摘要
煤矸石筛选是煤矿生产中一个重要的环节,目前的人工捡矸法、机械湿选法以及射线分选法无法兼顾选矸过程中的高效性、清洁性和无害性。针对上述选矸过程中存在的问题,提出了一种基于卷积神经网络的煤矸石识别方法,以颜色和纹理作为煤和矸石图像的类别特征,结合迁移学习方法,构建了VGG-16深度学习模型。实验结果表明,其识别准确率可以达到99.18%,可以有效地实现对煤和矸石的识别。
Coal gangue screening is an important link in coal mine production.The current manual gangue picking method,mechani⁃cal wet separation method and ray sorting method cannot take into account the high efficiency,cleanliness and harmlessness of the gangue selection process.Aiming at the problems in the above gangue selection process,a coal gangue recognition method based on convolutional neural network is proposed.Using color and texture as the category features of coal and gangue images,combined with the transfer learning method,VGG-16 deep learning is constructed model.The experimental results show that the recognition accuracy rate reaches 99.18%respectively,which can effectively realize the recognition of coal and gangue.
作者
孙立新
SUN Li-Xin(Hebei University of Engineering,Handan 056038,China)
出处
《电脑知识与技术》
2020年第21期16-18,22,共4页
Computer Knowledge and Technology
关键词
煤矸石识别
卷积神经网络
图像分类
迁移学习
coal gangue recognition
convolutional neural network
image classification
transfer learning
