PSI - Issue 72

V.P. Matveenko et al. / Procedia Structural Integrity 72 (2025) 229–234

230

reflectometry (OFDR) stands out for its superior spatial resolution in the millimeter range and high strain sensitivity, making it particularly effective for short- to medium-range sensing applications (Lv et al. (2024)). However, achieving consistent and accurate strain measurements with OFDR remains challenging. Rayleigh backscatter signals are inherently affected by intensity and phase noise, amplitude distortions, and random signal dropouts, all of which complicate spectral correlation analysis leading to outliers, missing values, spatial and temporal noise in the strain distribution data (Lobach et al. (2024), Richter et al. (2024), Luo et al. (2021)). Even in environments with minimal external mechanical influences, these noise sources can obscure subtle strain variations or introduce false anomalies (Qu et al. (2022)). To address these challenges, researchers have pursued various strategies. Optimizing fiber coatings and adhesives has been shown to mitigate strain-transfer mismatches (Du et al. (2023)), while advanced data-processing techniques have been developed to compensate for measurement artifacts. Some of the effective approaches include specialized outlier detection (Richter et al. (2024)), local feature extraction (Lv et al. (2024)), and wavelet-based denoising (Li et al. (2021)). Additionally, machine learning-driven techniques have demonstrated potential for real-time noise reduction and anomaly detection (Karapanagiotis et al. (2023)). Nevertheless, balancing the trade-offs between high spatial resolution, fast processing, and a broad strain measurement range remains an ongoing challenge (Qu et al. (2022)). This study evaluates the noise characteristics of an OFDR-based DFOS by analyzing strain measurements in an unloaded optical fiber under isothermal conditions. By eliminating external mechanical and thermal influences, the intrinsic noise behavior of the system is isolated. The study considers spatial and temporal noise variations in strain measurements and assesses the statistical dispersion of sensor readings. The effectiveness of noise reduction strategies, particularly moving average filtering, is evaluated in enhancing signal quality. The findings provide a quantitative basis for understanding noise limitations in DFOS and offer practical recommendations for improving data processing techniques to enhance measurement accuracy and reliability in engineering applications. 2. Materials and methods Distributed fiber-optic sensors (DFOS) based on Rayleigh scattering allow for precise strain measurements along an optical fiber. These sensors offer significant advantages due to utilization of fiber under test as sensing element without the need of preliminary influence on the optical fiber sections to induce strain and temperature sensitivity as with single-point FOS. Among different types of fiber optic sensors, DFOS based on Rayleigh scattering measurement with OFDR provide the highest spatial resolution, enabling strain measurements at thousands of points with a resolution of up to or even less than 1 mm. Fig. 1 illustrates a schematic representation of an optical fiber used for strain measurement. The scheme consists of an optical fiber (red line) along which measurement points ( S ₁, S ₂, ..., S ₙ ) are positioned.

Fig. 1. Schematic diagram of a distributed fiber-optic sensor.