基于LASSO法的桉木-相思混合制浆原料近红外分析模型的建立

doi:10.3969/j.issn.0253-2417.2020.05.012

林产化学与工业 ›› 2020, Vol. 40 ›› Issue (5): 83-90.doi: 10.3969/j.issn.0253-2417.2020.05.012

基于LASSO法的桉木-相思混合制浆原料近红外分析模型的建立

吴珽^1,²(),梁龙¹,朱北平¹,邓拥军¹,房桂干^1,*()

1. 中国林业科学研究院林产化学工业研究所; 生物质化学利用国家工程实验室; 国家林业和草原局林产化学工程重点实验室; 江苏省生物质能源与材料重点实验室; 江苏省林业资源高效加工利用协同创新中心, 江苏南京 210042
2. 金东纸业(江苏)股份有限公司, 江苏镇江 212132

收稿日期:2020-06-03 出版日期:2020-10-28 发布日期:2020-10-31
通讯作者: 房桂干 E-mail:wuting@icifp.cn;fangguigan@icifp.cn
作者简介:吴珽(1988-),男,江苏兴化人,助理研究员,博士,研究方向为制浆造纸清洁生产; E-mail:wuting@icifp.cn
基金资助:
中国博士后科学基金资助项目(2019M661780);国家重点研发计划资助项目(2017YFD0601005)

Establishment of Near Infrared Analysis Models of Eucalyptus-Acacia Mixed Pulpwood Materials Based on LASSO Algorithm

Ting WU^1,²(),Long LIANG¹,Beiping ZHU¹,Yongjun DENG¹,Guigan FANG^1,*()

1. Institute of Chemical Industry of Forest Products, CAF; National Engineering Lab. for Biomass Chemical Utilization; Key Lab. of Chemical Engineering of Forest Products, National Forestry and Grassland Administration; Key Lab. of Biomass Energy and Material, Jiangsu Province; Jiangsu Co-Innovation Center of Efficient Processing and Utilization of Forest Resources, Nanjing 210042, China
2. Gold East Paper(Jiangsu) Co., Ltd., Zhenjiang 212132, China

Received:2020-06-03 Online:2020-10-28 Published:2020-10-31
Contact: Guigan FANG E-mail:wuting@icifp.cn;fangguigan@icifp.cn

摘要/Abstract

摘要：

为提高国内木浆产量和利用率，缓解制浆造纸行业优质原料紧缺的现状，针对我国南方特定的制浆原料模式——桉木-相思混合原料进行快速分析研究。采集了175个桉木-相思混合原料和45个单一材种原料的近红外光谱，明确其混合程度和化学成分含量。通过平滑、矢量归一化（V-Norm）、多元散射校正（MSC）、一阶导数（1st Der）和二阶导数（2nd Der）等方法组合预处理原始光谱，结合LASSO算法，确定了混合程度（桉木质量分数）、综纤维素、Klason木质素、聚戊糖、苯醇抽出物和1% NaOH抽出物含量分析模型的建模方法，并建立了相应的模型。建模过程中最优调整参数（μ）分别为13.62、18.30、6.39、9.64、7.49和12.07。6个模型的预测均方根误差（RMSEP）值分别为1.93%、0.61%、0.51%、0.80%、0.28%和0.41%。绝对偏差范围（AD）分别为-3.19%~3.24%、-0.96%~1.01%、-0.89%~0.84%、-1.37%~1.46%、-0.43%~0.39%和-0.58%~0.60%。模型适应性好，能够满足制浆造纸工业需求，同时也证实了LASSO法用于混合制浆原料分析的可行性。

关键词: 近红外光谱, 制浆造纸, LASSO算法, 预处理, 桉木-相思

Abstract:

In order to improve the domestic pulp output and utilization rate, and alleviate the shortage of high-quality materials in the pulping and paper-making industry, a rapid analysis research was conducted on the specific material model of pulp making-Eucalyptus-Acacia mixed with pulpwood in southern China. The near-infrared spectra of 175 Eucalyptus-Acacia mixed samples and 45 single-species samples were collected. The mixing degree and chemical composition content of all samples were analyzed. The original spectra were preprocessed by the combined pretreatment methods of smoothing, vector normalization(V-Norm), multiple scattering correction(MSC), first derivative(1st Der) and second derivative(2nd Der). Combined with the least absolute shrinkage and selection operator(LASSO) algorithm, the mixing degree, holocellulose, Klason lignin, pentosan, benzene-alcohol extractives and 1% NaOH extractives content models were built. The optimal adjustment parameters determined during the modeling process were 13.62, 18.30, 6.39, 9.64, 7.49, and 12.07. The RMSEP values of the six models were 1.93%, 0.61%, 0.51%, 0.80%, 0.28%, and 0.41%, respectively. The absolute deviation ranges were -3.19%-3.24%, -0.96%-1.01%, -0.89%-0.84%, -1.37%-1.46%, -0.43%-0.39%, -0.58%-0.60%. The models have good adaptability and could meet the needs of the pulping and paper-making industry. It also confirmed the feasibility of the LASSO algorithm for the analysis of mixed pulpwood materials.

