Research on dormitory load time series anomaly detection method based on multi-source high-frequency feature fusion
Jinrun Xie
Nanjing Institute of Technology
Zihan Li
Nanjing Institute of Technology
DOI: https://doi.org/10.59429/esta.v12i3.11357
Keywords: multi-source feature fusion; dormitory electricity safety; high-frequency sampling; LSTM autoencoder; time series anomaly detection
Abstract
To deal with the variety of appliances, complex usage patterns, and hidden abnormal events in dormitory electricity use, we propose an anomaly detection method based on multi-source high-frequency feature fusion. We collect data from multiple power circuits, environmental sensors, and high-frequency voltage/current waveforms, then combine them into a multi-dimensional time-series feature set that mixes both low-frequency statistical data and high-frequency dynamic details. In the modeling stage, we use an LSTM autoencoder to learn the normal usage patterns over time, and detect anomalies by checking the reconstruction errors. Tests show that this approach works well for pinpointing sudden power changes, describing appliance behaviors, and spotting abnormal loads. Compared with single-variable detection methods, our multi-source fusion greatly boosts detection accuracy, helps find safety risks earlier, and supports smarter dormitory energy management and protection.
References
[1] Kaselimi M, Protopapadakis E, Voulodimos A, et al. Towards trustworthy energy disaggregation: A review of challenges, methods, and perspectives for non-intrusive load monitoring[J]. Sensors, 2022, 22(15): 5872.
[2] Cruz-Rangel D, Ocampo-Martinez C, Diaz-Rozo J. Online non-intrusive load monitoring: A review[J]. Energy Nexus, 2025, 17: 100348.
[3] Yan L, Tian W, Wang H, et al. Robust event detection for residential load disaggregation[J]. Applied Energy, 2023, 331: 120339.
[4] Gerasimov, G.; et al. Real-Time Event-Based NILM Framework for High-Frequency Metering. arXiv, 2025.
[5] Zamanzadeh Darban Z, Webb G I, Pan S, et al. Deep learning for time series anomaly detection: A survey[J]. ACM Computing Surveys, 2024, 57(1): 1-42.
[6] Lee Y, Park C, Kim N, et al. LSTM-autoencoder based anomaly detection using vibration data of wind turbines[J]. Sensors, 2024, 24(9): 2833.