Pattern recognition based on enhanced multifeature parameters for vibration events in φ‐OTDR distributed optical fiber sensing system

Abstract

from the manufacturer's datasheet values are considered as expected and are attributed to the fact that the measurements of the RFID ASIC were conducted on the terminal pads of the strap, which included conductive adhesive. Finally, the input impedance measurement of an RFID tag antenna onto 3 different material surfaces was conducted and the return loss was extracted for every case. The results, are evaluated largely as expected, as the input impedance of the tag antenna was measured and characterized close enough to the conjugate impedance of the RFID strap for the first 2 cases, whereas a large degradation was found for the third case measurement of the filled with water glass container. These results validate the importance of the work of antenna designers and justify the development of application-specific RFID tags when they are used with different objects. Moreover, the presented results and methodology can be a useful toolkit for the RFID design and test engineer, and of particular usefulness as a guide for accurate measurements using imperfect test fixtures. ACKNOWLEDGMENTThe authors want to thank Professor J. Sahalos for the provision of RFID tags and his valuable help and comments for this work. ORCID D. E. Anagnostou

Use AF-CNN for End-to-End Fiber Vibration Signal Recognition

“…Subsequently, Xu et al used spectral subtraction to reduce broadband background noise to enhance vibration signals' time-frequency features. The denoising effect is better than the wavelet method [10]. In the endpoint detection stage, Wang et al proposed a method based on the threshold crossing rate to detect whether a vibration signal is generated.…”

“…At present, the effective features are the energy ratio, kurtosis, skewness, and spectral entropy of the signal, etc. The final classification and recognition usually use the SVM classifier [10]- [17].…”

Multi-Class Disturbance Events Recognition Based on EMD and XGBoost in φ-OTDR

“…In 2015, a method based on morphological feature extraction of time-space domain signals and relevance vector machine (RVM) classifier was reported [13]. The average identification rate of three events reached up to 97.8%, but the recognition time was 0.7 s. In 2017, Xu et al [14] used spectral subtraction to reduce wide-band background noise of signals and support vector machine (SVM) to detect four disturbance events (taping, striking, shaking, and crushing). The average identification rate was 93.8% and identification time was below 0.6 s. Subsequently, they reported a pattern recognition method based on convolution neural network (CNN) and SVM in 2018 [15].…”

Research of the Vibration Source Tracking in Phase-Sensitive Optical Time-Domain Reflectometry Signals Based by Image Processing Method

“…Unlike traditional monitoring, the Φ-OTDR technique regards the optical fiber as an organic whole in the monitoring process. In other words, the optical fiber is considered as a single vibration signal appearing at multiple points on the same line [23][24][25]. Owing to the distributed feature, the vibration signal of the Φ-OTDR contains three kinds of information-amplitude, time and length-in which the amplitude varies with time and length.…”