Risk Prediction Techniques for Power Control System Network Security

As a fundamental energy infrastructure, the power system has a direct impact on people’s daily lives and production. If the security of the electrical power technology is compromised by malicious network attacks, it can result in widespread blackouts and have significant consequences on every aspect of society. Therefore, to ensure power control system security, this study introduces a network security risk prediction algorithm for power control systems. The algorithm is based on categorization-constrained Boltzmann machine and Markov time-varying model. It classifies network security risk states using classification-constrained Boltzmann machine and predicts security risk using Markov time-varying model. The experimental results demonstrate that the classification-constrained Boltzmann machine achieved higher average accuracy (78.4%), average precision (76%), and average recall (73.4%) for the network risk state compared to the Hidden Markov Model, Multi-kernel Support Vector Machines, and Bayesian Networks. Therefore, it is evident that the classification-constrained Boltzmann machine is better suited for the power control system. Additionally, when considering the prediction of security risks, the categorization-constrained Boltzmann machine and Markov time-varying model produced impressive results with a combined accuracy, precision, F-score, and consistency index of 0.964, 0.947, 0.96, and 0.974, respectively. These results demonstrate that this particular security risk prediction method has high accuracy and is adaptable to complex equipment and network environments found in power control systems.