Research on Machine-Learning-Based Load Frequency Control Strategy for Power Systems Under Data-Driven DoS Attacks

Focusing on data-driven denial of service (DoS) attacks, this article investigates the intelligent control strategy for the load frequency control (LFC) system of nonlinear power systems. An LFC model of the power system is first established. This model includes renewable energy sources, multiple remote terminal units (RTUs), and energy storage systems. Two nonlinear factors, the generation rate constraint and the governor dead band, are taken into account in the model. A data-driven DoS attack is then proposed based on the data importance awareness strategy. It specifically targets key RTUs for attacks. In comparison with indiscriminate DoS attacks, the proposed DoS attack strategy has a more pronounced negative impact on system stability. The power grid operational performance, measured by the integral of time-weighted absolute error, is used as the objective function. The optimal control strategy is solved via the adaptive moment estimation algorithm in machine learning. Finally, simulation examples are conducted, through which the reliability and effectiveness of the proposed data-driven attack strategy and the machine-learning-based LFC strategy are verified.