Anomaly Detection for Power System Forecasting Under Data Corruption Based on Variational Auto-encoder

摘要

With the expansion of information technology in the era of smart grid, cybersecurity vulnerability issues are becoming a major threat for the modern power system. Among various results of cyberattacks, data corruption may lead to poor forecasting performance in the power system issues. To tackle this problem, this paper introduces an anomaly detection method for power system forecasting based on variational auto-encoder. This reconstruction-based anomaly detection method adaptively discovers the potential common patterns of data series and find the anomalies different from common patterns. Taking load forecasting as example, the numerical results of case studies on the dataset of Global Energy Forecasting Competition 2014 demonstrate the outstanding effectiveness of proposed method over other competing methods.