Fading Noise Suppression Method of Φ-OTDR System Based on GA-VMD Algorithm

Tian-Xiong Li,Fu-Dong Zhang,Jun Lin,Xing-Ye Bai,Hao-Zhuang Liu
DOI: https://doi.org/10.1109/jsen.2023.3306199
IF: 4.3
2023-01-01
IEEE Sensors Journal
Abstract:A distributed optical fiber acoustic sensing system (DAS) based on phase-sensitive optical time domain reflectometry ( $\Phi $ -OTDR) can detect the vibration information along an optical fiber and has been widely used for various engineering detections; however, the high-coherence detection light source used in a DAS will cause coherent fading in the fiber, leading to system fading noise and affecting the monitoring of the target vibration events. In order to solve the problem of the fading noise in the DAS, this study considered a method of suppressing the fading noise based on the genetic algorithm (GA) and variational mode decomposition (VMD). Furthermore, the parameter selection problem in the VMD method is improved by introducing the GA. By introducing sample entropy as a fitness function and combining it with the GA, this method adaptively searches for the optimal decomposition parameters by iterating the minimum value of the fitness function according to the type of input signal. Then, it decomposes the noisy DAS data into multiple intrinsic mode functions (IMFs), and the permutation entropy and time-frequency domain information of each component were used to extract the effective IMF components for signal reconstruction to obtain the denoised signal. Experimental results show that compared with empirical mode decomposition (EMD), ensemble EMD (EEMD), and complete EEMD (CEEMD) methods, the proposed method can eliminate the mode aliasing problem, effectively suppress the fading noise of the DAS, improve the signal-to-noise ratio, and reduce the root-mean-square error. It is, thus, possible to more accurately monitor vibration events, improving the practical application value of the DAS.
engineering, electrical & electronic,instruments & instrumentation,physics, applied
What problem does this paper attempt to address?