186
Y. Zhang et al.
17.1 Introduction
As a typical ceramic matrix composite, continuous fiber reinforced silicon carbide
(C/SiC) has excellent performance of high specific strength, corrosion resistance,
oxidation resistance and high temperature resistance [1]. It has broad prospects in
the application of thermal structures such as aeroengine, gas turbine, launch vehicle
thermal protection system, and is an indispensable material for the development of
aerospace and other high-tech fields [1–6]. Acoustic emission (AE) technology can
monitor the development of material internal damage in real time and directly reflect
the process of material damage generation and expansion. However, the AE signal
of material damage is complex, and the parameters of the AE signal of damage types
are not obvious, so it is difficult to directly associate the AE signal with the damage
type [6, 7].
Previous research results show that the method of unsupervised clustering can
identify the damage types of materials [8–12]. The damage modes of the composite
under high temperature (700–1200 °C) were obtained by cluster analysis method by
Momons et al. [9]; Tong et al. studied the AE signal parameters of C/SiC composite
under room temperature tensile test by using k-means clustering algorithm [11].
Huang Xipeng, Wang Bo, etc. used the energy and amplitude of AE signals after
wavelet denoising to carry out pattern recognition through K-means clustering analysis method, and studied the damage modes of 3D-N C/SiC and C/SiC composites
with different densities under tensile load [11, 12]. At present, the unsupervised
analysis of AE signals of tensile damage of C/SiC composite is mainly based on
K-means algorithm and unsupervised clustering analysis of some characteristics of
AE signal directly selected by researchers [9–12]. K-means clustering algorithm
is a classical algorithm in cluster analysis, which has the advantages of simplicity
and rapidity. However, K-means algorithm is a local search technology, which may
converge prematurely due to the influence of initial clustering center [13]. At the
same time, less AE signal characteristics can’t reflect the AE events of materials
completely. There is no an improved k-means algorithm to analyze the AE signal of
C/SiC damage mode.
In this study, the genetic algorithm proposed by Holland is introduced to solve
the problem of K-means algorithm in the clustering analysis of AE signals of 2D
C/SiC tensile damage. The algorithm simulates the genetic and evolutionary process
of biological species in nature, and it has strong robustness and global optimization
ability [14]. Through the optimization of the common clustering algorithm for AE
data processing in the tensile test of 2D C/SiC samples at room temperature, the
more reliable clustering analysis results are obtained. At the same time, through the
distribution of all kinds of damage types energy and the growth trend of accumulated
energy in the process of tensile test, combined with SEM observation of sample fracture, the identification of damage types in the process of tensile loading is completed,
which provides the basis for damage detection in the engineering application of 2D
C/SiC composite materials.
Y. Zhang et al.
17.1 Introduction
As a typical ceramic matrix composite, continuous fiber reinforced silicon carbide
(C/SiC) has excellent performance of high specific strength, corrosion resistance,
oxidation resistance and high temperature resistance [1]. It has broad prospects in
the application of thermal structures such as aeroengine, gas turbine, launch vehicle
thermal protection system, and is an indispensable material for the development of
aerospace and other high-tech fields [1–6]. Acoustic emission (AE) technology can
monitor the development of material internal damage in real time and directly reflect
the process of material damage generation and expansion. However, the AE signal
of material damage is complex, and the parameters of the AE signal of damage types
are not obvious, so it is difficult to directly associate the AE signal with the damage
type [6, 7].
Previous research results show that the method of unsupervised clustering can
identify the damage types of materials [8–12]. The damage modes of the composite
under high temperature (700–1200 °C) were obtained by cluster analysis method by
Momons et al. [9]; Tong et al. studied the AE signal parameters of C/SiC composite
under room temperature tensile test by using k-means clustering algorithm [11].
Huang Xipeng, Wang Bo, etc. used the energy and amplitude of AE signals after
wavelet denoising to carry out pattern recognition through K-means clustering analysis method, and studied the damage modes of 3D-N C/SiC and C/SiC composites
with different densities under tensile load [11, 12]. At present, the unsupervised
analysis of AE signals of tensile damage of C/SiC composite is mainly based on
K-means algorithm and unsupervised clustering analysis of some characteristics of
AE signal directly selected by researchers [9–12]. K-means clustering algorithm
is a classical algorithm in cluster analysis, which has the advantages of simplicity
and rapidity. However, K-means algorithm is a local search technology, which may
converge prematurely due to the influence of initial clustering center [13]. At the
same time, less AE signal characteristics can’t reflect the AE events of materials
completely. There is no an improved k-means algorithm to analyze the AE signal of
C/SiC damage mode.
In this study, the genetic algorithm proposed by Holland is introduced to solve
the problem of K-means algorithm in the clustering analysis of AE signals of 2D
C/SiC tensile damage. The algorithm simulates the genetic and evolutionary process
of biological species in nature, and it has strong robustness and global optimization
ability [14]. Through the optimization of the common clustering algorithm for AE
data processing in the tensile test of 2D C/SiC samples at room temperature, the
more reliable clustering analysis results are obtained. At the same time, through the
distribution of all kinds of damage types energy and the growth trend of accumulated
energy in the process of tensile test, combined with SEM observation of sample fracture, the identification of damage types in the process of tensile loading is completed,
which provides the basis for damage detection in the engineering application of 2D
C/SiC composite materials.