Distributed Acoustic Sensing (DAS) systems face increasing challenges in massive data processing and real-time fault diagnosis due to the growing complexity of industrial environments and data volume. To address these issues, an end-to-end diagnostic framework is developed, integrating Mel-Frequency Cepstral Coefficients (MFCCs) for high-efficiency signal compression and Liquid Neural Networks (LNNs) for lightweight, real-time classification. The MFCC algorithm, originally used in speech processing, is adapted to extract key features from DAS vibration signals, achieving compression ratios of 60–100× without significant information loss. LNNs’ dynamic topology and sparse activation enable high accuracy with extremely low latency and minimal computational cost, making it highly suitable for edge deployment. The proposed framework was validated both in simulated environments and on a real-world conveyor belt system at Qinhuangdao Port, where it achieved 100% accuracy across four vibration modes over 14 weeks of operation. Comparative experiments show that LNNs outperform traditional models such as 1D-CNN and LSTMs in terms of accuracy, inference speed, and model size. The proposed MFCC-LNN pipeline also demonstrates strong cross-domain generalization capabilities in pipeline monitoring, seismic detection, and speech signal processing.
A reflective all-fiber optical current transformer based on a spatial non-reciprocal phase modulation technique is investigated by theoretical analysis and experimental measurement. The modulation unit, composed of a phase delay wave plate (LiNbO3) and two Faraday rotators, achieves flexible frequency adjustment by converting modulation from the time domain to the spatial domain. Therefore, the avoidance of the impact caused by delay coils is achieved in principle. The absence of intrinsic frequency limitations eliminates the demand for precise timing control in demodulation, thereby simplifying the demodulation circuit and reducing the cost and size of the transformer. In previous studies, redundancies were identified in the optical path coupling devices. The half-wave voltage of the modulator is excessively high, and its size is considerable due to constraints inherent in the manufacturing process. The measurement range is within 1800 A. The scheme simplifies some optical path components. By optimizing the phase delay wave plate, the half-wave voltage of the modulator is significantly reduced by a factor of 150. Experimental results demonstrate that the current transformer exhibits excellent detection consistency within the rated current range of 30–3600 A (1–120%), the response time is within 3 ms, and the measurement error and peak error reach 0.052% and 0.127%. This configuration provides a novel option for the design and practical application of all-fiber optical current transformers.