Enhancing smart manufacturing: a tensor-based ontology framework for predictive optimization using semantic digital twin

The rapid advancement of Industry 4.0 has driven the need for intelligent, responsive manufacturing systems that can anticipate and mitigate disruptions, reduce downtime, and optimize performance. However, traditional manufacturing systems often rely on reactive maintenance strategies, leading to unexpected downtimes and increased operational costs. These limitations necessitate a shift toward predictive optimization for real-time condition monitoring and proactive failure prevention. This study presents a Tensor-Based Semantic Digital Twin (SDT) framework that integrates a hybrid CNN-LSTM model with ontology-based decision-making to provide a context-aware predictive optimization system. The SDT framework outperforms traditional methods, such as Deep Learning-Based Predictive Maintenance (DL-PdM) and Hybrid Anomaly Detection in Manufacturing (HAD-M), by significantly reducing equipment downtime and lowering costs per unit while maintaining high predictive accuracy and operational efficiency. The study demonstrates that SDTs provide real-time operational intelligence, ensuring manufacturing systems are more resilient, sustainable, and cost-effective. Unlike traditional approaches that rely on static models, the proposed framework enables adaptive learning from evolving sensor data, allowing manufacturing facilities to anticipate failures proactively and optimize resource utilization dynamically. By leveraging a hybrid CNN-LSTM model, the framework accurately predicts equipment failures, ensuring a 30% reduction in unexpected downtimes and a 20% improvement in energy consumption efficiency. Additionally, ontology-based decision-making enables context-aware automation, leading to an 18% decrease in maintenance costs and adaptive load balancing in dynamic industrial environments. Additionally, the proposed system extends beyond predictive maintenance by incorporating automated reactive control, making real-time adjustments to minimize inefficiencies. The findings establish a new standard for intelligent manufacturing, ensuring enhanced resilience, adaptability, and cost efficiency of smart manufacturing.