Towards safer steel operations with a multi model framework for accident prediction and risk assessment simulation.

This research concentrates on an introduction of a multi-model approach integrating Bayesian Networks (BN), Machine Learning (ML) models, Natural Language Processing (NLP) with Sentiment Analysis, Agent-Based Modeling (ABM), and Survival Analysis to improve predictive modelling of accident causation in high-risk steel industries. The significance of the artificial intelligence (AI) based models is that every approach complements other substantiating the hypothesis. Also, the augmentation of prediction accuracy could be achieved through AI approaches contrary to conventional methods. Results reveal that the application of AI model improves the prediction accuracy compared to conventional approaches. BN application uncovers the machine conditions and human errors responsible for causing accidents. Gradient Boosting Machines discussed equipment-related incidents, while NLP analysis demonstrated negative sentiment due to non-compliance with safety protocols. Moving forward, ABM simulations in accidents focus on personal protective equipment (PPE) compliance and machine maintenance. Survival analysis indicated the role of timely interventions in reducing severe accidents. Additionally, temporal insights aid in timing interventions, improving safety strategy efficacy. The outcome of this research discusses advancements in proactive accident prediction and risk management in high-risk steel industrial environments by addressing latent risk factors.