Digital-Twin Technology Across Manufacturing, Healthcare, Smart Cities, and Energy Systems: A Systematic Review and Experimental Validation

Abstract

Digital-twin (DT) technology has evolved into a fundamental paradigm for integrating physical assets, processes, and systems with their dynamic digital counterparts, enabling real-time monitoring, simulation, prediction, optimization, and intelligent decision-making. This study presents a systematic review of 230 peer-reviewed articles published between 2018 and 2026, examining the evolution, application domains, enabling technologies, implementation challenges, and emerging research directions of DTs across manufacturing, healthcare, smart cities, and energy systems. The review synthesizes the growing convergence of DTs with artificial intelligence (AI), the Internet of Things (IoT), blockchain, edge computing, cloud computing, and cyber–physical systems, highlighting their role in developing intelligent, interconnected, and data-driven environments. Across the four domains, the principal challenges identified include heterogeneous and distributed data integration, cybersecurity and privacy, physical–virtual synchronization, interoperability, scalability, model fidelity, and computational and communication constraints. The review further identifies eight emerging research directions, including AI-enabled autonomous DTs, blockchain-assisted trust and security mechanisms, federated learning, 6G-enabled DT networks, adaptive and self-learning architectures, and sustainable and energy-efficient DT infrastructures. To complement the literature synthesis with practical evidence, two experimentally validated DT prototypes developed at the Laboratory of Advanced Electronic Systems (LSEA) and the Laboratory of Applied Automation and Industrial Diagnostics (LAADI), Algeria, are presented. The first integrates CIROS Studio, VGG16, and Node-RED for drilling-operation quality control, achieving an accuracy improvement from 98.63% to 99.83% following fine-tuning. The second combines Factory I/O, MobileNet, fuzzy logic, and a Gradio-based human–machine interface for conveyor-belt sorting, achieving true-positive and true-negative rates of 99.90% and 99.10%, respectively. The combined findings demonstrate that the integration of DTs with AI, industrial automation, and simulation can substantially enhance monitoring, prediction, quality control, and intelligent decision-making. At the same time, they emphasize the need for interoperable, secure, scalable, computationally efficient, and energy-aware architectures to enable reliable real-time deployment of DTs across heterogeneous application domains.