Refractory material damage mode identification method based on acoustic emission technology

Abstract

The invention relates to the technical field of microcosmic damage mechanism recognition of refractory materials, and discloses a refractory material damage mode recognition method based on an acoustic emission technology, which comprises the following steps of: acquiring acoustic emission signals of a refractory material under the action of static load through a three-point bending experiment, determining the number and characteristic parameters of data set categories of the acoustic emission signals, and determining the number of the data set categories of the acoustic emission signals; and performing clustering analysis on the acoustic emission signals by adopting a K-means clustering algorithm, dividing the acoustic emission signals into three classes, analyzing by combining an acoustic emission parameter association diagram, determining acoustic emission characteristics corresponding to different damage mechanisms, and determining a damage mode corresponding to each class of signals. According to the method for identifying the damage mode of the refractory material based on the acoustic emission technology, a method combining clustering analysis and acoustic emission parameter association diagram analysis is adopted, accurate judgment on the microscopic damage process of the refractory material can be effectively achieved, and theoretical support is provided for further research on mechanical behaviors and the microscopic damage mechanism of the refractory material.

Description

Technical Field

[0001] The invention relates to the technical field of microscopic damage mechanism identification of refractory materials, in particular to a refractory damage pattern identification method based on acoustic emission technology. Background Art

[0002] Refractory materials are key materials for industrial production processes such as steel, glass, and ceramics that guarantee high-temperature equipment. This type of material is composed of high-temperature resistant inorganic aggregates, bonding phases, additives, etc. It has a complex microscopic heterogeneous structure and has strong high-temperature resistance. Under high-temperature conditions, refractory materials will suffer various microscopic damages, such as matrix cracking, interface debonding, and particle damage. At present, research in this field focuses on pattern recognition of microscopic damage to refractory materials. Accurate identification and analysis of these microscopic damage mechanisms are crucial to establishing a mechanical behavior model for refractory materials and predicting changes in material properties.

[0003] As a non-destructive detection method, acoustic emission technology can effectively monitor and analyze the microscopic damage process of refractory materials during stress. Through the application of acoustic emission technology, the characteristic signals of different damage stages can be distinguished, thereby realizing the identification of microscopic damage patterns of refractory materials. Through the traditional acoustic emission parameter correlation diagram analysis method, correlation analysis of any two variables can be performed to distinguish acoustic emission signals with different characteristics, such as amplitude, duration, energy, arrival time, root mean square voltage value, number of impacts, impact rate, external parameters and other acoustic emission signal change patterns.

[0004] In recent years, acoustic emission technology has been widely used in the field of micro damage of refractory materials. Studies have shown that the combination of acoustic emission technology and pulse echo technology can effectively monitor the destructive behavior of refractory materials at high temperatures; acoustic emission parameters can be used to monitor the crack formation in refractory materials, and different brittle refractories have different crack behaviors; using digital image correlation technology and acoustic emission methods, the mechanical properties and fracture behavior of refractory materials can be evaluated; acoustic emission technology can monitor the damage process of refractory materials during thermal cycles and determine the critical damage level; for magnesium refractory materials containing graphite, acoustic emission technology can also be used to detect their fracture behavior. However, since refractory materials are a kind of composite materials with a very complex microstructure, it is not reasonable to rely solely on acoustic emission technology as an analysis method.

[0005] In view of this, it is necessary to develop a refractory damage pattern recognition method based on acoustic emission technology and K-means clustering algorithm, a comprehensive clustering analysis and acoustic emission parameter correlation diagram analysis method, to effectively realize the acoustic emission signal pattern recognition of refractory damage in order to solve the above problems. Summary of the invention

[0006] (I) Technical solution

[0007] The present invention provides the following technical solution: a method for identifying damage patterns of refractory materials based on acoustic emission technology, comprising the following steps:

[0008] S1. Acoustic emission signals of refractory materials under static load are collected through three-point bending test to determine the number and characteristic parameters of data set categories of acoustic emission signals.

