Recognition of Acoustic Effective Relaxational Lines
for Gas Detection Using Minimum Distance
Clustering
Kesheng Zhang*, Xiaoxue Guo, Wensheng Hu
School of Electrical and Information Engineering, Guizhou
Institute of Technology
Guiyang, China, 550003.
Email: keshengzhang@163.com
Weihua Ou
School of Big Data and Computer Science, Guizhou Normal
University,
Guiyang, China, 550001.
Email: ouweihuahust@gmail.com
Abstract—Acoustic propagation characteristics, i.e., the
frequency-dependent sound speed and absorption depend upon
the composition of the gas. The position of an acoustic absorption
peak can be synthesized by the measured acoustic propagation
characteristics at two frequencies. In this paper, we propose a
sensing method based on the minimum distance clustering to
recognize acoustic effective relaxational Lines, which are
composed of acoustic absorption spectral peaks, for gas
concentration detection. Simulation results demonstrate that the
proposed method can extract the information of gas
concentration and environmental temperature from the
recognized acoustic effective relaxational Lines.
Keywords—Minimum distance clustering; Acoustic Effective
Relaxational Lines; Gas Detection; Recognition Analysis
I.
I
NTRODUCTION
Quantitative detection of specific gas concentration in a
two-component mixture has important applications in many
industrial applications such as chemical processing and
handling industry, or for the real-time monitoring of natural gas
in a given environment [1]. The propagation characteristics (i.e.,
acoustic speed and acoustic absorption) of acoustic waves in
gases are inextricably related to gas composition, temperature,
and pressure [2,3]. This enables the use of acoustic waves as a
quantitative method of the thermodynamic and molecular
properties of gas mixtures [4,5,6].
A sound wave passing through an excitable (polyatomic or
diatomic) gas will continuously exchange its mechanical
energy with the kinetic energy of the constituent molecules of
the gas by alternating compressions and expansions. The
tendency toward equipartition of excess acoustic energy among
all the molecular modes is a phenomenon called thermal
relaxation [2]. The thermal relaxation process can generate
frequency-dependent relaxational absorption [3]. Thus, the
quantitative acoustic relaxational spectra can be used to acquire
the concentration of contaminant gases in a known base gas by
analyzing the differences between the two acoustic absorption
spectra. In our previous works [2], we present a physical model
to predict the relaxational absorption spectra and develop an
algorithm to reconstruct the spectral peak in a given excitable
gas, which can be used to capture the primary relaxation
process, by only measuring the sound absorption and sound
speed of two operating frequencies [7]. Based on the algorithm,
we also propose the methods to employ the relaxational lines
formed by spectral peak points for gas composition detection
[8,9]. In this paper, we demonstrate how to employ minimum
distance clustering to recognize the relaxational lines in two-
component gas mixtures for real-time gas concentration
detection.
This paper is organized as follows. In Section II we
introduce the physical background about how to synthesize the
acoustic-absorption spectral peak at two fixed frequencies and
construct the acoustic effective relaxational lines. In Section III,
we prose a maximum value judgment method of the peak value
of the relaxation absorption spectrum to recognize of acoustic
effective relaxational lines for gas detection. In Section IV,
simulation results in the mixtures of CO
2
–N
2
are presented to
validate our method. Section V concludes this paper.
II. P
HYSICAL
B
ACKGROUND
A. Synthesizing Algorithm for Position of Acoustic Absorption
Peaks
The presence of excitable gases is manifested by
relaxational peaks in the frequent-dependent acoustic
absorption spectra [10]. The relaxation time and relaxation
strength determine the characteristics of a molecular relaxation
process, which makes acoustic absorption become frequent-
dependent spectrum. Then, the position of the acoustic
absorption peak along the abscissa (i.e., the relaxation
frequency) is determined by the relaxation time and the height
of the absorption maximum by the relaxation strength [7]. Thus,
capturing the acoustic absorption peak can reconstruct the
acoustic absorption spectrum due to a molecular single-
relaxation process in the whole frequency domain.
With measuring the relaxational absorption coefficient
r
α
and the speeds of sound
c
at two selected frequencies
1
and
2
, the maximum absorption
m
and the relaxation
978-1-5386-3016-7/17/$31.00 ©2017 IEEE 215
2017 International Conference on Security, Pattern Analysis,
and Cybernetics (SPAC)