Shipboard Microgrid Frequency Control Based on Machine Learning Under Hybrid Cyberattacks

Nowadays, marine vessels are critical to the industries and trade of many countries. Modern ships require clean energy and modern communications systems to function properly. The use of renewable energy resources requires efficient coordination using reliable communications, which can be threatened by cyberattacks. An attacker's ultimate goal is to collapse systems by implementing hybrid cyberattacks. Accordingly, this article studies the impact of hybrid cyberattacks involving a combination of denial of service and false data injection, as well as the implementation of attacks on both actuator and measurement signals for load frequency control performance in a low-emission shipboard microgrid. In this regard, a parallel attack detection strategy based on machine learning is developed to detect the attacks imposed on the measurement and actuator signals. A novel resilient control scheme, called adaptive reinforcement learning active disturbance rejection control (ADRC), is also presented to compensate for the impacts of attacks on the system frequency performance and ensure the ship can operate under different attack scenarios. Due to the simplicity and ease of implementation of ADRC, the ability to use reinforcement learning to tune parameters, and the resilience of the technique against attacks, the proposed strategy is a viable method as opposed to other mathematically complicated methods. The feasibility of the proposed method under various conditions has been evaluated by hardware-in-the-loop real-time simulations.