Research on adaptive noise suppression methods in high-precision OFDR strain measurement

Le résumé fourni par la source

In this study, we proposed and experimentally validated a high-precision strain measurement method for optical frequency domain reflectometry (OFDR) based on enhanced adaptive median filtering method (AMF). As the OFDR system enhances measurement spatial resolution, the correlation between the reference signal and the test signal diminishes. This reduction in correlation can result in multiple peaks or ghost peaks appearing in cross-correlation results. Furthermore, as the distance of distributed fiber optic sensing increases, both random noise and ghost peaks become more pronounced, significantly impacting system performance. To mitigate the effects of random noise and ghost peaks while enhancing system performance and accuracy, the AMF method is employed to address these challenges. This approach effectively eliminates random noise and ghost peaks, improves spatial resolution, and facilitates more accurate measurements. Compared to traditional denoising methods such as wavelet denoising (WD), Gaussian filtering (GF), and median filtering (MF), AMF dynamically adjusts window size through multiple iterations of adaptive median filtering combined with residual analysis and various filtering operations utilizing different window sizes. Consequently, the filtering effect can be tailored according to noise intensity, demonstrating robust noise suppression capabilities. The experimental results show that this method accurately identifies spatial resolution strain gradient information of 3mm and 1.5mm along the sensi ng fiber at a distance of 52m at a wavelength scanning rate of 20 nm/s. In addition, this method significantly reduces system errors, enhances the signal-to-noise ratio (SNR), and improves overall system stability. At a spatial resolution of 3 mm, the mean SNR achieved by this method is 42.97dB, with a corresponding mean standard deviation reduced to 2.46με. At a spatial resolution of 1.5 mm, the mean SNR increases by 30.18dB compared to the RAW data value of 6.11dB, while the mean standard deviation decreases markedly from 163.62με in RAW data to just 5.68με. This work provides new insights into remote distributed strain sensing aimed at improving spatial resolution within OFDR systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.