Next-Generation Digital Twin Analytics in Smart Manufacturing using Cross-Layered Edge Cloud Deep Learning

The emergence of digital twins is transforming smart manufacturing by enabling real-time system monitoring, predictive maintenance, and optimization of industrial processes. Despite these advantages, traditional cloud-centric analytics face limitations such as high latency, bandwidth constraints, and reduced responsiveness, which hinder deployment in dynamic factory environments. To address these challenges, a cross-layered edge–cloud framework integrated with deep learning is proposed. The architecture employs long short-term memory (LSTM) networks to model temporal dependencies in sensor data, enabling accurate prediction of equipment degradation. Edge devices are responsible for local data acquisition, preprocessing, and lightweight LSTM-based inference, while the cloud layer is utilized for large-scale model training, digital twin synchronization, and advanced analytics. A bidirectional communication strategy ensures that updated cloud models are periodically transferred back to edge nodes, supporting continuous refinement of predictive intelligence while maintaining low-latency inference. The framework is validated using publicly available datasets, including the AI4I 2020 predictive maintenance dataset. Experiments show that our proposed learning model can improve the performance of the cloud manufacturing platform in real-life applications efficiently. Finally, the integration of deep learning with cross-layered edge–cloud computing establishes a scalable and adaptive foundation for digital twin analytics in next-generation smart manufacturing.