The Study Of Acoustic Emission Parameters And Micro-damage Pattern Recognition About Refractory

Refractory is an integral part of the high temperature industrial buttress materialand is widely used in modern in industries.Due to the complex microstructure ofRefractory materials.It is very difficult to study the microscopic damage of Refractorymaterials for many research workers,and so far, the microscopic damage type ofRefractory materials is still not clearly.In this paper, in order to solve the twoproblems which are the microscopic damage type of refractory is uncertain andproperly selecting the class index class is difficult,we take acoustic emissiontechnology as a means,and combined with some mathematical statistics analysisalgorithm, realize the classification about damage signals of Refractory materials,thedamage signals of refractory materials are divided into two classes.At the end of thepaper, combined with the results electron microscope scanning experiment, we verifythe rationality of the classification.Main works and achievements of the thesis are asfollows：1、By the acoustic emission experiment, collected a lot of original acousticemission signals of refractory materials which is the research object and providesenough data support for our study.2、By using principal component analysis method, the15acoustic emissionparameters (indicators) which obtained in experiment are eliminated the correlationdimension, and we construct two new parameters (indicators), which significantlyreduce the number of indicators and the complexity of the clustering analysis.3、By using two kinds of clustering algorithms(k means clustering algorithm andgaussian mixture model clustering algorithm) in which the principle has very bigdifference.The damage signals of refractory materials are divided into two classes.4、By analyzing and comparing the two clustering results generated by the twoclustering algorithms,and combining with electron microscope scanning results, thereasonability of our classification is furtherly verified.In this paper,we break the difficulties of traditional refractory damageclassification and avoid the complex direct analysis of acoustic emissionsignals,convert research ideas from acoustic emission parameters and combine with avariety of mathematical statistical analysis algorithms, provides a new idea for thedamage of the material.