Application of machine learning in industrial safety: Digital innovations and new safety paradigms in refinery processes

In modern refinery facilities, complex industrial processes require a high level of safety to minimize risks to human lives, property, and the environment. Traditional monitoring systems and safety procedures increasingly show limitations in detecting anomalies and predicting potential incidents. The application of machine learning (ML) in industrial safety enables a transformation of safety paradigms through the analysis of large volumes of data from sensors, SCADA systems, and other industrial sources. This paper explores current digital innovations in ML algorithms for predictive analytics, automatic anomaly detection, and optimization of safety procedures in refinery processes. Focus is placed on integrating ML models into existing risk management systems, implementation challenges, and opportunities for enhancing industrial safety through a proactive approach. The research results indicate that the application of machine learning significantly contributes to reducing incidents, improving the efficiency of safety operations, and opening new perspectives for digital transformation in industrial safety.