Enhancing State Estimation Accuracy in Power Systems: An ANN-Based Data Mining Approach Defending Cyber Attacks

In energy systems, measurement accuracy is jeopardized by bad data arising from cyber attacks. When bad data is detected in the measurement dataset as a result of cyber attacks, it's essential to identify and eliminate these data. However, this elimination process introduces the problem of missing measurement data, threatening the system's observability conditions. This study proposes a data mining approach supported by artificial neural networks to address the missing measurement data issue when bad data is detected. Our proposed method aims to maintain the system's observability by completing the measurement data lost due to bad data. Consequently, the measurement set purified from bad data enhances the accuracy of the crow search algorithm based state estimation results. This methodology has been shown to successfully mitigate the adverse effects of unforeseen situations, such as cyber attacks.