基于近红外光谱和ELM算法的菱镁矿石品级分类研究

发布时间：2018-04-03 01:16

本文选题：近红外光谱　切入点：菱镁矿　出处：《光谱学与光谱分析》2017年01期

【摘要】：由工业发展需求,针对菱镁矿石矿物含量不同以及分布不均匀而难以判定其品级的情况,提出一种由近红外光谱技术结合ELM的菱镁矿石品级分类模型。该模型可以实现菱镁矿石品级的快速分类。近红外光谱利用菱镁矿中不同种类含H基团对近红外光谱有不同吸收的特性,用来测定菱镁矿石的成分及其含量,其操作简便、不破坏样品、速度快、准确高效。以辽宁省营口市大石桥的菱镁矿石30组为研究对象,采集菱镁矿石的近红外光谱数据样本30×973。采用主成分分析(PCA)对其进行降维处理,以主元贡献率大于99.99%而得到10维的特征变量值。建立了ELM算法定量分析数学模型,取20组样本为训练样本(包括6组特级,14组非特),其余10组作为测试样本(其中4组特级,6组非特),ELM算法模型的隐含层节点数选取20。为了进一步提高分类效果,提出两种ELM算法模型的改进:采用循环模式对传统ELM的输入权值和阈值进行寻优的精选ELM和在精选ELM基础上进行集成的集成-精选ELM。并与用人工方法、化学方法和BP神经网络模型方法对菱镁矿石样品品级分类作对比。结果表明:近红外光谱和ELM菱镁矿石品级分类模型不论在时间上还是成本上,都具有明显的优势,且其准确率能够达到90%以上,为菱镁矿石品级分类提供了一条新的途径。
[Abstract]:According to the demand of industrial development, a classification model of magnesite grade based on Near-Infrared Spectroscopy (NIR) combined with ELM is proposed in view of the fact that the content of magnesite is different and the distribution of magnesite is not uniform, so it is difficult to judge the grade of magnesite.The model can realize fast classification of magnesite grade.Near-infrared spectroscopy (NIR) is used to determine the composition and content of magnesite by using different H-containing groups in magnesite with different absorption characteristics. It is easy to operate, does not destroy the sample, is fast, accurate and efficient.Taking 30 groups of magnesite from Dashiqiao, Yingkou City, Liaoning Province as the research object, the near infrared spectrum data of magnesite were collected.The principal component analysis (PCA) was used to reduce the dimensionality, and the characteristic variable value of 10 dimensions was obtained by using the principal component contribution rate greater than 99.99%.The mathematical model of quantitative analysis of ELM algorithm was established. Twenty groups of samples were taken as training samples (including 6 groups of special grade 14 groups of non-special test samples) and the other 10 groups of which were used as test samples.In order to further improve the classification effect, two improved ELM algorithm models are proposed: select ELM, which uses circular mode to optimize the input weights and thresholds of traditional ELM, and integrated ELM based on selected ELM.The classification of magnesite samples is compared with artificial method, chemical method and BP neural network model.The results show that both the near infrared spectrum and the ELM magnesite classification model have obvious advantages in both time and cost, and the accuracy rate can reach more than 90%, which provides a new way for magnesite classification.
【作者单位】：东北大学资源土木与工程学院;东北大学信息科学与工程学院;
【基金】：国家自然科学基金项目(41371437,61203214) 国家“十二五”科技支撑计划课题项目(2015BAB15B01)资助
【分类号】：P575;O657.33

【相似文献】