Application Research Of Concrete Health Detection System Based On Ultrasonic

Concrete is one of the most important materials in engineering construction.Concrete is a complex multi-phase non-homogeneous medium.When ultrasonic testing concrete,a series of complex acoustic phenomena are generated,make the detection signal complex and difficult to identify.Based on the analysis of ultrasonic concrete testing,this thesis establishes the physical model of ultrasonic testing system.The software and hardware of the ultrasonic testing system are designed based on the physical model and compared with the theoretical model output,ensure the consistency of theory and experiment.Aiming at the characteristics of concrete complexity and uncertainty in reality,a method combining wavelet packet decomposition and BP neural network is proposed to provide an effective detection method for ultrasonic concrete detection.The main contents of the thesis include:(1)Analysis of the physical model of ultrasonic testing system.Based on the piezoelectric equation,wave equation and acoustic theory,this paper analyzes the vibration response of the piezoelectric transducer,and establishes the dynamic model of the piezoelectric transducer and the acoustic radiation model of the pressure transducer.(2)Based on the physical model analysis of ultrasonic testing system,the software and hardware of ultrasonic testing system are designed by combining microelectronic technology and computer technology.The acoustic time and waveform transmitted in the air by the ultrasonic testing system are compared with the output value of the theoretical model to ensure the reliability of the ultrasonic testing system.(3)Aiming at the complexity and uncertainty of ultrasonic wave in concrete.The complexity and uncertainty of ultrasonic in concrete make high-frequency noise signals in the defect signal.In this thesis,the wavelet packet decomposition feature of multi-resolution is used to remove the high-frequency noise in the signal,and seven characteristic parameters reflecting the signal characteristics of defects,such as acoustic time,mean and peak value,are extracted.Analysis of the single characteristic parameter cannot distinguish the size of the defect,the BP neural network is applied to identify the defect,which improves the accuracy of concrete defect detection.