Hot-rolled strip steel surface defect classification method based on convolutional neural network

Abstract

The invention discloses a hot-rolled strip steel surface defect classification method based on a convolutional neural network, and belongs to the field of computer deep learning. The method comprisesthe following steps: firstly, obtaining a hot-rolled strip steel surface defect typical image sample from an NEU database, and preprocessing the sample; building a convolutional neural network VGG16 model, and building a plurality of classification models for the surface defects of the hot-rolled strip steel based on the VGG16 model in combination with an SGD or Adam optimization algorithm; then,based on a plurality of built classification models, identifying and classifying the hot-rolled strip steel surface defect typical image samples obtained in the step; evaluating results obtained by the plurality of classification models to obtain an optimal classification model; and finally, based on the optimal classification model, hot-rolled strip steel surface defect classification is carriedout. The method for identifying the surface defects of the hot-rolled strip steel is high in accuracy and high in classification speed, and can be effectively applied to on-site real-time detection ofthe surface defects of the hot-rolled strip steel.

Description

technical field [0001] The invention relates to the field of computer deep learning, in particular to a method for classifying surface defects of hot-rolled strip steel based on a convolutional neural network. Background technique [0002] In the actual hot-rolled strip production process, due to various physical and chemical factors and the complexity of the hot-rolling process, the surface of the strip is prone to various problems such as scale, scratches, pockmarks, inclusions, plaques, and cracks. All kinds of defects bring huge economic and commercial reputation losses to product manufacturers. Hot-rolled strip steel has high surface temperature, strong radiant light, and the existence of problems such as water, scale, and uneven light effects, which have become one of the main difficulties in using machine vision to detect defects. Large intra-class differences and high inter-class similarities increase the difficulty in the development of hot-rolled strip surface def...

AI Technical Summary

Problems solved by technology

However, it is difficult for these methods to extract all the features contained in the image and make effective use of them.

After the convolutional neural network was proposed, the classification accuracy and detection speed of hot-rolled strip surface defects have been greatly improved, but the problem of low classification accuracy or inability to meet real-time requirements still exists. At the same time, the existing convolutional neural network There are many model training parameters, large system consumption, and memory usage. At the same time, there will be a "gradient dispersion" problem in the deep network, and a large amount of sample data needs to be used for prediction, which also brings certain difficulties to the classification of hot-rolled strip surface defects.

Images