by Ferdinand Fabrice Ayissi Zogo¹*, Jacques Matanga¹, and Jean François Essiben Dikoundou¹
¹ National Higher Polytechnic School of Douala, University of Douala, P.O. Box 2701, Douala, Cameroon
*Corresponding author: [email protected]
MJ Engineering Sciences, 1(1), 31-48.https://doi.org/10.63156/mjes04
Received: 18.01.2025 Accepted: 26.07.2025 Published online: 11.08.2025
| The fast digitalization of industrial systems, facilitated by Industry 4.0 technology, has created increasingly complex IT risks, such as cyberattacks, operational interruptions, and data breaches, all of which can have a negative impact on productivity, safety, and financial stability. Traditional risk assessment approaches frequently fall short of tackling these dynamic threats due to their reactive nature and inability to process vast amounts of real-time data. To bridge this gap, this study provides a Comprehensive Hybrid framework that combines mathematical modeling, deep learning, and IoT-driven analytics to enable proactive industrial IT risks anticipation. To evaluate possible vulnerabilities across interconnected industrial networks, the framework uses mathematical optimization techniques such as stochastic modeling, Bayesian risk assessment, and Game theory optimization. It uses Deep Neural Networks (DNN), such as Long Short-Term Memory (LSTM) and the Autoencoder for Unsupervised Anomaly identification, to analyze historical and real-time data for anomaly identification, predictive risk assessment, and adaptive threat forecasting. Additionally, distributed IoT sensor networks deliver continuous, high-resolution data streams from vital infrastructure, allowing for real-time monitoring. Through this synergistic integration, the framework improves IT risk prediction accuracy through multi-modal data fusion and adaptive learning, while maintaining data privacy via federated learning implementations. When tested on real-world datasets, the model outperformed current techniques. |