[0009] S2. Use K-means clustering algorithm to perform cluster analysis on acoustic emission signals and divide them into three categories.

[0010] S3. Combined with the analysis of the acoustic emission parameter correlation diagram, determine the acoustic emission characteristics corresponding to different damage mechanisms, and determine the damage pattern corresponding to each type of signal.

[0011] Preferably, in step S1, the refractory material is a magnesium-carbon refractory material, the basic size of the material is 125 mm×25 mm×25 mm, the three-point bending experiment is to perform acoustic emission monitoring on the bending process of the magnesium-carbon refractory material specimen until the specimen breaks, and the loading rate of the material testing machine is 0.25 N / (m 2 ·s).

[0012] Preferably, in step S1, among the acoustic emission signals, the amplitude of the acoustic emission signals accounting for 93.7% of all signals is concentrated in 50-75dB, the average frequency of the signals is concentrated in 21-60kHz, the characteristic parameters include rise time, peak count, count, energy, duration, amplitude and ASL, and the acoustic emission process of the material bending process can be divided into three stages, the damage incubation stage 21-33.66s, the damage accumulation development stage 33.66-62.14s and the instability failure stage 62.14-fracture.

[0013] Preferably, in step S2, the K-means clustering algorithm comprises the following steps:

[0014] S21. Select K initial cluster centers: The superscript indicates the number of iterations in the clustering process.

[0015] S22. Assume that r iterations have been performed. If for a sample x,

[0016]

[0017] but in, So The sample subset is the cluster center, and all samples are distributed in K clusters according to this method, that is, the principle of minimum spacing.

[0018] S23. After reclassification, calculate each cluster center:

[0019]

[0020] in, for The number of samples included in .

[0021] S24, if If j=1, 2, ..., K, the process ends; otherwise, the process goes to S22.

[0022] Preferably, in step S21, the K value is 3, corresponding to the three stages of refractory damage, namely, the damage initiation stage, the matrix damage stage and the interface damage stage. The K-means clustering algorithm divides the acoustic emission signals into three categories according to the characteristic attributes of amplitude, ringing count, rise time, duration and average frequency.

[0023] Preferably, in step S3, determining the damage mode corresponding to each type of signal comprises the following steps:

[0024] S31. Combined with the analysis of the acoustic emission parameter correlation diagram, the acoustic emission characteristics corresponding to different damage stages of refractory materials are identified to achieve pattern recognition of refractory material damage.

[0025] S32, taking out each classified parameter, and then matching it with time, to obtain the time history classification result of the acoustic emission signal.

[0026] S33. Analyze the characteristics of different types of acoustic emission signals, such as amplitude, ring count, rise time, duration, and average frequency, and infer the characteristics of different damage mechanisms corresponding to the acoustic emission signals.

[0027] S34. Combine theoretical analysis to explain the acoustic emission signal characteristics corresponding to different damage modes.

[0028] Preferably, in step S33, the different types of acoustic emission signals include three categories, namely, the first category, the second category and the third category. The first category of signals appears throughout the entire bending test process and is the largest category of all signals, accounting for 95.9% of the total signals. The amplitude of the signal is concentrated in 50-70dB, and the average frequency is concentrated in 30-60KHZ. Among the three categories of signals, the ringing count, rise time and duration of this type of signal are the smallest, while the average frequency is the highest.

[0029] The second type of signal accounts for 2.8% of the total signal. The amplitude and average frequency of the signal are concentrated in 80-90dB and 15-35KHZ respectively. However, the average frequency of this type of signal is the lowest. This type of signal does not appear until 33.67s and continues until the event breaks.

[0030] The third type of signal has the smallest proportion among the three types of signals, accounting for 1.3% of the total signal. The signal amplitude is concentrated in 95-100dB, which is the largest among the three types of signals. Such a signal began to appear at 62.17s and also lasted until the event broke.

