QTE-Net: A Quantum-Inspired Temporal Learning Network for Intelligent Fault Detection in Industrial IoT

Abstract

The Industrial Internet of Things (IIoT) begs for intelligent systems, which will require highly accurate, low latency, scalable systems that operate on constrained edge devices and components. This work proposes a novel intelligent fault detection approach for multimodal industrial sensor streams named Quantum-Inspired Temporal Encoding Network (QTE-Net), which processes vibration, thermal, acoustic, current, and voltage data. It integrates adaptive signal enhancement, the temporal representation of features inspired by the quantum world, the cross-scale attention mechanism in deep learning, the explainable predictive fault reasoning, and continuous learning for dynamic adaptation in industry. According to the experimental results, the proposed model in this paper could obtain the best results as 97.92% of accuracy, 97.14% of precision, 96.95% of recall, and 192ms of average computing time in inferences of different types of industrial fault. The results demonstrate the excellent defect detection performance for complex machine defects and the satisfactory computational efficiency. The proposed solution offers compelling, personalized maintenance for next-generation smart industrial monitoring and resilient Industry 4.0 applications intelligent, reliable, and sustainable