Analysis And Research On The Grade Quantization Of Defect Signal Based On Nonlinear Ultrasonic Guided Wave

In national defense construction and economic development,pipeline transportation plays an important role.In order to ensure safe transportation,buried pipelines are often effectively protected by covering the anti-corrosion layer.However,the working environment where the anti-corrosion layer is located is prone to erosion.For the defects of the anticorrosion layer,the use of external detection methods will not only consume a lot of resources,but also be limited by environmental conditions.It is currently the most mature detection technology for pipeline anti-corrosion coating defects by using ultrasonic guided waves to collect and transport pipeline defects,and the identification,processing and analysis of ultrasonic guided wave defect echo signals are the focus of current detection technology.The purpose of this paper is to classify the defects of pipeline anti-corrosion layer and realize the quantification of the defect levels.Firstly,the paper introduces the detection technology of pipeline anticorrosion layer defects and its actual application in industry,and analyzes the propagation mechanism of ultrasonic guided waves in solid single-layer media and double-layer media and their interfaces.Then,the non-linear ultrasonic guided wave signal is transmitted through the transmitting probe and the echo signal of the pipeline anticorrosion layer defect is obtained through the receiving probe,and the local wavelet method is used to obtain the denoised echo signal,for different types of corrosion layer defects,the kurtosis coefficient,skewness coefficient,dispersion coefficient,shape coefficient and wavelet packet energy coefficient of the echo signal present different ranges,and a corrosion prevention based on principal component SVM is proposed.The layer defect classification method also introduces dynamic perturbation particle swarm optimization algorithm to obtain the optimal penalty coefficient C and RBF kernel function g value of SVM.Finally,a mutual information quantification method based on kernel density estimation is proposed.By calculating the similarity between the mutual information kernel densities of the corrosion layer defect feature vectors,the corrosion layer defects can be quantified to achieve different types of pipeline corrosion layer defect levels.The research results show that the SVM corrosion layer defect classification algorithm based on principal component analysis can achieve the effective classification of three types of pipeline corrosion layer defects,such as holes,cracks and pits.Compared with the traditional SVM classification,the accuracy rate is improved by 12.9%.Layer defect detection provides an effective defect classification method.The mutual information classification method based on kernel density estimation can accurately achieve the quantification of the grade of pipeline anti-corrosion layer defects,and the quantification accuracy rate is above 87%.