
Russian engineers have developed a system for quickly searching for pipeline defects
Gazstroyprom specialists, together with colleagues from the Center for Industrial and Industrial Solutions of Innopolis University, have developed a system for artificial analysis of pipeline defects using AI technologies. This was reported by the press service of Innopolis University.
This system allows you to find pores, slag inclusions, lack of fusion and other internal defects in welds with 94% accuracy. The principle of its operation is to analyze digital radiographic images obtained using flat-panel matrix detectors - special devices that make it possible to convert X-ray radiation into a digital image of a welded joint.
“The examination of pipes using flat panel array detectors is called digital radiographic non-destructive testing. The technique is based on the ability of X-ray waves to pass through metal. Despite the perfection of the hardware, the method has limitations and problems in processing the results by a non-destructive testing specialist, for example, subjective interpretation of defects, labor intensity and processing time. The developed solution provides a consistent and accurate interpretation of the results, reducing the likelihood of human error, and allows us to obtain a conclusion about weld defects 30 times faster than with manual processing by a specialist,” said Dmitry Kostyukov, head of the non-destructive testing technology sector at Innopolis University.