基于深度学习的水泥颗粒检测方法  

Cement Particle Detection Method Based on Deep Learning

作　　者：向科峰[1] 张津晨 刘自红[1] XIANG Kefeng;ZHANG Jinchen;LIU Zihong(School of Manufacturing Science and Engineering,Southwest University of Science and Technology,Mianyang 621010,Sichuan,China)

机构地区：[1]西南科技大学制造科学与工程学院,四川绵阳621010

出　　处：《西南科技大学学报》2023年第1期98-104,共7页Journal of Southwest University of Science and Technology

基　　金：国家工信部绿色制造系统集成项目(17zg0102)。

摘　　要：水泥生产过程中需要实时检测水泥颗粒粒径、圆度等几何参数。针对粒度离线检测效率低、参数调整滞后的现状,基于深度学习搭建并优化了水泥颗粒图像分割网络模型。实验表明:基于深度学习的水泥颗粒图像分割网络模型对水泥颗粒图像的分割精度达98%、分割准确度达94%,与离线检测的误差在8%以内。提高了分割精度和检测效率,满足水泥生产中过程控制智能化和信息化要求。In the process of cement production,geometric parameters such as cement particle size and roundness need to be measured in real time.Aiming at the low efficiency of granularity offline detection and the lag of parameter adjustment,the cement particle image segmentation network model was built and optimized based on deep learning.The experiment shows that the segmentation precision for cement particle images of cement particle image segmentation network model based on deep learning is 98%,the segmentation accuracy is 94%,and the error with offline detection is less than 8%.The detection method improves the segmentation precision and detection efficiency,and can meet the requirements of intellectualized and information-based process control in cement production.

关 键 词：深度学习 图像分割 颗粒检测 水泥 

分 类 号：TB44[一般工业技术]