Key words: near-infrared spectroscopy, pulping and paper-making, LASSO algorithm, pretreatment methods, Eucalyptus-Acacia

中图分类号:

TQ35
TS721

吴珽,梁龙,朱北平,邓拥军,房桂干. 基于LASSO法的桉木-相思混合制浆原料近红外分析模型的建立[J]. 林产化学与工业, 2020, 40(5): 83-90.

Ting WU,Long LIANG,Beiping ZHU,Yongjun DENG,Guigan FANG. Establishment of Near Infrared Analysis Models of Eucalyptus-Acacia Mixed Pulpwood Materials Based on LASSO Algorithm[J]. Chemistry and Industry of Forest Products, 2020, 40(5): 83-90.

图/表 7

表1

表2

表3

图1

表4

LASSO算法结合不同预处理方法建立模型情况"

模型models	预处理方法pretreatment methods	最优调整参数μ the optimum adjusting parameter	R_cv²	RMSECV/%
混合程度 mixing degree	平滑smoothing,矢量归一化V-Norm,一阶导数1st Der	13.62	0.9970	1.63
	平滑smoothing,矢量归一化V-Norm,二阶导数2nd Der	10.50	0.9967	1.72
	平滑smoothing,多元散射校正MSC,一阶导数1st Der	7.61	0.9966	1.75
	平滑smoothing,多元散射校正MSC,二阶导数2nd Der	11.48	0.9969	1.67
综纤维素 holocellulose	平滑smoothing,矢量归一化V-Norm,一阶导数1st Der	18.30	0.9370	0.53
	平滑smoothing,矢量归一化V-Norm,二阶导数2nd Der	15.41	0.9273	0.57
	平滑smoothing,多元散射校正MSC,一阶导数1st Der	19.08	0.9316	0.55
	平滑smoothing,多元散射校正MSC,二阶导数2nd Der	11.42	0.9240	0.58
Klason木质素 Klason lignin	平滑smoothing,矢量归一化V-Norm,一阶导数1st Der	8.40	0.9049	0.50
	平滑smoothing,矢量归一化V-Norm,二阶导数2nd Der	12.31	0.9166	0.47
	平滑smoothing,多元散射校正MSC,一阶导数1st Der	6.16	0.9118	0.48
	平滑smoothing,多元散射校正MSC,二阶导数2nd Der	6.39	0.9238	0.45
聚戊糖 pentosan	平滑smoothing,矢量归一化V-Norm,一阶导数1st Der	11.20	0.9490	0.76
	平滑smoothing,矢量归一化V-Norm,二阶导数2nd Der	9.64	0.9592	0.68
	平滑smoothing,多元散射校正MSC,一阶导数1st Der	10.92	0.9537	0.72
	平滑smoothing,多元散射校正MSC,二阶导数2nd Der	12.45	0.9535	0.72
苯醇抽出物 benzene-alcohol extractives	平滑smoothing,矢量归一化V-Norm,一阶导数1st Der	6.81	0.9514	0.27
	平滑smoothing,矢量归一化V-Norm,二阶导数2nd Der	5.60	0.9560	0.26
	平滑smoothing,多元散射校正MSC,一阶导数1st Der	7.49	0.9596	0.25
	平滑smoothing,多元散射校正MSC,二阶导数2nd Der	11.75	0.9528	0.27
1% NaOH抽出物 1% NaOH extractives	平滑smoothing,矢量归一化V-Norm,一阶导数1st Der	12.07	0.9310	0.36
	平滑smoothing,矢量归一化V-Norm,二阶导数2nd Der	14.13	0.9163	0.39
	平滑smoothing,多元散射校正MSC,一阶导数1st Der	9.46	0.9273	0.37
	平滑smoothing,多元散射校正MSC,二阶导数2nd Der	9.72	0.9237	0.37

