露天台阶爆破块度智能分析平台研发与应用  

Research and Application of Intelligent Analysis Platform for Open-pit Bench Blasting Fragmentation

作　　者：黄磊 刘玉龙 陶明[2] 向恭梁 徐源泉 HUANG Lei;LIU Yulong;TAO Ming;XIANG Gongliang;XU Yuanquan(CGNPC Uranium Resources Co.,Ltd.,Beijing 100029,China;School of Resources and Safety Engineering,Central South University,Changsha 410083,China)

机构地区：[1]中广核铀业发展有限公司,北京100029 [2]中南大学资源与安全工程学院,湖南长沙410083

出　　处：《金属矿山》2025年第4期164-173,共10页Metal Mine

基　　金：国家自然科学基金项目(编号:12072376);中南大学中央高校基本科研业务费专项资金(编号:CX20240247)。

摘　　要：在露天矿山生产过程中,精准获取爆破块度信息对矿岩铲装和加工至关重要。针对目前传统爆破块度测量技术操作复杂和耗时长,无法直接应用于矿山生产作业的问题,研发了集成矿山现场碎石块度动态识别、实时传输和静态分析功能的爆破块度智能识别平台。结合双目深度视觉与Yolov8-seg模型构建智能分析平台的核心算法,系统部署在矿岩运输路径中,自动采集矿卡车斗内的爆堆图像,通过Restful标准接口协议打通该系统与生产调度系统的数据交互。建立了具备现场块度动态识别功能和静态图像分析功能的两大核心业务板块,在湖山铀矿成功应用。使用过程中对干扰物体、阴影和强光照等影响因素进行了针对性强化训练,通过多种增强方法结合点云信息还原的方式实现了块度图像的自动筛选与关键区域裁剪。现场应用表明:爆破块度智能分析平台能够实现爆破块度的快速自动识别和数据反馈,碎石识别覆盖率达95.5%,识别速度为100 ms/张。有效提高了湖山铀矿生产调度和管理水平,形成了爆破块度实时采集、信息汇聚和效果评估全流程智能作业体系。In the production process of open-pit mines,accurate acquisition of blasting fragmentation information is very important for rock shoveling and processing.Aiming at the problem that the traditional blasting fragmentation measurement technology is complex and time-consuming,and cannot be directly applied to mine production operations,an intelligent identification platform for blasting fragmentation is developed,which integrates the dynamic identification,real-time transmission and static analysis functions of rock fragmentation in mine site.The core algorithm of the intelligent analysis platform is constructed by combining binocular depth vision and Yolov8-seg model.The system is deployed in the ore rock transportation path,and the image of the blasting pile in the mine truck bucket is automatically collected,and the data interaction between the system and the production scheduling system is opened through the Restful standard interface protocol.Two core business sectors with onsite block dynamic recognition function and static image analysis function were established and successfully applied in Husab Mine.In the process of using,the influencing factors such as interference objects,shadows and strong light are trained in a targeted manner.Through a variety of enhancement methods combined with point cloud information restoration,the automatic screening and key area clipping of block images are realized.The field application shows that the intelligent analysis platform of blasting fragmentation can realize the rapid automatic identification and data feedback of blasting fragmentation.The coverage rate of gravel identification is 95.5%,and the recognition speed is 100 ms/sheet.It effectively improves the production scheduling and management level of Hushan uranium mine,and forms the whole process intelligent operation system of real-time collection,information aggregation and effect evaluation of blasting fragmentation.

关 键 词：铀矿 爆破块度 智能识别 数据协同 智能爆破 

分 类 号：TD85-9[矿业工程—金属矿开采]