and it identifies anomalies with a high degree of accuracy within electricity consumption data. This paper validates an algorithm's practicality through empirical analysis and establishes various applications for identifying power equipment malfunctions and analyzing study power consumption patterns, which help predict future power demands. Results indicate that such would hugely improve the performance and financial gains of the power system according to time series analysis using LSTM.