Real-time vibration mode recognition of <i>φ</i> -OTDR based on Opus audio compression and spiking neural networks

This study proposes a vibration monitoring system based on distributed acoustic sensing (DAS) and spiking neural networks (SNN) for real-time fault detection in industrial applications. By integrating Opus audio compression technology with SNN, the framework achieves highly efficient compression of vibration signals, allowing for a significant reduction in data volume (50-80 times compression) without losing critical information such as friction and impact events. This allows for high-quality input for subsequent fault diagnosis. Experimental results show that the system demonstrates exceptional performance, achieving over 98% accuracy in scenarios such as long-distance conveyor monitoring, pipeline monitoring, and railway safety, with minimal computational requirements and rapid real-time processing capabilities. The proposed system's modular design, which includes data acquisition, preprocessing, compression, feature extraction, and result output, ensures reliable fault detection with high generalization across different engineering fields.