2026· IEEE Transactions on Instrumentation and Measurement· Vol 75, pp. 2519311-2519311· 0 citations· 38 references
TL;DR
Results indicate that the proposed DF-CLDNN enhances feature discriminability and measurement quality, providing an effective solution for high-precision event recognition in <inline-formula> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR-based intrusion detection under complex conditions.
Abstract
To enhance the accuracy and robustness of optical fiber distributed vibration measurement under complex environmental noise, this work proposes a phase-sensitive optical time-domain reflectometry (<inline-formula> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR)-based signal processing and event recognition framework. First, a multistage adaptive denoising algorithm is constructed by combining wavelet threshold decomposition, adaptive impulse suppression, and robust principal component analysis (RPCA). This module effectively decomposes the raw signal into a low-rank background component and a sparse intrusion component, thereby achieving enhanced denoising. Subsequently, a convolutional long short-term memory fully connected deep neural network (CLDNN) architecture is designed to perform adaptive denoising decomposition and fuse multidomain features, referred to as denoising and feature fusion-based CLDNN (DF-CLDNN). Experiments based on the Rayleigh backscattering (RBS) phase signals measured by an <inline-formula> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR system are conducted. Experimental results show that the proposed method achieves the highest signal-to-noise ratio (SNR) of 18.04 dB and the best phase measurement accuracy (PMA) of 0.097 rad, improving the SNR by 5.93 and 7.65 dB over adaptive wavelet thresholding (AWT) and RPCA. On a six-class event dataset, the proposed DF-CLDNN achieves a testing accuracy of 98.67 % and maintains 95.34 % accuracy on an independent unseen dataset. Furthermore, under a severe 5-dB low-SNR condition, the proposed DF-CLDNN still achieves a recognition accuracy of 90.10 %. These results indicate that the proposed DF-CLDNN enhances feature discriminability and measurement quality, providing an effective solution for high-precision event recognition in <inline-formula> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR-based intrusion detection under complex conditions.
A lightweight multi-scale time–frequency network (MTFNet) for direct processing of one-dimensional DAS signals is proposed, demonstrating its robustness against additive white Gaussian noise (AWGN) and practical engineering application potential.
Qian-Kun Zhang, Yang-Yang Sun, Zhi-Long Chen et al.· Italian National Conference...· 0 citations
Real-time anomaly detection is a core requirement in industrial monitoring, surveillance, biomedical diagnostics and
predictive maintenance, where unusual events must be identified within strict latency limits. This paper presents a unified
framework that combines digital image processing (DIP) and digital signal proce...
Monika Kailas Khairnar· International Journal for Re...· 0 citations
The Industrial Internet of Things (IIoT) is being used a lot in important areas like advanced manufacturing, smart energy systems, and intelligent cities. Watching for intrusion detection is very important for the safety of the IIoT. Nevertheless, multivariate sensor sequences frequently demonstrate pronounced physical...
Hao-Ran Lei, Shi-Ming Li, Wen-Tao Li et al.· Italian National Conference...· 0 citations
The reliable detection of hidden rebar debonding in reinforced concrete structures remains a critical challenge. Addressing this issue is essential for advancing structural health monitoring. The prevailing data-driven rebar debonding detection methods predominantly rely on single-dimensional physical features, which s...
Mao-Jiang Zhang, P. Qiao, An Chen et al.· Structural Health Monitoring· 0 citations
Aiming at the challenge of signal detection and feature extraction in complex electromagnetic environment, this paper proposes an adaptive joint optimization method of signal detection and feature. Faced with the problems of traditional methods relying on prior knowledge, insufficient deep learning generalization abili...
Cheng Yang· Journal of Computing and Ele...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.