Based On The Acoustic Emission And The BP Neural Network Analysis Of The Reinforced Concrete Beam Damage Evolution

With the health monitoring and safety evaluation of important engineering structures has been widely concerned by the academic and engineering community and management department, nondestructive testing technology is more and more valued by people. As a new nondestructive testing technology, acoustic emission technology has unique advantages, such as high sensitivity, non intrusive and on-line continuous detection, early warning of structural damage can be carried out by AE, it also can effectively prevent the personnel casualties,economic loss and negative social influence caused by engineering structure damage.In this paper, the acoustic emission signal of the damage and failure process of the specimen is collected by the SAEU2 S centralized multi-channel USB acoustic emission detector. The damage signal characteristics of beam under different load levels are studied by using correlation analysis method of AE characteristic parameters and the improved evaluation criterion of acoustic emission signal and combined with the pattern recognition technology. Development of a set of structural damage monitoring, safety assessment and intelligent identification of damage signal method, for the acoustic emission technology services to provide theoretical basis and reference for engineering practice.(1)The damage evolution process of reinforced concrete beam is revealed by acoustic emission characteristic parameters, the specific research is to study the relationship between the rise time(RT) and the amplitude(DB) of different loading cycles, observed evolution characteristics of acoustic emission signals in the process of beam damage.The proportion of AE signals with different amplitude(DB) range is studied. It is found that the increase of high amplitude signals means the occurrence and development of cracks. At the same time, the rising time(RT) distribution and the curve fitting are analyzed and studied.The distribution of RT in each range is exponential function, and the fitting precision is high,and the mean square deviation is close to 0.9. With the increase of damage, the |b| value of function fitting coefficient is decreased, and the a value is increased. The variation of the fitting coefficient of the exponential function of the rise time distribution can accurately capture the crack initiation point of the reinforced concrete beam. The research shows that the damage evolution of the beam is closely related to the characteristic parameters ofacoustic emission.(2)On the basis of the above research, using improved acoustic emission evaluation criteria, Japan Nondestructive Testing Association(JSNDI) recommended NDIS-2421 quantitative evaluation standard, signal intensity analysis(ISA) technique and acoustic emission characteristics of the signal Kurtosis index specific evaluation of damage evolution process of reinforced concrete beam. Further qualitative and quantitative assessment of structural damage and service status, determination of the 4 typical failure stages of the beam in the damage evolution process, data division is carried out after the critical point is determined, in order to establish neural network training sample database.(3)An artificial neural network(ANN) method in pattern recognition technology is used in this paper, the BP network model is designed by Matlab, then the sample database is used to train and shape the network structure, in order to identify the acoustic emission signals of 4 typical failure stages in the process of damage evolution, for detection and recognition of the damage of the beam, and further evaluation of the security of the structure provides a new way of thinking.