Refractory Microdamage Pattern Recognition Based On The Acoustic Emission Signal Characteristic Parameter

Refractory materials is an important high temperature auxiliary materials in the industry.Tostudy the form of refractory microscopic damage,and improve the efficiency and safety of use ofrefractory materials has an important significance.Refractories will generate a lot of acousticemission signals under loading.In this dissertation, studying the development of injury, damageto the signal characteristics and damage form using acoustic emission parameters correlation,thepower spectrum analysis, principal component analysis and K-means clustering method.Determine the different characteristics of injury signal in corresponding to the damage formMain tasks:1.To analysis the signal by acoustic emission parameters correlation and the powerspectrum analysis. Find that there are four stages: Compaction stage, injury embryonicstage,injury rapid development stage, failure stage.And three form: The frequency in50-60kHzsingle burst signal,The frequency in150-160kHz single burst signal,The superimposed burstsignal superimposed by Two single-burst signal randomly.2.The risetime, duration, amplitude, energy,counts and centroid frequency were selected asthe parameters in K-means method to divide the data into three classes. The characteristics ofless-counts, short-duration and small-energy of the first two classes both correspond one singleparoxysmal signal. The injury strength of the second class was greater than the first one.Thecharacteristics of multi-counts, long-duration and large-energy of the third class correspondsuperimposed paroxysmal signals.3.To divide the data by K-means clustering method which used the first five principalcomponent.Each class had little overlap and the classification was better.Contrasted theadvantages and disadvantages of the two classification methods. AE parameters can directlyreflect the characteristics of the signal. Principal component eliminated the correlation betweenthe parameters. To simplified the difficulty of data analysis and improved the efficiency of theanalysis.4.Identified the two main types damage of the refractories by SEM. the frequency in50-60kHz corresponded to the matrix damage,and the frequency in150-160kHz corresponded tothe interface damage.In this dissertation, acoustic emission parameters analysis combine principal componentanalysis can eliminate the correlation between the parameters,and reduce the data dimension. Inthe analysis of large data sets had a distinct advantage.To provide a new idea for refractorydamage acoustic emission signal analysis.