[0031] Preferably, in step S34, the acoustic emission signals corresponding to different damage modes have different characteristics, while the acoustic emission characteristics at the same stage are similar, matrix damage accompanies the entire bending test process, interface damage will not occur until the matrix damage reaches a certain degree, and two of the three types of signals have similar acoustic emission characteristics and can be classified into one category.

[0032] (II) Beneficial effects

[0033] Compared with the prior art, the present invention provides a refractory damage pattern recognition method based on acoustic emission technology, which has the following beneficial effects:

[0034] 1. The refractory damage pattern recognition method based on acoustic emission technology can effectively identify the characteristic signals of different damage stages through cluster analysis of acoustic emission signals, so as to accurately judge the microscopic damage pattern of refractory materials. Cluster analysis can classify the acoustic emission signals collected during the material bending test into several categories. These signals can reflect the characteristics of the damage mechanism. Combined with the analysis of the acoustic emission parameter correlation diagram, the acoustic emission characteristics corresponding to the refractory materials at different damage stages can be identified, thereby realizing pattern recognition of refractory material damage. The damage process of the refractory materials studied in the present invention can be divided into the damage initiation stage, the matrix damage stage and the interface damage stage. The amplitude and ringing counter of the acoustic emission signal are used to identify the damage process of the refractory materials. The parameters such as number, rise time, duration and average frequency are preliminarily divided into three categories using the K-means clustering algorithm, and then combined with the acoustic emission correlation diagram, it can be concluded that the matrix damage accompanies the entire bending test process. Interface damage will not occur until the matrix damage reaches a certain level. The acoustic emission characteristics of two of the three types of signals are similar and can be classified into one category. The present invention provides an important basis for in-depth research on the destruction mechanism of refractory materials. The present invention can be used to improve the safety of refractory materials, prevent potential failures or accidents by timely identifying different damage stages, and can also help optimize material performance and quality control in the production process, improve material design and manufacturing technology, and reduce unnecessary losses and costs.

[0035] 2. The refractory damage pattern recognition method based on acoustic emission technology adopts a method combining cluster analysis and acoustic emission parameter correlation diagram analysis, which can effectively realize the accurate judgment of the microscopic damage process of refractory materials, provide theoretical support for further research on the mechanical behavior and microscopic damage mechanism of refractory materials, complete the pattern recognition of acoustic emission signals, and lay the foundation for establishing a mathematical model of microscopic damage of refractory materials and studying the microscopic damage mechanism of the mechanical behavior of refractory materials. At the same time, it can also provide tools for the field of materials science to deeply understand the microscopic damage mode, so as to improve the reliability of existing materials and promote the research and development of new refractory materials, with extremely high application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the experimental principle of the three-point bending experiment of the present invention;

[0037] Figure 2 It is a curve diagram of cumulative energy and cumulative ringing count time variation of the present invention;

[0038] Figure 3 It is a time history diagram of acoustic emission parameters and an acoustic emission parameter correlation diagram of the present invention;

[0039] Figure 4 This is a schematic diagram of the K-means algorithm flow chart of the present invention;

[0040] Figure 5 This is a schematic diagram of the duration-amplitude K-means clustering result of the present invention;

[0041] Figure 6 This is a schematic diagram of the time-amplitude clustering results of the present invention;

[0042] Figure 7 It is a schematic diagram of the acoustic emission signal characteristic values ​​(cluster centers) and classification results of the three types of damage in the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] See also Figure 1-7 The present invention provides a technical solution: a method for identifying damage patterns of refractory materials based on acoustic emission technology, comprising the following steps:

[0045] S1. Acoustic emission signals of refractory materials under static load are collected through three-point bending test to determine the number and characteristic parameters of data set categories of acoustic emission signals.

