Microwave NDT Imaging For Multiple Cracks In Metals Based On Improved Canny Algorithm

In the complex and extreme environment,there are often various types of defects on the surface and near surface of metal equipment,which brings many difficulties to the normal operation of the equipment,and causes serious accident and loss.The conventional Nondestructive testing method has not been fully satisfied in the case of dust,high and low temperature,high pressure and other complex conditions,so it is necessary to study the microwave Nondestructive testing method which has a minimal influence in the propagation of complex conditions.At the same time,the traditional quantitative detection methods to obtain data is not intuitive,the relevant professionals need to analyze the data to draw the corresponding conclusions,so there is a higher demand of imaging methods for defect localization and boundary recognition.This paper focuses on the methods of microwave nondestructive imaging and metal defect boundary recognition:(1)The finite element simulation model and parameter optimization.Firstly,the finite integral simulation the wide frequency band time domain model is established,the influence of the lift-off on the reflection coefficient magnitude and phase,and the regularity of the periodic variation of the half wavelength extraction are analyzed;the optimum frequency and the lift-off of the magnitude and phase of the reflection coefficients of the rectangular waveguide in the wideband frequency range are studied.The relationship between the width and depth of defect and the magnitude and phase of reflection coefficient is studied,and the reflection coefficient phase is used to characterize the defect boundary by comparing the sensitivity of the magnitude of reflection coefficient with the phase.The relationship between the circular crack boundary and the position of maximum curve gradient and the corresponding relationship between the center of the Curve and the center of the defect are studied.(2)Microwave nondestructive imaging system and experiment.The microwave nondestructive testing imaging platform is set up,and the hardware system of the microwave Nondestructive imaging system is established.Based on Visual Baisc,a planar scanning program is programmed to drive the scanning platform by controlling 9030 board;Based on Visual C++,data acquisition program is utilized to acquire the magnitude and phase of the reflection coefficient from PNA E8363 B through WAN;the relationship between the crack width and the depth of the defect and the magnitude and phase of the reflection coefficient are studied in the experiment,and the relationship between the defect width and the phase difference of the reflection coefficient is validated.The relationship between the defect depth and the magnitude and phase of the reflection coefficient is studied,the relationship between the phase and amplitude of the reflection coefficient and the location of the defect center is studied.The relationship between the circular crack boundary and the maximum curve gradient position and the corresponding relation between the center of the Curve and the center of the defect are studied,and the linear law of the phase difference between the maximum and the lowest reflection coefficients of the rectangular defect with different width is verified by simulation and experiment.The corresponding relationship between the boundary of the maximum gradient position of different wide circle defects and the diameter boundary of the circle defect is verified.The sensitivity of the magnitude of the experimental reflection coefficient and the phase to the defect boundary is compared,and selection of the phase representation of the defect boundary is validated.(3)The boundary recognition algorithm of microwave nondestructive testing imaging.Several first-order and second-order boundary recognition algorithms are studied,and the advantages and disadvantages of these boundary recognition algorithms are compared.The first-order and second-order boundary recognition algorithm is used to identify the boundary image of the experimental defect scan results,and the deficiency of the traditional boundary imaging algorithm in microwave nondestructive testing is studied.An improved canny algorithm based on the least squares phase unwrapping algorithm is proposed,and the improved boundary imaging algorithm is used to identify the boundary of the microwave nondestructive imaging results of the rectangular defect and the circular defect.Finally,the SVM algorithm based on Hog feature extraction is used to classify the defect boundary in the experiment by using the boundary imaging simulation results of rectangle defect and circle defect as sample.