基于最小二乘支持向量机的油页岩含油率近红外光谱分析

发布时间：2018-04-24 11:34

本文选题：最小二乘支持向量机 + 油页岩　；参考：《高等学校化学学报》2016年10期

【摘要】：为了提高油页岩含油率近红外光谱分析建模的预测精度和稳定性,开展了基于最小二乘支持向量机(LS-SVM)建模方法的对比研究.采用主成分-马氏距离(PCA-MD)和基于蒙特卡洛采样(MCS)2种方法进行了奇异样本的检测,采用径向基核函数的LS-SVM、偏最小二乘(PLS)和反向传播神经网络(BPANN)3种方法进行建模方法对比.结果表明,对于64个油页岩岩芯样本,与PCA-MD方法相比,采用MCS方法剔除奇异样本后所建PLS模型的预测精度提高了28%.对于MCS方法剔除奇异样本后的58个样品,采用KennardStone法划分了44个样品的校正集和14个样品的预测集,采用2阶导数和标准化预处理方法,建立了100个LS-SVM的校正模型,模型的预测决定系数R2平均值达到0.90以上,高于PLS和BPANN模型的对应值;且R2的变化量(0.02)小于BPANN模型的对应值(0.32).因此,MCS奇异样本检测结合LS-SVM方法可提高油页岩含油率样本建模的精度和稳定性.
[Abstract]:In order to improve the prediction accuracy and stability of oil shale oil content near infrared spectroscopy (NIR) modeling, a comparative study on the modeling method based on least squares support vector machine (LS-SVM) was carried out. Two methods, principal component Markov distance (PCA-MD) and Monte Carlo sampling (MCSN), are used to detect singular samples. Three modeling methods, LS-SVM, partial least squares (PLS) of radial basis function (RBF) and backpropagation neural network (BPANNN), are compared. The results show that for 64 oil shale core samples, compared with PCA-MD method, the prediction accuracy of PLS model established by MCS method after eliminating singular samples is improved by 28%. For 58 samples which were excluded by MCS method, the calibration sets of 44 samples and the prediction sets of 14 samples were divided by KennardStone method. The correction model of 100 LS-SVM was established by using the second order derivative and standardized pretreatment method. The average predictive decision coefficient R2 of the model is above 0.90, which is higher than the corresponding value of PLS and BPANN model, and the variation of R2 is 0.02) less than the corresponding value of BPANN model. Therefore, the accuracy and stability of oil shale oil content sample modeling can be improved by using MCS singular sample detection and LS-SVM method.
【作者单位】：吉林大学仪器科学与电气工程学院;
【基金】：国家潜在油气资源(油页岩勘探开发利用)产学研用合作创新子课题(批准号:OSR-02-04) 吉林省科技发展计划项目重大科技专项(批准号:20116014)资助~~
【分类号】：O657.33;TE662.3

【相似文献】