[0046] The material used in the experiment is magnesium-carbon refractory material, and the basic size of the material is 125mm×25mm×25mm. The material used in the experiment is pressed and formed by a pressure testing machine (forming pressure is 150MPa), with sintered magnesia as aggregate, magnesia fine powder, graphite, metal aluminum powder as base material, and resin as binder. Finally, the formed sample is placed in a constant temperature blast drying oven for 24 hours of insulation (the temperature is set to 110℃). The experiment uses a differential high temperature stress strain testing machine (HMOR / STRAIN) to load the material at three points. While loading, the DISP acoustic emission detection system is used to collect acoustic emission signals in real time. The collected signal parameters include rise time, peak count, count, energy, duration, amplitude and ASL. The content of this experiment is to monitor the bending of magnesium-carbon refractory specimens by acoustic emission until the specimen breaks, that is, the whole process of bending of magnesium-carbon refractory specimens. The loading rate of the material testing machine is 0.25N / (m2·s). Figure 1 Shown is the experimental schematic diagram.

[0047] By Figure 2 It can be seen that during the loading process, the cumulative energy of acoustic emission and the cumulative ringing counts have basically the same changing trend over time. Since the signal characteristics such as the total ringing count and the cumulative energy can clearly reflect the damage activity of the material during acoustic emission, the damage experienced by the refractory material during loading can be divided into: the initial stage of damage, the matrix damage stage and the interface damage stage. Therefore, the acoustic emission process of the material bending process can be divided into these three stages according to the trend of the curve change in the figure:

[0048] (1) The first stage is the damage incubation stage (21-33.66s): There is no obvious change in the AE cumulative energy and cumulative ringing count curves, the average frequency of the signal is concentrated in 9-20KHZ, and no AE signal with an amplitude greater than 70dB is generated.

[0049] (2) The second stage is the damage accumulation development stage (33.66-62.14s): the AE cumulative energy and cumulative ringing counts slowly increase, and the AE activity increases significantly with the increase of load. Figure 3 It can be seen that the amplitude of the acoustic emission signal at this stage is mainly concentrated in 50-60dB, the average frequency is concentrated in 30-40KHZ, and the rise time is concentrated in 10-25μs.

[0050] (3) The third stage is the instability failure stage (62.14-fracture): Figure 2It can be clearly seen that the cumulative energy and cumulative ring count of the acoustic emission signal increased sharply, and the acoustic emission activity increased significantly. In this stage, a large number of acoustic emission signals with an amplitude exceeding 70dB appeared. Among them, an acoustic emission signal with an amplitude of 100dB and a duration of 22039μs appeared at 111.6s, and the cumulative energy and cumulative ring count of the acoustic emission signal reached a peak at this time.

[0051] Through simple mathematical statistical analysis of the material acoustic emission experimental data, it can be found that the AE signal amplitude, which accounts for about 93.7% of the total signal of the material, is concentrated in the range of 50-75dB during the bending and fracture process of the material, and the average signal frequency is concentrated in the range of 21-60kHz.

[0052] S2. Use K-means clustering algorithm to perform cluster analysis on acoustic emission signals and divide them into three categories.

[0053] The K-means clustering algorithm is designed to minimize the square distance between all vectors in a data set and the center of the data set. This method assumes that the number of clusters k to be constructed is known and determined in advance. The K-means algorithm is one of the most commonly used algorithms based on the function optimal method and is also one of the most famous algorithms. The main advantage of this algorithm is its simple operation, which makes it extremely attractive in various applications and suitable for processing large data sets. The clustering criterion function used by the K-means algorithm is the error square sum criterion. Through repeated iterative optimization of the clustering results, the sum of the square distances of all samples to the center of their respective categories is minimized. Figure 4 The figure shows the K-means algorithm flow chart. The specific algorithm steps are as follows:

[0054] S21. Select K initial cluster centers: The superscript indicates the number of iterations in the clustering process.

[0055] S22. Assume that r iterations have been performed. If for a sample x,

[0056]

[0057] but in, So The sample subset is the cluster center, and all samples are distributed in K clusters according to this method, that is, the principle of minimum spacing.

