Risk Assessment in Automated Manufacturing Systems: A Hybrid Framework for Industry 4.0 and Beyond

Abstract

Industry 4.0 has transformed manufacturing systems through the integration of cyber-physical systems, Internet of Things (IoT), machine learning, and digital twin technologies. While this interconnected architecture improves efficiency and flexibility, it also introduces complex and interdependent risk scenarios that are not adequately addressed by conventional assessment methods. This paper proposes a hybrid risk assessment framework for automated manufacturing systems that combines classical techniques such as Failure Mode and Effects Analysis (FMEA) and Hazard and Operability Study (HAZOP) with machine learning-based prediction and digital twin simulation. The framework incorporates a composite risk scoring model integrating static risk prioritisation with real-time predictive analytics and contextual factors. A four-layer architecture is developed, comprising risk taxonomy, dynamic monitoring, simulation-based validation, and governance. The proposed approach is evaluated using benchmark manufacturing datasets, including the PHM 2010 milling dataset and the CWRU bearing dataset, under stratified k-fold cross-validation. Results indicate that the hybrid FMEA + Random Forest model achieves superior performance, with an F1-score of 0.972 and Area Under the Curve (AUC) of 0.981, outperforming standalone models. Case studies across automotive, electronics, smart factory, and pharmaceutical domains demonstrate the practical applicability of the framework. The study further highlights the role of explainable artificial intelligence (XAI) in improving transparency and trust in risk prediction systems. The findings suggest that hybrid, data-driven approaches are essential for effective risk management in Industry 4.0 and emerging Industry 5.0 environment.