表4

表5

图2

参考文献 22

1	中国造纸协会.中国造纸工业2019年度报告[R].北京:中国造纸协会, 2020, 5: 13-14.
	China Paper Association.China Paper Industry Annual Report 2019[R]. Beijing: China Paper Association, 2020, 5: 13-14.
2	邓拥军.桉木化学机械法制浆节能提质技术及作用机理的研究[D].北京:中国林业科学研究院, 2015: 2-5.
	DENG Y J.Energy-saving and pulp quality improvement measures for eucalypt woodchips chemimechanical pulping process and relative effectiveness mechanisms[D]. Beijing: Chinese Academy of Forestry, 2015: 2-5.
3	常俊婷. 广西地区种植品种变化对造纸原料市场的影响[J]. 造纸信息, 2015, 5, 48.
	CHANG J T . The impact of varieties of planting varieties in Guangxi on the papermaking material market[J]. China Paper Newsletters, 2015, 5, 48.
4	璩爱玉, 房桂干. 速生材制浆性能与其材性的多元非线性回归分析及预测方法[J]. 林业科学, 2009, 45 (10): 113- 119.
	QU A Y , FANG G G . Forecasting methods of pulping properties of fast-growing woods with multiple nonlinear regression[J]. Scientia Silvae Sinicae, 2009, 45 (10): 113- 119.
5	TSUCHIKAWA S , SCHWANNINGER M . A review of recent near-infrared research for wood and paper (part 2)[J]. Applied Spectroscopy Reviews, 2013, 48 (7): 560- 587. doi: 10.1080/05704928.2011.621079
6	NABAVI M , DAHLEN J , SCHIMLECK L . Regional calibration models for predicting loblolly pine tracheid properties using near-infrared spectroscopy[J]. Wood Science and Technology, 2018, 52, 445- 463. doi: 10.1007/s00226-018-0986-1
7	TALENS P , MORA L , MORSY N , et al. Prediction of water and protein contents and quality classification of Spanish cooked ham using NIR hyperspectral imaging[J]. Journal of Food Engineering, 2013, 117 (3): 272- 280. doi: 10.1016/j.jfoodeng.2013.03.014
8	朱玉慧, 闻靓, 张耀丽, 等. 杨木应拉木微区结构可视化及化学成分分析[J]. 林业工程学报, 2020, 5 (3): 54- 58.
	ZHU Y H , WEN L , ZHANG Y L , et al. Visualization of micro-structure and chemical composition of poplar tension wood[J]. Journal of Forestry Engineering, 2020, 5 (3): 54- 58.
9	莫军前, 张文博. 基于近红外光谱技术的热处理竹材物理力学性能[J]. 林业工程学报, 2019, 4 (1): 32- 38.
	MO J Q , ZHANG W B . Physical and mechanical properties of heat-treated bamboo using near infrared reflectance spectroscopy[J]. Journal of Forestry Engineering, 2019, 4 (1): 32- 38.
10	BALABIN R M , SAFIEVA R Z , LOMAKINA E I . Near-infrared (NIR) spectroscopy for motor oil classification:From discriminant analysis to support vector machines[J]. Microchemical Journal, 2011, 98 (1): 121- 128. doi: 10.1016/j.microc.2010.12.007
11	赵阅书, 薛晓明, 宋小娇, 等. 6种阔叶树材红外光谱特征的比较[J]. 林业工程学报, 2019, 4 (5): 40- 45.
	ZHAO Y S , XUE X M , SONG X J , et al. Comparison and analysis of FT-IR spectra for six broad-leaved wood species[J]. Journal of Forestry Engineering, 2019, 4 (5): 40- 45.
12	赵荣军, 霍小梅, 邢新婷, 等. 粗皮桉木材气干密度测定方法比较研究[J]. 西北林学院学报, 2012, 27 (2): 242- 244.
	ZHAO R J , HUO X M , XING X T , et al. Comparison of measurement methods of eucalyptus pellita air-dry density[J]. Journal of Northwest Forestry University, 2012, 27 (2): 242- 244.
13	冷春晖, 易敏, 张露, 等. 材性快速评估技术在刨花楠活立木木材密度和弹性模量预测中的应用[J]. 林业工程学报, 2018, 3 (6): 56- 61.
	LENG C H , YI M , ZHANG L , et al. Application of wood properties rapid assessment technology in prediction of wood density and modulus of elastic of standing tree of Machilus pauhiol[J]. Journal of Forestry Engineering, 2018, 3 (6): 56- 61.
14	HE W M , HU H R . Prediction of hot-water-soluble extractive pentosan and cellulose content of various wood species using FT-NIR spectroscopy[J]. Bioresource Technology, 2013, 140 (7): 299- 305.
15	刘耀瑶, 熊智新, 王勇, 等. 不同型号便携式光谱仪间木质素近红外光谱分析模型传递研究[J]. 林业工程学报, 2019, 4 (4): 93- 98.
	LIU Y Y , XIONG Z X , WANG Y , et al. Study on the transform of near-infrared calibration models for lignin determination between different types of portable near-infrared spectrometers[J]. Journal of Forestry Engineering, 2019, 4 (4): 93- 98.
16	YU S H , ZHANG J . The study on impact factors of foreign direct investment based on Lasso[J]. Journal of Hunan University (Social Sciences), 2014, 28 (2): 53- 56.
17	SHI W F , HU X G , YU K . K-part Lasso based on feature selection algorithm for high-dimensional data[J]. Computer Engineering and Applications, 2012, 48 (1): 157- 161.
18	吴珽.基于NIR的制浆材材性快速检测方法研究[D].北京:中国林业科学研究院, 2018: 103-104.
	WU T.Study on rapid measurement method for pulpwood characteristics using the near-infrared techniques[D]. Beijing: Chinese Academy of Forestry, 2018: 103-104.
19	TIBSHIRANI R J . The Lasso method for variable selection in the cox model[J]. Statistics in Medicine, 1997, 385- 395.
20	KRASZNAI D J , CHAMPAGNE P , CUNNINGHAM M F . Quantitative characterization of lignocellulosic biomass using surrogate mixtures and multivariate techniques[J]. Bioresource technology, 2012, 110 (5): 652- 661.
21	BAILLERES H , DAVRIEUX F , HAM-PICHAVANT F . Near infrared analysis as a tool for rapid screening of some major wood characteristics in a Eucalyptus breeding program[J]. Annals of Forest Science, 2002, 59 (5/6): 479- 490.
22	MCLAUCHLIN A R , GHITA O , GAHKANI A . Quantification of PLA contamination in PET during injection moulding by in-line NIR spectroscopy[J]. Polymer Testing, 2014, 38 (18): 46- 52.

