PSI - Issue 72

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

232

where N is the number of measurement points (discrete samples); x i is a separate measurement point with the number i; µ is the average value of all measurement points; x max and x min are the maximum and minimum strain values in the sample under study.

Fig. 2. Visualization of noise in the readings of distributed fiber-optic sensors based on Rayleigh scattering.

The statistical analysis of the collected data shows that the average signal amplitude range is 11.4 µε along the fiber length and 17.6 µε over time. The average standard deviation (σ) is 2.1 µε for length and 2.6 µε for time. For a signal following a normal distribution, 95% of measurements should fall within ±2σ, meaning they lie within the range of ±4.2 µε along the spatial coordinate and ±5.2 µε along the time coordinate. The results of statistical characteristics of noise for DFOS are presented in Table 1. Also, similar analysis for FBG strain sensors is provided from previous authors work (Serovaev et al. (2024))

Table 1. Statistical characteristics of measured strain data for studied FOS DFOS, length DFOS, time

FBG

Std (σ), µε

2.1

2.6

0.9

Min, µε

-5.3

− 8.6

− 4.1

Max, µε

6.0

8.9

4.2

R, µε

11.4

17.6

8.3

Many applications require measuring small strain values with high accuracy that are close to or below the noise threshold of distributed Rayleigh scattering FOS. One of the most effective and straightforward methods for reducing noise in signal processing is the application of uniform or moving average filters. This study demonstrates the effectiveness of such filters in minimizing noise in the readings of distributed FOS based on Rayleigh scattering. The moving average filter is based on averaging a certain number of discrete samples of the input signal and can be expressed as: