Contents lists available at ScienceDirect
Optical Fiber Technology
journa l homepa ge: www.elsevier.com/locate/yofte
Multi-peak detection algorithm based on the Hilbert transform for optical
FBG sensing
Fang Liu, Xinglin Tong
⁎
, Cui Zhang, Chengwei Deng, Qiao Xiong, Zhiyuan Zheng, Pengfei Wang
National Engineering Laboratory for Fiber Optic Sensing Technology, Wuhan University of Technology, 430070, PR China
ARTICLE INFO
Keywords:
Optical fiber sensing
FBG
Spectrum analysis
Peak detection algorithm
Hilbert transform
WDM system
ABSTRACT
A multi-peak detection algorithm based on the Hilbert Transform is proposed to overcome the shortage of
traditional single peak detection algorithms, which can extract the Bragg wavelengths from the reflection
spectrum in WDM systems. This algorithm introduces the thought of Divide and Conquer, including splitting the
reflection spectrum and single peak detection. The multi-peaks can be pre-positioned by splitting the input
reflection spectrum after the Hilbert Transform and derivation. Besides, theoretical analysis and the experi-
mental results show that this algorithm using zero as threshold of splitter can greatly enhance the algorithmic
portability under different dimensions as well as improve demodulation speed and accuracy. This implies that
the proposed algorithm provides a precise demodulation algorithm for the distributed FBG sensor network.
1. Introduction
Fiber Bragg gratings (FBGs) sensors, as a vital branch of optical fiber
sensors, have been widely used in various engineering monitoring areas
for nearly 30 years, such as civil engineering, petrochemical industry,
aerospace industry, and so on [1–3] . More conventional electric sensors
have been gradually substituted by FBG sensors, mainly because FBG
sensors not only have the common advantages of optical fiber sensors,
such as intrinsic safety, immunity to electromagnetic interference, high
resistance to corrosion, adaptability to harsh environments, high tem-
perature resistance, remote measurement, small size, and so on, but
they also have the unique advantages of simple optical properties and
good linearity, which make the sensor system easy to multiplex [4,5].
Thousands of FBGs can be written in a fiber and be located through the
Wavelength-Division-Multiplexing (WDM) technique [6,7], so a dis-
tributed FBG sensing network has the ability to monitor the online
health of large structures. The general principle of distributed FBG
sensing demodulation systems is that the relative shift of the central
wavelength reflected by each FBG changes with the monitored physical
parameters, such as temperature, strain, and others [8]. Therefore, the
multi-peak detection algorithm must be designed to accurately obtain
each Bragg wavelength from the reflection spectrum, which helps to
improve the accuracy of the measurement system. Although there are
many kinds of single peak detection algorithms, such as direct peak
(DP), centroid detection, polynomial fitting, Gaussian nonlinear fitting,
and so on [9], they are mainly adapted to detect the reflection spectrum
of single FBG. Because the reflection spectrum of a WDM system
contains more than one peak, none of the above algorithms can work
directly in WDM systems.
Aiming at solving this problem, we must make adjustments based on
the traditional single peak detection algorithms to dynamically adapt to
different applications and to maintain the accuracy of the measurement
process. In this paper, a multi-peak detection algorithm based on the
Hilbert transform is proposed to solve the above problem. The proposed
algorithm introduces the thought of Divide and Conquer, which can
accurately detect all Bragg wavelengths from the input reflection
spectrum of multi-FBGs. The algorithm mainly consists of two sections.
The first section splits the reflection spectrum of multi-FBGs and pre-
positions multi-peaks after the Hilbert transform and derivation. The
second section chooses one suitable single peak detection algorithm to
extract the Bragg wavelength from each split spectrum. Finally, the
monitored physical parameter of each point can be demodulated
through the Bragg wavelength without de-noising.
2. Theory
2.1. Theoretical FBG model
FBG is a typical reflective grating, equivalent to a narrow-band re-
flector. When the incident light in the fiber travels through the FBG, it
will reflect the light that meets the Bragg phase condition, and reflec-
tion is weak to the light that doesn't meet the Bragg phase condition
[10]. From the perspective of the reflection spectrum, FBG plays a role
as a wavelength selector. FBG demodulation systems are mostly
https://doi.org/10.1016/j.yofte.2018.06.003
Received 9 April 2018; Received in revised form 11 May 2018; Accepted 1 June 2018
⁎
Corresponding author.
E-mail address: tongxinglin@whut.edu.cn (X. Tong).
Optical Fiber Technology 45 (2018) 47–52
1068-5200/ © 2018 Elsevier Inc. All rights reserved.
T