树种 wood species	树龄/a age	来源 sources
尾巨桉Eucalyptus urophylla×grandis	6	广东省湛江市Zhanjiang, Guangdong
尾叶桉U6E. urophylla U6	5	广东省湛江市Zhanjiang, Guangdong
尾叶桉L11 E. urophylla L11	5	云南省昆明市Kunming, Yunnan
蓝桉E. globules	6	广西省钦州市Qinzhou, Guangxi
马占相思Acacia mangium	6	海南省海口市Haikou, Hainan
厚荚相思A. crassicarpa	6	广西省南宁市Nanning, Guangxi
纹荚相思A. aulacocarpa	6	广西省南宁市Nanning, Guangxi
大叶相思A. auriculiformis	6	广西省钦州市Qinzhou, Guangxi
杂交相思A.auriculiformis×A. mangium	6	广西省南宁市Nanning, Guangxi

样品数 number	桉木质量分数/% eucalyptus mass fraction	最小值/% minimum	最大值/% maximum	均值/% mean	中位数/% median	标准偏差/% SD
25	0	0	0	0	0	0
35	0~20	0.19	19.88	10.03	9.96	6.42
35	20~40	20.11	39.96	29.85	29.95	6.11
35	40~60	40.24	59.89	50.17	50.13	6.67
35	60~80	60.03	79.92	70.06	70.22	6.23
35	80~100	80.09	99.81	89.65	89.58	6.26
20	100	100	100	100	100	0

化学成分 chemical composition	最小值/% minimum	最大值/% maximum	均值/% mean	中位数/% median	标准偏差/% SD
综纤维素holocellulose	73.85	81.26	77.55	77.60	2.18
Klason木质素Klason lignin	21.46	27.53	24.88	25.22	1.68
聚戊糖pentosan	17.94	29.97	24.52	24.99	3.29
苯醇抽出物benzene-alcohol extractives	0.74	4.81	2.72	2.74	1.23
1% NaOH抽出物1% NaOH extractives	11.39	16.24	14.01	14.06	1.38

