@article{MAED6861F,
title = "Electric Power Time Series Data-Based Unsupervised Anomaly Detection Method for Equipment",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2024",
issn = "1226-4717",
doi = "10.7840/kics.2024.49.7.993",
author = "Joonhee Lee, Dongwoo Kwon, Youngmin Ji",
keywords = "EMS, Anomaly Detection, AutoEncoder, LSTM",
abstract = "Recent greenhouse gas and energy regulations have increased the need to reduce energy use. If an anomaly occurs in the equipment, unnecessary energy consumption may occur or productivity may be lowered. This has increased the need for anomaly detection for energy-consuming equipment. This paper monitor and analyze result applying LSTM-AutoEncoder-based anomaly detection algorithm method for real-time power usage data of energy-consuming equipment in buildings and factories. If an anomaly occurs in the equipment, signs of anomaly appear in the power-useage data, and a pattern different from the normal pattern. By applying the reconstruction-based anomaly detection algorithm to power usage data, we have shown that anomaly patterns in energy-consuming equipment can be detected in real-time."
}