Optimization of the Random Forest Hyperparameters for Power Industrial Control Systems Intrusion Detection Using an Improved Grid Search Algorithm

The intrusion detection method of power industrial control systems is a crucial aspect of assuring power security. However, traditional intrusion detection methods have two drawbacks: first, they are mainly used for defending information systems and lack the ability to detect attacks against power industrial control systems; and second, although machine learning-based intrusion detection methods perform well with the default hyperparameters, optimizing the hyperparameters can significantly improve its performance. In response to these limitations, a random forest (RF)-based intrusion detection model for power industrial control systems is proposed. Simultaneously, this paper proposes an improved grid search algorithm (IGSA) for optimizing the hyperparameters of the RF intrusion detection model to improve its efficiency and effectiveness. The proposed IGSA boosts the speed of calculation from O(n(m)) to O(n x m). The suggested model is evaluated based on the public power industrial control system dataset after hyperparameter optimization. The experiment results show that our method achieves a superior detection performance with the accuracy of 98% and has more outstanding performance than the same type of work.

Automated summary

This paper introduces an intrusion detection method for power industrial control systems and proposes an improved algorithm to optimize the performance of the model. The experimental results demonstrate that the method achieves superior detection performance and outperforms similar approaches.