Ship Intrusion Detection Technology Based on Bayesian Optimization Algorithm and XGBoost

Cybersecurity is a prerequisite for maritime safety. This study focuses on intelligent ship integrated automation system, analyzing the principles and shortcomings of these system. We propose a machine learning-based intrusion detection solution that utilizes Recursive Feature Elimination (RFE) as a feature selection algorithm and XGBoost as a classification algorithm. The Bayesian optimization algorithm is applied to optimize the hyperparameters of the XGBoost model for detecting anomalous network data. Results indicate that, compared to decision trees, logistic regression, random forests, KNN, and recent research findings, the proposed method exhibits smaller estimation errors with an overall accuracy of 99.8%. It effectively reduces vulnerabilities in ship networks, achieving a F1-score of 96.37% and a recall rate of 96.33% under sufficient dataset conditions. This approach enhances network security performance and elevates the cybersecurity level of intelligent ships.