A Novel Distributed Vibration Sensor Based on Fading Noise Reduction in Multi-Mode Fiber

Lidong Lü, Xingchen Su, Chenglong Zhang, Qinghao Gao, Hongwei YangSensors 2022 · DOI

TL;DR

This paper introduces a new distributed vibration sensing system using multi-mode fiber (MMF) that leverages vibration events to actively reduce noise in optical time domain reflectometry (OTDR) traces. By averaging data around vibration points, the system can accurately locate and analyze vibration spectra, offering a lower false alarm rate and improved efficiency compared to existing methods.

What Problem Does This Paper Address?

Optical fiber sensors are excellent for distributed monitoring, but accurately locating and identifying vibration events, especially in power infrastructure, remains a challenge. Traditional methods like trace subtraction can be prone to false alarms due to the inherent randomness of Rayleigh scattering in optical fibers, particularly in multi-mode fibers (MMF) which are already widely deployed. Existing techniques for vibration spectrum analysis can be computationally intensive, requiring specialized hardware or long processing times. This paper aims to develop a more robust and efficient distributed vibration sensing system that utilizes the existing MMF infrastructure and mitigates noise issues.

Methods

The core of this novel approach lies in its unique use of vibration events themselves to improve signal quality. Instead of treating vibration as a purely external disturbance to be filtered out, the authors propose that the vibration event itself diversifies the optical polarization state and phase of the Rayleigh scattering light. This diversification is then exploited through a data averaging technique.

Specifically, the system uses a polarization-sensitive OTDR setup. When a vibration occurs, the backscattered Rayleigh light is averaged over a specific region around the vibration point. This averaging process smooths out the "fading noise" (random fluctuations in the OTDR trace) and makes the trace converge rapidly after the vibration point, clearly delineating the event location.

For spectrum analysis, data within a defined range around the vibration event is further averaged to improve the signal-to-noise ratio (SNR). This averaged data is then fed into a Fast Fourier Transform (FFT) algorithm to extract the vibration spectrum. This targeted averaging approach is presented as more efficient than analyzing every single point along the fiber.

Datasets

Not discussed in the paper. The experiments were conducted using a custom-built experimental setup with a 10 km length of multi-mode fiber.

Key Insights and Ideas

  • Vibration as a Noise Reducer: The most striking insight is the idea that vibration events can actively reduce noise by diversifying the optical polarization state and phase of Rayleigh scattering. This flips the conventional approach of trying to eliminate noise.
  • MMF Advantage: The system is designed to work with multi-mode fiber (MMF), which is already prevalent in many industrial applications (like power cables). This avoids the need for costly infrastructure upgrades to single-mode fiber.
  • Targeted Averaging for Efficiency: By averaging data specifically around the detected vibration event, the system achieves a good SNR for spectrum analysis without needing to process the entire fiber length exhaustively. This leads to faster event identification and spectrum extraction.
  • Lower False Alarm Rate: The averaging method, by smoothing out random fluctuations, is expected to significantly reduce false alarms compared to methods that are highly sensitive to polarization variations.

Key Results

  • Accurate Vibration Location: The experimental setup successfully located vibration events at 5.167 km and 10.145 km along the MMF. The averaged OTDR traces showed a clear convergence after the vibration point, making location straightforward.
  • Effective Spectrum Extraction: The FFT analysis of the averaged data successfully extracted the vibration spectrum, with measured frequencies (e.g., 7.8 Hz and 4.688 Hz) closely matching the applied vibration frequencies.
  • Reduced Fading Noise: Figures demonstrate a significant smoothing of the OTDR trace after the vibration event due to the averaging method, indicating effective reduction of fading noise in that region.
  • Time Savings: The proposed method is claimed to be more time-efficient than analyzing vibration spectra for all positions along the fiber.

Limitations and Future Work

  • Dynamic Measurement Performance: The paper acknowledges that the trace averaging method, while reducing noise, negatively affects the dynamic measurement performance, specifically the frequency response range. This is a trade-off for improved SNR and reduced processing time.
  • MMF Specifics: The authors highlight that the randomness of optical polarization state and phase in MMF necessitates data averaging for spectrum analysis, a step not as critical in conventional polarization-sensitive OTDR systems using single-mode fiber.
  • Future Work: The paper suggests that this DVS system can be integrated with existing distributed temperature sensing (DTS) systems, enabling simultaneous monitoring of both temperature and vibration events. Further applications are envisioned for monitoring oil pipelines and power cables.

Summary updated

Abstract

Multi-mode fiber (MMF) is used in a polarization-sensitive optical time domain reflectometer (OTDR) for vibration event location and spectrum analysis. The vibration events acting on MMF are considered to be the optical polarization state and phase diversifying process for fading noise reduction. In addition, data averaging with continuous positions and the fast Fourier transform (FFT) method is proposed to extract the spectrum of the vibration events. In the experiment, the vibration events are loaded at the positions of 5.167 and 10.145 km, respectively, along MMF. The experimental results demonstrate that the vibration event can effectively diversify the optical polarization state and phase of the Rayleigh scattering light to make the averaged OTDR trace behind the vibration position converge rapidly, which helps to locate corresponding vibration events and extract the vibration spectrum. It is inferred that the new distributed vibration sensor shall have a lower false alarm rate, as it can greatly reduce the errors caused by randomness of the sensing light signals. Additionally, it also saves time in comparison with the method that analyzes the vibration spectra for all the positions along the fiber under test.
BibTeX
@inproceedings{lu2022novel,
  author = {Lidong Lü and Xingchen Su and Chenglong Zhang and Qinghao Gao and Hongwei Yang},
  title = {{A Novel Distributed Vibration Sensor Based on Fading Noise Reduction in Multi-Mode Fiber}},
  booktitle = {Sensors},
  year = {2022},
  doi = {10.3390/s22208028},
}

Loading related content...

Figure