Data-Driven Probabilistic Anomaly Detection for Electricity Market under Cyber Attacks

Information and communication technologies have been widely used in smart grid for efficient operation. However, these technologies are vulnerable to malicious cyber attacks, which may lead to severe reliability and economic issues. Recently, a variety of data-driven anomaly detection approaches hav...

Full description

Saved in:
Bibliographic Details
Main Authors: Sun, Mucun, Ren, Lingyu, Chiang, Nai-yuan
Format: Conference Proceeding
Language:English
Subjects:
Electric potential
Electricity supply industry
Numerical models
Probabilistic logic
Reliability
Sensitivity analysis
Smart grids
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Information and communication technologies have been widely used in smart grid for efficient operation. However, these technologies are vulnerable to malicious cyber attacks, which may lead to severe reliability and economic issues. Recently, a variety of data-driven anomaly detection approaches have been explored to detect potential cyber attacks in smart grids. In this paper, we researched on the electricity market data aiming to identify anomalies from the locational marginal prices (LMPs) and provide a new indicator for potential cyber attacks in power grids. Specifically, a novel data-driven probabilistic anomaly detection framework is proposed for electricity market, which consists of three major components: long short-term memory (LSTM) based deterministic electricity price forecasting, probabilistic electricity price forecasting and anomaly detection. This framework is tested on a model-based electricity market simulator under two types of cyber attacks, i.e., load redistribution attack (LRA) and price responsive attack (PRA). Numerical results on the simulated LMPs show that the proposed framework is capable of detecting data anomalies over these attacks.
ISSN:2378-5861
DOI:10.23919/ACC50511.2021.9482640