机器学习算法用于水泥强度预测的研究进展  

Research Progress of Machine Learning Algorithm for Cement Strength Prediction

作　　者：李自强 崔素萍[1] 马忠诚 王亚丽[1] 王晶 刘云 乔志杨 LI Ziqiang;CUI Suping;MA Zhongcheng;WANG Yali;WANG Jing;LIU Yun;QIAO Zhiyang(College of Materials Science and Engineering,Beijing University of Technology,Beijing 100124,China;China Building Materials Academy Co.,Ltd.,Beijing 100024,China;China National Building Materials Group Corporation,Beijing 100036,China)

机构地区：[1]北京工业大学材料科学与工程学院,北京100124 [2]中国建筑材料科学研究总院有限公司,北京100024 [3]中国建材集团有限公司,北京100036

出　　处：《材料导报》2025年第5期191-202,共12页Materials Reports

基　　金：中国建材集团原创技术策源地“揭榜挂帅”项目(2021YCJS01-4)。

摘　　要：水泥的强度是衡量其性能的重要指标之一,传统的基于人工定时采样测定的方法准确性好但存在较大的滞后性,不能及时调控水泥生产过程。机器学习可通过不同算法对水泥等流程工业的原料及产品检测数据、微观结构图像、工艺运行参数等多维生产数据进行有目标的关联分析,建立水泥强度预测模型,可以解决人工检测方法的滞后性问题。本文通过梳理常用算法的基本工作原理和优势,归纳基于机器学习的水泥强度预测模型,探讨其应用效果和发展方向,以期为水泥强度预测模型的进一步优化和在水泥工业中的应用提供参考。The strength of cement is one of the important indicators of its performance.The traditional strength detection method based on manual timed sampling has a good accuracy,but there is a large lag,which can not regulate the cement production process in time.Machine learning en-ables targeted correlation analysis of multi-dimensional production data in the cement industry,for example,raw material and product test data,microstructure images and process operating parameters.The lagging problem of manual testing methods can be solved by establishing a cement strength prediction model based on machine learning.This paper summarizes the cement strength prediction model based on machine learning by sorting out the basic working principles and advantages of commonly used algorithms,and discusses its application effect and development direc-tion,with a view to providing guidance for further optimization of the cement strength prediction model and its application in the cement industry.

关 键 词：水泥强度 大数据分析 机器学习 预测模型 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TQ172[自动化与计算机技术—控制科学与工程]