236
G. Bai et al.
has been used in aviation engines, supersonic aircraft, satellite orbital control engines,
and it is considered to be one of the most promising thermal structural materials for the
future [6, 7]. However, in addition to being subjected to continuous stress, CMC structures also need to be exposed to high temperatures and other environmental factors,
so the damage and degradation processes of CMC are relatively complicated. Therefore, these processes are currently the subject of much research [8–10]. Revealing the
damage mechanism and evolution of the material characteristics, and establishing
the physical mechanism of the damage process are essential, since knowledge of
these informs the basis of structural design, life prediction and application safety
evaluation. In recent years, many researchers have conducted extensive research on
the performance of CMC composite materials under service loading and the damage
mechanism, and used this to predict cumulative damage, using fracture analysis, and
stress/strain analysis [11, 12]. However, these cannot reveal the damage evolution
process, while maintaining continuous monitoring of the source of the damage, and
accurately record the complete process of material failure.
Acoustic emission (AE) arises from changes in the internal structure of a material
leading to sources of local stress concentration. This energy is released rapidly in the
form of an elastic wave. Different types of damage will give rise to different sources
that produce distinct AE signals, and the basis of the AE technique is to distinguish the
different signals as an indicator of the different types of damage within the material.
AE technology has been widely applied to non-destructive testing, due to its real
time continuous analytical capability. It is ideal for analyzing the thermal creep
damage mechanism of C/SiC. In order to obtain as much information as possible,
a large number of characteristic parameters are recorded. However, if the number
of features is too high, the error rate for pattern recognition increases, which would
reduce the effectiveness of the technique. Therefore, it is necessary to select and
optimize these parameters. During pattern recognition, it is advantageous to use
unsupervised identification algorithm in order to obtain known category samples, so
that a clustering algorithm for AE signal analysis can be used.
It is known from previous studies and SEM analysis that damage to CMC can be
mainly classified into three types: fiber breakage, matrix cracking, and fiber/matrix
interface debonding. Momon [13] classified AE data from CMC fatigue testing by
using unsupervised and supervised techniques, Racle [14] used mechanical parameters to investigate CMC damage factors, and Awaja [15] used a combination of fourpoint bending and X-ray radiography to study C/SiC damage. However, these studies
did not analyze the time evolution of the damage how the materials eventually failed.
Traditional AE pattern recognition methods, mostly based on the methods used to
analyze metals are not well suited to providing CMC clustering analysis parameters.
This paper reports a method of determining a set of AE parameters suitable for
C/SiC creep damage analysis by extraction from an AE signal, and establishes a
relationship between the material damage mechanism and the AE signal. As a result,
information is obtained on the dynamic evolution and Creep life of the damage
process through cluster analysis.
G. Bai et al.
has been used in aviation engines, supersonic aircraft, satellite orbital control engines,
and it is considered to be one of the most promising thermal structural materials for the
future [6, 7]. However, in addition to being subjected to continuous stress, CMC structures also need to be exposed to high temperatures and other environmental factors,
so the damage and degradation processes of CMC are relatively complicated. Therefore, these processes are currently the subject of much research [8–10]. Revealing the
damage mechanism and evolution of the material characteristics, and establishing
the physical mechanism of the damage process are essential, since knowledge of
these informs the basis of structural design, life prediction and application safety
evaluation. In recent years, many researchers have conducted extensive research on
the performance of CMC composite materials under service loading and the damage
mechanism, and used this to predict cumulative damage, using fracture analysis, and
stress/strain analysis [11, 12]. However, these cannot reveal the damage evolution
process, while maintaining continuous monitoring of the source of the damage, and
accurately record the complete process of material failure.
Acoustic emission (AE) arises from changes in the internal structure of a material
leading to sources of local stress concentration. This energy is released rapidly in the
form of an elastic wave. Different types of damage will give rise to different sources
that produce distinct AE signals, and the basis of the AE technique is to distinguish the
different signals as an indicator of the different types of damage within the material.
AE technology has been widely applied to non-destructive testing, due to its real
time continuous analytical capability. It is ideal for analyzing the thermal creep
damage mechanism of C/SiC. In order to obtain as much information as possible,
a large number of characteristic parameters are recorded. However, if the number
of features is too high, the error rate for pattern recognition increases, which would
reduce the effectiveness of the technique. Therefore, it is necessary to select and
optimize these parameters. During pattern recognition, it is advantageous to use
unsupervised identification algorithm in order to obtain known category samples, so
that a clustering algorithm for AE signal analysis can be used.
It is known from previous studies and SEM analysis that damage to CMC can be
mainly classified into three types: fiber breakage, matrix cracking, and fiber/matrix
interface debonding. Momon [13] classified AE data from CMC fatigue testing by
using unsupervised and supervised techniques, Racle [14] used mechanical parameters to investigate CMC damage factors, and Awaja [15] used a combination of fourpoint bending and X-ray radiography to study C/SiC damage. However, these studies
did not analyze the time evolution of the damage how the materials eventually failed.
Traditional AE pattern recognition methods, mostly based on the methods used to
analyze metals are not well suited to providing CMC clustering analysis parameters.
This paper reports a method of determining a set of AE parameters suitable for
C/SiC creep damage analysis by extraction from an AE signal, and establishes a
relationship between the material damage mechanism and the AE signal. As a result,
information is obtained on the dynamic evolution and Creep life of the damage
process through cluster analysis.