Metal additive manufacturing image detection method and device combining adversarial neural network with local binary value

Abstract

The invention provides a metal additive manufacturing image detection method and device combining an adversarial neural network with local binary. The method comprises the steps: obtaining a molten pool and a sputtering image of metal additive manufacturing, and carrying out the preprocessing of the molten pool and the sputtering image; utilizing the trained generative adversarial neural network to repair defects in the preprocessed molten pool and sputtering image to obtain a repaired molten pool and sputtering image; utilizing a local binary pattern algorithm to obtain LBP values of the molten pool, the sputtering original image and the repaired molten pool and sputtering image; and the difference between the molten pool and the sputtering original image and the repaired molten pool andthe sputtering image is identified according to the numerical difference of the LBP values of the molten pool and the sputtering original image, the image of the difference part is the image of the defect area, and the defect area is accurately positioned and identified according to the image of the defect area. Defects in a molten pool and a sputtering image in the metal additive manufacturing process can be positioned and recognized, so that manufacturing process parameters are adjusted in real time, and the part manufacturing yield is increased.

Description

technical field [0001] The invention relates to the field of image detection for metal additive manufacturing, in particular to an image detection method and device for metal additive manufacturing with anti-neural network combined with local binary. Background technique [0002] Metal additive manufacturing technology is widely used in high-end manufacturing fields such as aerospace manufacturing and medical equipment. Non-destructive testing technology for quality monitoring of formed parts has higher and higher requirements for accuracy, real-time performance, and ease of operation. In recent years, there have been a lot of research progress in the detection of additive manufacturing defects. Most of them use various sensors to sense and measure the information of the processing area of ​​parts in real time, and obtain defect information through a series of data processing processes, which are used to assist in the quality inspection of manufactured parts. To reject unqua...

AI Technical Summary

Problems solved by technology

However, the current machine learning programs are difficult to take into account the defect location, area measurement and stable and efficient recognition speed in the metal additive manufacturing process, which ultimately affects the recognition efficiency and recognition accuracy.

Images