[0058] S23. After reclassification, calculate each cluster center:

[0059]

[0060] in, for The number of samples included in .

[0061] S24, if If j=1, 2, ..., K, the process ends; otherwise, the process goes to S22.

[0062] Refractory materials are a kind of microscopic porous, multi-phase heterogeneous composites. There are many forms of damage, which will occur during the loading process. However, these damages can be divided into three stages: damage initiation stage, matrix damage stage, and interface damage stage. Damage will occur in these stages. Therefore, when using the K-means clustering algorithm, K is 3. Selecting appropriate features and including all classification information as much as possible can better classify the acoustic emission signals. In the acoustic emission characteristics of composite materials, the main parameters for distinguishing the various damage stages, damage types, and mechanical properties of composite components are ring count, amplitude, duration, constant load acoustic emission duration, and Felicity ratio. Therefore, amplitude, ring count, rise time, duration, and average frequency are selected as feature attributes during clustering.

[0063] There are 1195 acoustic emission signals emitted by the three-point bending test of the materials selected for cluster analysis. Figure 5 The figure shows the result of clustering the acoustic emission signals using the K-means algorithm. K-means clustering divides the acoustic emission signals into three categories. In order to correspond to the change curves of the cumulative energy and cumulative ringing count of acoustic emission during bending, the classified parameters are taken out and then matched with time to obtain the time history classification results of the acoustic emission signal, thereby better explaining the microscopic destruction process of the material. Figure 6 The figure shows the time-amplitude clustering results. Figure 7 Shown is a schematic diagram of the AE characteristic values ​​(cluster centers) and classification results of three types of injuries.

[0064] S3. Combined with the analysis of the acoustic emission parameter correlation diagram, determine the acoustic emission characteristics corresponding to different damage mechanisms, and determine the damage pattern corresponding to each type of signal.

[0065] The microstructure of refractory materials is a porous, multi-phase microscopic heterogeneous material, which is composed of high temperature resistant inorganic aggregates, bonding phases and additives. Its microscopic damage process is a very complex process. Generally speaking, the interface bonding strength and particle destruction strength are greater than the matrix cracking strength. Then, when the material reaches the initial critical state, as the external load increases, microcracks will first appear in the matrix. Once the matrix cracks are nucleated, they will immediately expand to the steady-state size and then expand to the interface phase. In fact, the source of the acoustic emission signal of the material during bending loading is very complex. The type of acoustic emission source in each stage is not unique, but should be the superposition of multiple acoustic emission sources. Based on the above theory, it can be found by comparing the acoustic emission parameter correlation diagram and the acoustic emission signal clustering results.

[0066] Furthermore, in step S3, determining the damage mode corresponding to each type of signal includes the following steps:

[0067] S31. Combined with the analysis of the acoustic emission parameter correlation diagram, the acoustic emission characteristics corresponding to different damage stages of refractory materials are identified to achieve pattern recognition of refractory material damage.

[0068] S32, taking out each classified parameter, and then matching it with time, to obtain the time history classification result of the acoustic emission signal.

[0069] S33. Analyze the characteristics of different types of acoustic emission signals, such as amplitude, ring count, rise time, duration, and average frequency, and infer the characteristics of different damage mechanisms corresponding to the acoustic emission signals.

[0070] S34. Combine theoretical analysis to explain the acoustic emission signal characteristics corresponding to different damage modes.

