1956 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 26, NO. 19, OCTOBER 1, 2014
Configurable Filter-Based Endpoint Detection
in DMZI Vibration System
Xiangdong Huang, Jia Yu, Kun Liu, Tiegen Liu, and Qinnan Chen
Abstract—In order to accurately determine the vibration
starting point of a disturbance frame in the dual Mach–Zehnder
interferometry vibration system, this letter proposes a novel
endpoint detection method based on convolution window filtering.
Through analyzing the fast Fourier transform spectrum of the
undisturbed noise, we can acquire the cutoff frequency parameter
to configure one high-pass filter to remove the noise, which
mainly consists of low-frequency components. Then, the time
instant corresponding to the vibration starting point can be
detected through applying a thresholding search to the output
of the high-pass filter. Experiments show that the proposed
detection method outperforms the wavelet-based method. Hence,
the proposed detection method has wide application values for
its high precision, flexibility, and reliability.
Index Terms—Endpoint detection, configurable high-pass
filter, convolution window, thresholding search.
I. INTRODUCTION
N
OWADAYS, several distributed optical fiber sensors, such
as Dual Mach–Zehnder Interferometry (DMZI) vibration
sensor [1], [2], sagnac interferometry vibration sensor and the
sensor based on optical time-domain reflectometer (OTDR)
are widely applied in disturbance detection. Among these,
DMZI vibration sensor receives increasing attention, for it
has the superiority of high sensitivity and fast response. Thus
it is widely applied in external invasion event detection,
such as pipeline leakage detection [3], perimeter security [4],
submarine cable security [5] etc. Hence, determination of
these events’ sudden moments is of vital importance in
DMZI vibration system. As far as the collected vibration
samples are concerned, vibration starting moment actually
corresponds to some endpoint, i.e., the boundary between the
undisturbed noise and the disturbance signal. Furthermore,
endpoint detection is also the basis of other techniques in
DMZI system, such as event positioning [6], [7] and event dis-
crimination [8], which are all realized through processing the
samples following the detected endpoint. As [9] demonstrates,
endpoint detection also contributes to enhancing the weight
Manuscript received June 8, 2014; revised July 11, 2014; accepted
July 21, 2014. Date of publication July 29, 2014; date of current version
September 8, 2014. This work was supported in part by the National Basic
Research Program of China under Grant 2010CB327806 and in part by the
National Natural Science Foundation of China under Grant 61271069 and
Grant 61271322. (Corresponding author: Kun Liu.)
X. Huang and J. Yu are with the School of Electronic Informa-
tion Engineering, Tianjin University, Tianjin 300072, China (e-mail:
xdhuang@tju.edu.cn; jia7721927@126.com).
K. Liu, T. Liu, and Q. Chen are with the College of Precision Instrument
and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China
(e-mail: beiyangkl@tju.edu.cn; tgliu@tju.edu.cn; chenqinnan@gmail.com).
Color versions of one or more of the figures in this letter are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LPT.2014.2343274
of high-frequency components as well as attenuating the
magnitude of low-frequency main lobe, which contributes to
reduce position error. Thus, the accuracy of endpoint detection
will inevitably influence the position error and recognition rate
of events (such as climbing the fence, cable invasion).
In recent years, various endpoint detection algorithms,
such as cepstrum coefficients method [10], short-time zero
crossing ratio method [11], wavelet-based method [12], [13],
have been proposed [14]. However, cepstrum coefficients
method only applies to the occasion when multiplicative
noise instead of additive noise is involved, so it is only
applied in some special fields such as speech and radar
systems. As to the short-time zero crossing ratio method,
which is based on counting the numbers of zero crossing in
some fixed time blocks, it actually determines the position
where the highest frequency component occurs. In fact, there
exists some distance between this position and the endpoint.
It should be emphasized, since the wavelet can be easily
zoomed out by simply choosing some fine scale so that the
high-frequency detail information of the sudden event can
be extracted, the wavelet-based method is the mainstream
in endpoint detection. However, wavelet based method also
cannot distinguish undisturbed noise and disturbance signal
accurately, since wavelet decomposition can only split the
frequency band in half. Thus wavelet decomposition has to
be implemented in high levels to ensure that the observed
frequency band can approximate the boundary frequency
between the undisturbed noise and disturbance signal as
close as possible, which also introduces high computation
complexity and makes it difficult to detect the endpoint in
real-time.
In fact, since the core problem of endpoint detection
is to remove undisturbed noise, whose energy is mainly
distributed in low frequency band, it is essential to accu-
rately analyze the frequency band of the undisturbed noise.
So this letter puts forward one novel detection algorithm
based on high-pass convolution window FIR filter, whose
cutoff frequency is acquired from the FFT result of undis-
turbed noise. This letter will also interpret the principle of
convolution window based filter, whose parameters can be
configured flexibly to apply to different application envi-
ronments. In addition, comparing with the existing high-
pass digital filters such as Butterworth and Chebyshev filters,
the proposed filter is permanently stable. Experiments also
show that the proposed endpoint detection method outper-
forms the wavelet based method in accuracy, flexibility, and
reliability.
II. DMZI S
YSTEM
A DMZI vibration sensor is shown in Fig. 1: the output of
the laser with narrow linewidth is split equally at coupler C1
1041-1135 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
资源评论
weixin_38729399
- 粉丝: 7
- 资源: 902
最新资源
- 纯css3发光霓虹灯文字闪烁特效代码.zip
- 用VBS制作自己的进度条
- 电脑说话VBS什么电脑都能用
- 利用HTML+CSS+JS的国漫分享网站(响应式)
- 练习springboot1 项目 模拟高并发秒杀，实现基本的登录、查看商品列表、秒杀、下单等功能，简单实现了系统缓存、降级和限流
- 一个社区论坛项目，技术栈：spring boot + thymeleaf+Redis 实现的功能：发帖，关注，点赞，私信，系统通知，日活统计.zip
- 会员管理系统.zip-会员管理系统.zip
- 解压软件 ZArchiver.apk
- 《系统分析和设计》课程作业-面向中国各大城市的医院预约挂号系统.zip
- SM4学习备份，有用的
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
