Pile Foundation Defects Identiifcation Based On Wavelet Analysis And Support Vector Machine

This paper studies the pile ultrasonic transmission method of data processing.Faced with those problems of fuzzy identification on the type of defect, a novelapproach of pile foundation defects identification using ultrasonic transmissionmethod was presented based on wavelet analysis and support vector machine (SVM)pattern recognition theory. The main idea of this method is: First according to thesynthetic criterion, the effects of different defects on acoustic parameters are studied.Analyzing the links between different kinds of defects in concrete and numerousacoustic parameters, then analyzing the testing data using wavelet transform, Parsethe data in the different frequency components, in order to distinguish thediscriminating factors of the defect types, the SVM classification modal isestablished, and then different types of defects are identified. Laboratory and fieldexperiment indicate that the SVM has higher accurate rate and better classificationperformance. The main work and conclusions include:(1) The application of ultrasonic transmission method in pile foundation testingis introduced.(2) Describes the propagation characteristics of ultrasound in concrete, from atheoretical analysis of ultrasound propagation in concrete.(3) The wavelet analysis theory and its application in signal procession areintroduced.(4) According to the Machine learning theory and statistical knowledge supportvector machine (SVM) pattern recognition theory is introduced . And then the pilefoundation defects identification ultrasonic transmission method based on waveletanalysis and support vector machine is proposed.(5) A group of experiments is designed to simulate these defects and ultrasonictransmission method is used to complete the data collection.(6) Establish the SVM classification modal to identify different types of defectsintelligently, and use the well trained modal to forecast data practical.