[0071] The first type of acoustic emission signal is the largest type of all signals, accounting for about 95.9% of the total signals. The amplitude of the signal is concentrated in 50-70dB, and the average frequency is concentrated in 30-60KHZ. And among the three types of signals, this type of signal has the smallest ring count, rise time, and duration, and the highest average frequency. Figure 6 It can be seen that this type of signal appears in the entire bending test process and occupies the largest proportion, so it can be inferred that the microcracks caused by the change in the material matrix structure extend. In the initial stage of bending loading, the original defects of the material are damaged by the external load. At this time, the degree of material damage is very low, so the amplitude of the signal is small and the average frequency is low. In the second stage, due to the continuous increase in load, a large number of cracks appear in the matrix and continue to expand, thus generating a large number of signals with high amplitude and high average frequency. In the stage of material instability and failure, as the load further increases, the matrix is ​​severely damaged, resulting in a large number of acoustic emission signals with amplitudes exceeding 70dB.

[0072] The second type of signal accounts for about 2.8% of the total signal. The amplitude and average frequency of the signal are concentrated in 80-90dB and 15-35KHZ respectively, but the average frequency of this type of signal is the lowest. Figure 6 It can be seen that this type of signal does not appear until 33.67s. The amplitude of this type of signal is significantly higher than that of the first type of signal, and this type of signal continues until the event breaks. The third type of signal has the smallest proportion among all signals, accounting for about 1.3% of the total signal. The amplitude of the signal is concentrated in 95-100dB, which is the largest among the three types of signals. Figure 6It can be seen that this type of signal only began to appear at 62.17s and continued until the event fracture. The third type of signal appears in the stage of material instability and failure. Since the material matrix is ​​severely damaged at this stage, when the cracks in the matrix expand to a certain extent, the new cracks will expand into the interface phase, resulting in interface peeling. At 111.6s, the cumulative energy and cumulative ring count of the acoustic emission signal reached a peak, and an acoustic emission signal with an amplitude of 100dB and a duration of 22039μs appeared. At this time, the material was instability due to the accumulation of damage energy to the limit. By comparing the acoustic emission characteristics of these two types of signals, it is not difficult to see that the amplitude, ring count, rise time, and duration of these two types of signals are significantly higher than those of the first type of signal, and have the same change trend, which can be classified into one category. Therefore, it can be inferred that these two types of signals correspond to interface failure. Compared with the instability and failure stage, the load is relatively small in the damage accumulation development stage, and the material damage is relatively light. Therefore, the acoustic emission amplitude, ring count, rise time, and duration of the second type of signal are lower.

[0073] The refractory damage pattern recognition method based on acoustic emission technology first determines the number and characteristics of data set categories through a three-point bending experiment, and then uses the K-means clustering algorithm to perform cluster analysis on the acoustic emission signals collected during the three-point bending experiment of the refractory material, and divides the emitted acoustic emission signals into three categories. Finally, combined with the analysis of the acoustic emission parameter association diagram, the acoustic emission characteristics corresponding to different damage mechanisms are determined, and the damage mode corresponding to each type of signal is explained, thereby realizing the acoustic emission signal pattern recognition of refractory damage. The present invention lays a theoretical foundation for establishing a mathematical model of micro-damage of refractory materials, and has a high application value in the field of studying the micro-damage mechanism of the mechanical behavior of refractory materials.

[0074] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A refractory damage pattern recognition method based on acoustic emission technology, characterized in that: The following steps are involved: S1. Collecting acoustic emission signals of refractory materials under static load through a three-point bending test, and determining the number and characteristic parameters of data set categories of the acoustic emission signals; S2. Use K-means clustering algorithm to perform cluster analysis on acoustic emission signals and divide the acoustic emission signals into three categories; S3. Combined with the analysis of the acoustic emission parameter correlation diagram, determine the acoustic emission characteristics corresponding to different damage mechanisms, and determine the damage pattern corresponding to each type of signal.

2. The refractory damage pattern recognition method based on acoustic emission technology according to claim 1 is characterized in that: In step S1, the refractory material is a magnesium-carbon refractory material, and the basic size of the material is 125 mm×25 mm×25 mm. The three-point bending experiment is to perform acoustic emission monitoring on the bending process of the magnesium-carbon refractory material specimen until the specimen breaks, and the loading rate of the material testing machine is 0.25 N / (m 2 ·s).

