Metallographic structure automatic rating method based on deep convolutional adversarial neural network

Abstract

The invention provides a metallographic structure automatic rating method based on a deep convolutional adversarial neural network. The method comprises the steps: constructing a data set and establishing a network by realizing metallographic image data enhancement; independently learning each metallographic map by using the network, generating a new metallographic map according to the characteristics learned by the network, expanding training set samples under the condition of a certain data volume and adjusting parameters through the network so that the data set is expanded to the greatest extent at least volume of failed image and data enhancement is realized, wherein the metallographic maps are divided into four categories and serve as feature vectors to serve as input of the classifier of the support vector machine, and finally data classification is conducted on the metallographic maps through the support vector machine.

Description

technical field [0001] The invention belongs to the fields of metal image recognition, deep convolutional confrontational neural network, and machine learning, and relates to the problem of extracting metallographic features and using them to enhance data and perform rating and classification. Background technique [0002] There are many kinds of detection techniques for metal materials, such as macroscopic detection, nondestructive testing, microscopic detection, ultrasonic detection, etc. From the perspective of materials science, the microstructure and properties of metal materials are closely related, and metallographic analysis can effectively predict the properties of metal materials. For example, in the metallographic image under the metallographic microscope, the part divided by the boundary is the grain. The crystal grains can be divided into 1-10 grades according to the grade size. The smaller the grain area and the larger the grade, the higher the strength and h...

AI Technical Summary

Problems solved by technology

However, although a fully automatic rating and classification has been achieved for specific metal materials, the progress of the rating results is not high, and the highest classification cannot be achieved, so the accuracy of the rating and classification needs to be improved.

Images