化工矿物与加工2026，Vol.55Issue(8)：22-29,8.DOI:10.16283/j.cnki.hgkwyjg.2026.08.004
基于近红外反射光谱与机器学习的磷矿石品位检测
Phosphate rock grade detection based on near-infrared reflectance spectroscopy and machine learning
摘要
Abstract
In order to realize the green and efficient development and utilization of phosphate rock resources,a phos-phate rock grade detection and analysis method combining near-infrared reflectance spectroscopy and machine learning is proposed.High-grade apatite and low-grade dolomite were crushed and ground,and mixed in a variety of propor-tions as test samples.The near-infrared reflectance spectra of the test samples were collected using a fiber optic spec-trometer in the 900-2 600 nm band.The relationship between the characteristic absorption and scattering response characteristics of near-infrared light and the mineral components in the test samples was explored,and the quantitative analysis model of spectroscopy and phosphate rock grade was constructed.The results show that the combination of partial least squares regression and random forest has the highest accuracy of 99.79％and the strongest stability in predicting the grade of phosphate rock on the verification set.The combination of machine learning and near-infrared spectroscopy technology can effectively extract potential spectral information,realize real-time and accurate detection of phosphate rock grade,and provide a scientific detection method for efficient separation of phosphate rock.关键词
选矿/磷矿石/近红外光谱/机器学习/特征提取/偏最小二乘回归/随机森林Key words
mineral processing/phosphate rock/near infrared spectroscopy/machine learning/feature extraction/partial least squares/random forest分类
矿业与冶金引用本文复制引用
屈鑫,吝曼卿,施浪,田春满,王嵩,熊伦,卢永雄..基于近红外反射光谱与机器学习的磷矿石品位检测[J].化工矿物与加工,2026,55(8):22-29,8.基金项目
国家自然科学基金项目(52174085) （52174085）
湖北省安全生产专项资金科技项目(SJZX20220910) （SJZX20220910）
国家磷资源开发利用工程技术研究中心开放基金项目(NECP2022-08). （NECP2022-08）
