Research On Concrete Acoustic Emission Signal Recognition Technology Based On SVM

Concrete structures or components mainly bear pressure.In actual projects,concrete components often suffer from brittle failure.Cracks in concrete can work normally under controllable conditions.When cracks expand to a certain width,they will cause a series of problems such as bearing capacity reduction and reliability reduction,and even seriously threaten people’s life and property safety,such as It can catch the emergence and development of cracks in concrete in time,identify the location and severity of concrete damage in time,so as to monitor and diagnose the major engineering structure or important structural parts,and make early warning of structural damage,so as to effectively improve the safety of the project.In this paper,through the experiments of concrete lead breaking,concrete uniaxial compression brittle failure and concrete acoustic emission signal collection,based on the relevant theoretical knowledge of concrete fracture mechanics,the acoustic emission signal data set produced by concrete cube axial compression brittle damage is taken as the research object,combined with the classical concrete acoustic emission signal recognition technology and clustering algorithm The research results are compiled into software for application research.The main contents of this paper are as follows:(1)By studying the fracture mechanics of concrete,it is found that the main acoustic emission source of acoustic emission signal is the energy release in the process of crack tip deformation and crack growth.The formation and propagation of acoustic emission signal in concrete also carries the internal information of concrete.Therefore,the formation of cracks can be recognized from the position of the acoustic emission source,and the crack in concrete can be recognized from the angle of energy The expansion of the Department.Through the research of clustering algorithm,it is found that the Gaussian mixture model clustering algorithm and the attribute weighted fuzzy clustering algorithm of maximum entropy regularization can be applied to the acoustic emission signal recognition.(2)The concrete lead breaking signal is a typical burst signal,the acoustic emission signal produced by lead breaking is seriously attenuated in concrete.The attenuation trend is fitted by polynomial.The fitting function is y=1.0618+(-2.08111x)+1.43889x~2+(-0.33951x~3).The dominant frequency of lead breaking signal is 4895hz.The research on the failure mode of concrete brittle failure experiment shows that in the process of concrete uniaxial compression brittle failure experiment,the failure modes of each group of test blocks in different ages are basically the same.For the test blocks with the same water cement ratio,the concrete strength increases with the increase of age.For the test blocks with the same age,the water cement ratio is 0.4,0.45,0.5 and 0.55,the compressive strength decreases gradually.The experimental results of concrete acoustic emission signal acquisition show that the acoustic emission signals of concrete cracks in the noise signal and waveform are significantly different in the waveform and amplitude changes;the comparison of cumulative energy,load arrival time relationship of three groups of test blocks in different curing age shows that their load arrival time relationship is composed of linear growth stage and load drop stage The distribution of energy in the whole loading stage has the phenomenon of agglomeration in different time nodes,and the cumulative energy has the phenomenon of jump and sudden increase in different time nodes.With the increase of maintenance age,the slope of the cumulative energy curve keeps to the straight section after the sudden increase is more obvious.The location distribution of acoustic emission events further confirms the law of acoustic emission events.The failure tendency of concrete in compression test is the same.(3)The correlation analysis of the characteristic parameters of the acoustic emission signal shows that by excluding the characteristic parameters whose correlation coefficient value is greater than or equal to 0.8,the characteristic parameters of the acoustic emission signal can be effectively simplified and the redundancy of the sample data can be avoided.The results of principal component analysis show that the introduction of variance contribution rate(A)and cumulative variance contribution rate(T)can eliminate the small impact on the clustering results Three new synthetic parameters representing the original feature parameter set are obtained.They are named as the first principal component,the second principal component and the third principal component.Taking the principal component as the pattern feature,the clustering results of the Gaussian mixture model clustering algorithm and the maximum entropy regularized membership weighted fuzzy clustering algorithm show that the Gaussian mixture model clustering algorithm can be divided into two parts Three different types of damage signals are identified,but the boundaries between the three types of damage signals are not obvious.After the introduction of time parameter arrival time,the correlation between clustering results and arrival time is not strong,which can not be used to explore the early warning mechanism of AE signal type changing with time.Based on the maximum entropy regularization,attribute weighted fuzzy clustering algorithm is used to analyze the advantages of AE signal sample data Obviously,three different types of damage signals can be distinguished,and the boundaries between the three types of damage signals are obvious.After introducing the time parameter arrival time,it is found that the clustering results have strong correlation with the arrival time,which can be used to design the early warning mechanism of acoustic emission signals.(4)Using C#and MATLAB 2014a Based on the mixed programming technology,a recognition system of acoustic emission signal of compression deformation of concrete cube is designed based on the maximum entropy regularization attribute weighted fuzzy clustering algorithm.The data of acoustic emission signal of axial compression damage of concrete test block with the curing age of 28d is used as the sample data set for clustering analysis.The clustering results of the sample data show that the recognition system can not only successfully complete the loading of the data set At the same time,it can judge whether the clustering result is beyond the normal range.If it is beyond the preset normal standard,the system will mark red in the corresponding category and send an early warning signal.