A survey of intruder detection in smart grid systems: comparative analysis of rule-based, machine learning and deep learning

The increasing sophistication of smart grids as cyber-physical systems renders them increasingly susceptible to cyberattacks. Attack detection has potential to be enhanced by implementing deep learning-based IDS methods. Deep learning-based IDS can improve attack detection in smart grids. However, challenges arise in managing vast data volumes generated across multiple communication levels when integrating deep learning into these systems. These issues are intensified by the exertion of detecting to illegal access, particularly in SCADA systems that use multi-tier protocols such as DNP3. Furthermore, the rapid growth of IoT in smart grids increases data processing demands, challenging DL model scalability and IDS integration due to standardisation and communication compatibility issues. Organisations struggle with resource allocation for employee training, technology updates, and security maintenance. Addressing these challenges requires goal-oriented strategies. This article explores enhancing smart grid flexibility through security upgrades, strategic planning, and regulatory compliance for effective IDS integration.