Quantitative Detection Method of Inclusion in Aluminum Alloy Ingots Based on Machine Learning
Author of the article:LIU Jinlin;DU Qungui
Author's Workplace：School of Mechanical&Automotive Engineering,South China University of Technology,Guangzhou 510640,China
Key Words：aluminum alloy ingot slag content detection K-mode convolutional neural network error back propagation algorithm
Abstract：In order to solve the problems of low efficiency and large error in manual detection of Inclusion content in cast aluminum alloy ingot,based on k-mode detection technology,a detection method combining convolutional neural network and error reverse propagation network based on machine learning was proposed to realize quantitative detection of Inclusion content in cast aluminum alloy ingot.Firstly,the convolution neural network was used to qualitatively analyze whether the K-Die fracture surface contained Inclusion.Then,the backpropagation network and sliding window were used to extract features and calculate the Inclusion content quantitatively.Finally,the non-maximum suppression algorithm was used to optimize and post-process the detection results.The results show that the accuracy of the convolutional neural network is 85.88%,which can accurately judge whether the fracture surface of K-Die contains Inclusion and calculate the Inclusion content of the fracture surface of K-Die,so as to meet the basic requirements of actual production.