3. The refractory damage pattern recognition method based on acoustic emission technology according to claim 1 is characterized in that: In the step S1, among the acoustic emission signals, the amplitude of the acoustic emission signals accounting for 93.7% of the total signals is concentrated in 50-75dB, the average frequency of the signals is concentrated in 21-60kH Z, the characteristic parameters include rise time, peak count, count, energy, duration, amplitude and ASL, and the acoustic emission process of the material bending process can be divided into three stages, the damage incubation stage 21-33.66s, the damage accumulation development stage 33.66-62.14s and the instability failure stage 62.14-fracture.

4. The refractory damage pattern recognition method based on acoustic emission technology according to claim 1 is characterized in that: In step S2, the K-means clustering algorithm includes the following steps: S21. Select K initial cluster centers: The superscript indicates the number of iterative operations in the clustering process; S22. Assume that r iterations have been performed. If for a sample x, but in, So The sample subset is the cluster center, and all samples are distributed in K clusters according to this method, that is, the minimum spacing is the principle; S23. After reclassification, calculate each cluster center: in, for The number of samples included in ; S24, if If j=1, 2, ..., K, the process ends; otherwise, the process goes to S22.

5. The refractory damage pattern recognition method based on acoustic emission technology according to claim 4 is characterized in that: In the step S21, the K value is 3, which corresponds to the three stages of refractory damage, namely the damage initiation stage, the matrix damage stage and the interface damage stage. The K-means clustering algorithm divides the acoustic emission signals into three categories according to the characteristic attributes of amplitude, ringing count, rise time, duration and average frequency.

6. The refractory damage pattern recognition method based on acoustic emission technology according to claim 4 is characterized in that: In step S3, determining the damage mode corresponding to each type of signal includes the following steps: S31. Combined with the analysis of the acoustic emission parameter correlation diagram, the acoustic emission characteristics corresponding to different damage stages of the refractory materials are identified to achieve pattern recognition of refractory material damage; S32, taking out each classified parameter, and then corresponding it to time, to obtain the time history classification result of the acoustic emission signal; S33. Analyze the characteristics of different types of acoustic emission signals, such as amplitude, ring count, rise time, duration, and average frequency, and infer the characteristics of different damage mechanisms corresponding to the acoustic emission signals; S34. Combine theoretical analysis to explain the acoustic emission signal characteristics corresponding to different damage modes.

7. The refractory damage pattern recognition method based on acoustic emission technology according to claim 6 is characterized in that: In step S33, the different types of acoustic emission signals include three categories, namely, the first category, the second category and the third category. The first category of signals appears in the entire bending test process and is the largest category of all signals, accounting for 95.9% of the total signals. The amplitude of the signal is concentrated in 50-70dB, and the average frequency is concentrated in 30-60KHZ. Among the three types of signals, the ringing count, rise time and duration of this type of signal are the smallest, and the average frequency is the highest; The second type of signal accounts for 2.8% of the total signal. The amplitude and average frequency of the signal are concentrated in 80-90dB and 15-35KHZ respectively. However, the average frequency of this type of signal is the lowest. This type of signal does not appear until 33.67s, and this type of signal continues until the event breaks. The third type of signal has the smallest proportion among the three types of signals, accounting for 1.3% of the total signal. The signal amplitude is concentrated in 95-100dB, which is the largest among the three types of signals. Such a signal began to appear at 62.17s and also lasted until the event broke.

8. The refractory damage pattern recognition method based on acoustic emission technology according to claim 7 is characterized in that: In step S34, the acoustic emission signals corresponding to different damage modes have different characteristics, while the acoustic emission characteristics at the same stage are similar. Matrix damage accompanies the entire bending test process, and interface damage will only occur when the matrix damage reaches a certain degree. In addition, two of the three types of signals have similar acoustic emission characteristics and can be classified into one type.

Images