Research On Weld Defect Identification System By TOFD

Time of Flight Diffraction Ultrasonic detection method is based on the measurements of the travel time of the echoes diffracted by the defect. It has three kinds of display modes called A-scan, B-scan and D-scan, the detection result is direct and precise. TOFD Ultrasonic detection method has simple principle, convenient operations, fast detection speed, high crack defects detection rate, especially suitable for the thick weld testing and with many other advantages it is widely used in many fields.This paper designed a manual scanning device for the TOFD detection system, write the image processing system based on VC++6.0. In order to locate the depth of defect accurately, due to the image distortion caused by the roughness of surface, or the scanning velocity mutation, and according to the principle of TOFD detection, straighten the lateral waves based on the peak search algorithm, corrected the image very well and make the lateral waves remain at the same level; The actual weld inspection process collects a large amount of data, so extraction the effective weld area based on the previous process, reduced the data of the image, and improve the efficiency of the processing.As an important part of TOFD detection, defect interpretation is mainly accomplished by people, waste a lot of time and energy. In addition, due to the personnel knowledge, experiences and the proficiency of the detector, especially when there is a large amount of data, it may get the wrong result. Therefore, in order to improve work efficiency, and reduce the error rate, In this paper, rebuild the image background of D-scan based on the frequency domain filtering, and using the background difference method to extract effective weld defects, extract the welding defects automatically, improves the efficiency and accuracy of the weld detection.The TOFD inspection system in this paper includes inspection modules such as cross curse and parabola pointer curse, it can straight the lateral wave and extract the effective weld region, according to the original image use the frequency filtering process rebuild the background image, and get the right defects automatically. The experiment shows that this method can extract defect targets accurately.