Metallographic structure recognitionon method based on bilinear convolutional neural network

Abstract

The invention discloses a metallographic structure recognition method based on a bilinear convolutional neural network. The method comprises the following steps: (1) collecting metallographic pictures to form a data set; (2) dividing the data set into a training set and a test set, training two different convolutional neural classification networks by using data of the training set, and testing the networks by using data of the test set; (3) establishing a bilinear convolutional neural network on the basis of the two trained convolutional neural classification networks; (4) training the bilinear convolutional neural network by using the training set data; (5) recognizing a metallographic structure by using a bilinear convolutional neural network; according to the method, the bilinear model is applied to a metallographic structure recognition task for the first time, the model can extract richer feature information, and compared with a single convolutional neural network model, the recognition accuracy of the metallographic structure is improved.

Description

technical field [0001] The invention belongs to the technical field of metallographic detection of metal material structures, and in particular relates to a metallographic structure recognition method based on a bilinear convolutional neural network. Background technique [0002] Metallography refers to the chemical composition of metals or alloys and the physical and chemical states of various components inside the alloy. Metallographic structure is closely related to the comprehensive properties of materials, and metallographic inspection is also an important means of predicting and analyzing material properties. In the oil refining and chemical industry, there are many process equipment serving in high temperature, high pressure, corrosion and other environments, such as reactors, towers, heat exchangers, etc. During the operation of these equipment, affected by environmental factors such as temperature, pressure and corrosive medium, the structure of the metal material ...

AI Technical Summary

Problems solved by technology

Methods based on machine learning models or a single convolutional neural network model cannot extract enough features to identify the type of metallographic structure

Images