Background: Engineering systems have progressed from conventional rule-based and computer-aided approaches through Industry 4.0 automation toward Industry 5.0 human-centric, resilient, and sustainable paradigms. Artificial intelligence (AI) and machine learning (ML) now enable the transition from reactive monitoring to predictive, adaptive, and autonomous operation by integrating deep learning, reinforcement learning, generative models, and foundation models with industrial Internet of Things (IIoT), digital twins, cyber-physical systems, edge computing, and robotics.
Objective: This article presents a comprehensive multi-layer framework that systematically links data acquisition, intelligent analysis, digital-twin representation, decision support, and autonomous action for next-generation engineering systems, while critically evaluating performance, limitations, and future research needs.
Methods: A conceptual layered architecture is defined and evaluated using representative engineering datasets for predictive maintenance, structural health monitoring, and smart manufacturing. Classical ML models (Random Forest, XGBoost, SVM), deep architectures (CNN, LSTM, GRU, Transformers), and hybrid approaches are compared under standardized preprocessing, cross-validation, and hyperparameter optimization. Explainability is assessed via SHAP and feature importance. Results are distinguished as literature-derived, simulated, or theoretically expected.
Results: AI-enabled methods consistently outperform conventional statistical and rule-based baselines in predictive accuracy, anomaly detection, and remaining-useful-life estimation. Digital-twin synchronization and edge deployment improve real-time responsiveness. Trade-offs emerge among accuracy, computational cost, interpretability, and scalability. Persistent limitations include data heterogeneity, model generalization, cybersecurity exposure, and explainability in safety-critical contexts.
Conclusion: AI and ML fundamentally transform engineering from reactive to autonomous systems. Realizing trustworthy next-generation engineering requires advances in physics-informed models, multimodal foundation models, federated and edge intelligence, and human-AI collaboration aligned with Industry 5.0 principles.