基于傅里叶红外光谱预测翅荚木顺纹抗压强度  

Prediction for Compressive Strength Parallel to Grain of Zenia insignis Plantation Based on Fourier Infrared Spectroscopy

作　　者：陈博文 胡娟 金咏琪 郑安 颛孙浩 林金国[1] 关鑫[1] Chen Bowen;Hu Juan;Jin Yongqi;Zheng An;Zhuan Sunhao;Lin Jinguo;Guan Xin(College of Materials Engineering,Fujian Agriculture and Forestry University,Fuzhou Fujian 350108,China;School of Software and Internet of Things Engineering,Jiangxi University of Finance and Economics,Nanchang Jiangxi 330013,China)

机构地区：[1]福建农林大学材料工程学院,福建福州350108 [2]江西财经大学软件与物联网工程学院,江西南昌330013

出　　处：《西南林业大学学报（自然科学）》2022年第4期178-183,共6页Journal of Southwest Forestry University：Natural Sciences

基　　金：福建省林业科技推广项目(2021TG13)资助。

摘　　要：以傅里叶红外光谱研究为基础,测量翅荚木样本抗压强度与光谱数据,剔除异常值,选择多元散射校正对光谱数据进行预处理,采用SPA算法对光谱波数进行选择,利用PLS建立抗压强度预测模型,加入试样的厚度与宽度作为自变量进行比较。结果表明:采用SPA算法对预处理后的光谱数据进行波段选择可以加强光谱的预测能力,最后预测模型的决定系数与预测标准偏差分别为0.900 0、1.536 6,能够满足对翅荚木无损检测的需求,若添加样本的宽度与厚度作为自变量,建立的预测模型决定系数与预测标准偏差分别为0.904 6、0.932 5与1.710 9、1.368 5。Based on Fourier infrared spectroscopy, the compressive strength and spectral data of the Zenia insignis samples were measured. After eliminating the outliers, the spectral data were preprocessed by multiple scattering correction, and the SPA algorithm was used to select the spectral wave number. The compressive strength prediction model was established by using PLS, and the thickness and width of the samples were added as independent variables for comparison. Results show that after pretreatment and the SPA algorithm is adopted to the spectral data of band selection can strengthen the prediction ability of the spectrum, the prediction model of standard deviation determination coefficient R2 and forecasting SEP 0.900 0 and 1.536 6, respectively, can satisfy the demand of Z. insignis nondestructive testing, if add the width and thickness of the sample as the independent variable, establish the forecast model of R2 and SEP were 0.904 6, 0.932 5 and 1.710 9, 1.368 5.

关 键 词：翅荚木 速生树种 傅里叶红外光谱 顺纹抗压强度 连续投影算法 

分 类 号：S781.2[农业科学—木材科学与技术] O657.33[农业科学—林学]