Pattern Recognition Of Ultrasonic Testing Of Material Defects Based On COMSOL Simulation

In this paper,the pattern recognition of two kinds of typical defects(cracks and voids)in materials is studied by means of ultrasonic testing technology.In order to obtain more abundant echo signals of typical defects,the finite element simulation software COMSOL is selected to simulate the propagation of ultrasound in materials and the interaction with defects.Compared with the experimental results,the results are in good agreement,which proves that the research on the simulation of ultrasonic defect detection based on finite element method is credible.On the basis of the above research,the following work is mainly carried out:(1)Use reasonable component and defect models.The 45# steel cylinder is selected as the tested component.The defect model is based on the ellipse.Three kinds of defects of the long axis size are set: the long axis is smaller than the diameter of the probe,equal to the diameter of the probe(10 mm)and larger than the diameter of the probe,respectively.By adjusting the short-to-long axis ratio to represent the crack(short-to-long axis ratio is 0.05/0.1/0.2)and the pore(short-to-long axis ratio is0.4/0.6/0.8/1),there are 21 defect models with different shapes,which comprehensively cover the main factors affecting the detection.By analyzing a large number of defect echo signals,the different laws of crack and pore echo signals are summarized,which can provide reference for defect pattern recognition.(2)Defect signal characteristics and pattern recognition of omnidirectional detection.The longitudinal wave probe was used to detect cracks and pore defects,and the echo peaks of defects were extracted.It was found that the echo peaks of different morphological defects varied in the detection range of 0~180°,and the peak coefficients(the ratio of 90° to 0°)were used as characteristic parameters to effectively identify cracks and pore defects.By changing the detection angle and using dynamic echo spectrum analysis method,the defect type can be qualitatively identified.(3)Detection of defective signal characteristics and pattern recognition under limited azimuth.By fitting the relationship curve between the peak value of defect echo detected by a straight probe and each detection angle interval,and taking the slope value of the curve as the characteristic parameter,it is found that cracks and pore defects can be distinguished in the small detection intervals of 0-30° and 20-50°.However,it is noteasy to distinguish two kinds of defects by using slope characteristic parameters in the detection range of 50~90°.For this reason,the cracks and pore defects can be distinguished obviously by adding 10 oblique probe detection in the detection range of50~90° according to the relationship between the ratio of low-frequency peak value to high-frequency peak value of spectrum waveform and the detection angle.Through the finite element simulation analysis of ultrasonic detection of cracks and pore defects of different sizes and sharpness in cylinder,the pulse sound reflected from different parts of the defect is collected,which provides more systematic data for pattern recognition of defect types.Under the condition of omni-directional detection and limited detection orientation,cracks and pore defects of different shapes can be well identified according to the time-domain and frequency-domain characteristics of defect echo signals.This study can provide a reference for pattern recognition of actual ultrasonic testing defects,and is of great significance for predicting component life and ensuring equipment safety.