增量式矿石自动化分拣系统研究  

Study on the Incremental Automatic Ore Sorting System

作　　者：蒋卫祥[1] JIANG Weixiang(School of Software and BigData,Changzhou College of Information Techology,Changzhou,Jiangsu 213164,China)

机构地区：[1]常州信息职业技术学院软件与大数据学院,江苏常州市213164

出　　处：《矿业研究与开发》2020年第11期150-155,共6页Mining Research and Development

基　　金：江苏省高等学校自然科学研究项目(19KJB520023);常州信息职业技术学院智能制造边缘计算开放实验室建设项目(KYPT201802Z)。

摘　　要：矿产资源的分选对于矿业具有重要意义。以采煤为例,人工分选煤矸石存在效率低、工作量大、工作环境恶劣等问题,而基于传统机器学习的煤矸石分拣系统识别准确率低,且现有煤矸石知识先验模型无法有效覆盖新的矿石,需通过扩充数据集重新训练模型,造成训练时间长且训练效率低等问题。针对上述问题,研究提出了增量式自动化煤矸石分拣流水线。结合实际工程需要,构建深层卷积神经网络实现煤矸石图像的高效分选,同时采用深度增量学习算法实现对自动化煤矸石分拣流水线的增量式学习。研究结果表明,煤矸石增量式自动化分拣系统分类准确率达到99%,高于传统机器学习算法分类准确率7.7个百分点,且具有增量学习的能力。The sorting of mineral resources is of great significance for mining industry.Taking coal mining for an example,manual sorting coal gangue existes such problems as the low efficiency,big workload,poor working conditions,and the accuracy of the coal gangue sorting system based on traditional machine learning is low.And the existing coal gangue knowledge prior model cannot effectively cover new ore,and it is necessary to retrain the model by expanding the data set,which results in long training time and low training efficiency.Aiming at the above problems,an incremental automatic coal gangue sorting pipeline was proposed.Based on the actual engineering requirements,a deep convolutional neural network was constructed to realize the efficient sorting of gangue images,and the deep incremental learning algorithm was used to realize the incremental learning of the automatic gangue sorting pipeline.The results show that,the classification accuracy of the incremental automatic sorting system of coal gangue reaches 99%,which is 7.7 percentage points higher than that of the traditional machine learning algorithm,and has the ability of incremental learning.

关 键 词：煤 矿石分选 增量学习 自动分拣 智能识别 

分 类 号：TD421[矿业工程—矿山机电]