模型models	R_val²	RMSEP	RPD	AD
混合程度mixing degree	0.9958	1.93%	15.44	-3.19%~3.24%
综纤维素holocellulose	0.9160	0.61%	3.45	-0.96%~1.01%
Klason木质素Klason lignin	0.9010	0.51%	3.12	-0.89%~0.84%
聚戊糖pentosan	0.9426	0.80%	4.17	-1.37%~1.46%
苯醇抽出物benzene-alcohol extractives	0.9473	0.28%	4.36	-0.43%~0.39%
1% NaOH抽出物1% NaOH extractives	0.9085	0.41%	3.31	-0.58%~0.60%

[1]	唐保山,李坤,张弘,张雯雯,关庆芳,史正军. 基于近红外法的紫胶原胶主组分快速预测模型构建[J]. 林产化学与工业, 2020, 40(4): 47-56.
[2]	储秋露,陈雪艳,宋凯,王静,张晓东,施爱平. 两步法预处理对杨木酶水解及木质素吸附性能的影响[J]. 林产化学与工业, 2019, 39(6): 68-74.
[3]	黄晨, 武新星, 赖晨欢, 黄曹兴, 李鑫, 勇强. 高灰分麦糠水热预处理及其酶水解性能研究[J]. 林产化学与工业, 2019, 39(5): 26-32.
[4]	陈雪艳, 储秋露, 王静, 张晓东, 施爱平. 两步法预处理脱除木质素对杨木酶水解的影响[J]. 林产化学与工业, 2019, 39(5): 73-79.
[5]	张浩楠, 韦家慧, 赵辉, 李丹, 任浩. 浒苔纤维聚丙烯酸吸水树脂的制备及性能分析[J]. 林产化学与工业, 2019, 39(5): 101-107.
[6]	乔慧,欧阳水平,刘蕾,欧阳嘉. 低浓度乙酸预处理玉米芯的工艺研究[J]. 林产化学与工业, 2019, 39(1): 81-87.
[7]	冯彦洪, 王端, 肖和平, 何和智, 瞿金平. 螺杆蒸汽爆破预处理对稻秆结构及其无胶板性能的影响[J]. 林产化学与工业, 2018, 38(3): 69-74.
[8]	董益帆, 刘志佳, 兰天晴. 预处理对蔗渣吸附纤维素酶特性的影响[J]. 林产化学与工业, 2018, 38(2): 75-79.
[9]	李娇阳, 黄吉振, 刘玉新, 孙兵, 梁星, 张宝. 漆酶预处理前后蔗渣半纤维素的提取及性质研究[J]. 林产化学与工业, 2018, 38(1): 121-126.
[10]	周静, 沈葵忠, 房桂干, 梁龙, 田庆文, 李漫. 不同预处理方法对麦草化学组分及其酶解性能的影响[J]. 林产化学与工业, 2017, 37(5): 53-60.
[11]	江明珠, 韩邦兴, 颜辉, 贾俊强, 屠洁, 马永昆. 便携式近红外光谱仪快速无损鉴别霍山石斛枫斗和河南石斛枫斗[J]. 林产化学与工业, 2017, 37(5): 101-106.
[12]	何士成, 孙曼钰, 孙忠岩, 舒月力, 贾士儒, 钟成. 蒸汽爆破与碱法协同预处理对小麦秸秆结构及酶解的影响[J]. 林产化学与工业, 2017, 37(5): 126-132.
[13]	李士伟, 杨静翎, 李辉, 谭梦娇, 黄忠良, 李昌珠. 水热预处理对樟木屑制备成型燃料行为的影响[J]. 林产化学与工业, 2017, 37(4): 129-136.
[14]	李水芳, 付红军, 马强, 单杨. 构建三种木本油料植物种子含油率NIR通用模型的可行性研究[J]. 林产化学与工业, 2017, 37(4): 137-142.
[15]	吴博士, 张逊, 杨俊, 许凤. 亚氯酸盐预处理杉木细胞壁木质素溶解机理研究[J]. 林产化学与工业, 2017, 37(3): 38-44.

基于LASSO法的桉木-相思混合制浆原料近红外分析模型的建立

Establishment of Near Infrared Analysis Models of Eucalyptus-Acacia Mixed Pulpwood Materials Based on LASSO Algorithm

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 22

相关文章 15

编辑推荐

Metrics

本文评价

关于本刊

投稿指南

在线期刊

相关单位

联系我们