Real-Time Threat Detection in Critical Infrastructure with Machine Learning and Industrial Control System Data

The present study introduces a framework for the detection of threats in critical infrastructure in real-time. This framework leverages the capabilities of machine learning and utilizes data from Industrial Control Systems (ICS). In order to evaluate the accuracy of threat identification, we conducted a comparative analysis of three well-known machine learning models: Logistic Regression, Random Forest, and Knearest Neighbors. Our study specifically emphasized the importance of precision as a key parameter in this assessment. The results of our study indicate that Logistic Regression has higher performance compared to the other models, emerging as the most effective method for effectively identifying possible hazards in real-time ICS data. This study highlights the effectiveness of Logistic Regression in strengthening security protocols within vital infrastructure, presenting a hopeful resolution to protect against rising risks and weaknesses.