Damage Detection And Reliability Of Piles Based On Wavelet Transform

Traditional low strain testing method is especially appropriate for assessing pile quality because of its low cost,easy operation,flexibility and minor damage to pile.However,the amplitude change of the reflected wave caused by pile defects is often influenced by ambient noise and interaction between pile and soil,hence,the damage positions of piles can not be estimated well by the amplitude information alone.In addition,although the low strain testing method is usually used to localize the damage position through the velocity response curve from the top of the pile,it is hard for us to evaluate the damage degree of the pile.In order to address these issues,a new signal processing method is proposed to detect the damage position of piles based on a combination of analytical mode decomposition,recursive Hilbert transform and complex continuous wavelet transform.More over,to solve the problem of subjective selection of energy concentration points in the wavelet scalogram and interference points in the grayscale images of phase angles,a new method of pile damage localization based on K-means clustering algorithm,fast Fourier transform and complex continuous wavelet transform is proposed.As for the damage degree evaluation of piles,dynamic fingerprint of low strain reflected wave is introduced to address this issue.On a basis of this,a reliability method of pile damage identification based on dynamic fingerprint of low strain reflected wave combined with response surfaceMonte Carlo method is used to solve the problem of uncertainty in pile damage identification.Finally,the effectiveness and accuracy of the proposed methods are verified by numerical examples and low strain dynamic tests.The main research work and innovations of this dissertion are as follows.(1)The analytical mode decomposition theorem is first introduced into the measured response signal to extract the reflected wave component signal in concerned frequency band.Then,the recursive Hilbert transform is used to demodulate the signal.After that,the complex continuous wavelet transform is performed on the demodulated signal and the resultant phase angle of wavelet coefficients is displayed on the time-frequency plane.Finally,the damage position of pile is detected by picking up the corresponding phase angle turning point on the time-frequency plane.(2)The complex continuous wavelet transform is applied to the response signal and then the wavelet scalogram is obtained.The energy concentration points of the wavelet scalogram and their time coordinates are selected by K-means clustering algorithm.Meanwhile,a fast Fourier transform is performed on the response signal to obtain a spectrum diagram,by which a frequency band of interest is defined.On a basis of the determination of the time coordinates and the concerned frequency band,a specific area used for the searching of phase angle turning points on the time-frequency plane is defined objectively.(3)In view of the pile material,soil around the pile,pulse load and other relative parameters are known beforehand,the evaluation of different damage degree is carried out fora fixed damage position of a pile.Based on the velocity signal curve from the pile head,the low strain reflected wave dynamic fingerprint is obtained.Thus,the relationship between pile integrity coefficient and low strain dynamic fingerprint is established,which can effectively evaluate pile damage degree.(4)In order to solve the problems such as ambient noises and the variability of excitation load,pile material and soil around the pile,a response surface-Monte Carlo method is employed on a basis of the low strain reflected wave dynamic fingerprint method.This method not only reduces the error caused by highly nonlinear response surface function,but also overcomes the shortcomings of Monte Carlo method which function is implicited.Thus,the reliability of pile damage identification is accurately calculated as well.