Identification and classification of fused magnesia blocks based on convolutional neural network
The manual sorting process of fused magnesia results in low classification accuracy and has harsh working environment.The image recognition technology of artificial intelligence has the advantages of high efficiency and reliability,so the identification and classification of fused magnesia by artificial intelligence is a good solution to this problem.Based on the image recognition technology of convolutional neural network,after collecting the macroscopic characteristics of a large number of fused magnesia samples,different fused magnesia varieties were identified and classified,and 150 iterations of high calcia fused magnesia images were trained.The results show that using convolutional neural network to train the high calcia fused magnesia image,and the training accuracy is the highest at the 104th iteration,reaching 97.2％.In actual identification,all the prediction probabilities of six fused magnesia samples are more than 99.4％,which can not only re-duce labor,but also improve the classification efficiency and identification accuracy for fused magnesia.
fused magnesiamacroscopic characteristicsconvolutional neural networkidentification and classification
万方数据
维普
