第36卷，第2期 2016年2月
光谱学与光谱分析 Spectroscopy and Spectral Analysis
Vol. 36,No.2-pp593-598 February 2016
The Characteristic Spectral Selection Method Based on Forward and Backward Interval Partial Least Squares
QU Fang-fang', REN Dong'*,HOU Jin-jian'-2,ZHANG Zhong'
LU An-xiang,WANG Ji-hua-2,XU Hong-lei
1. College of Computer and Information Technology , Three Gorges University , Yichang443002, China
2. Beijing Research Center for Agricultural Standards and Testing , Beijing100097, China 3. Department of Mathematics and Statistics , Curtin University , Perth6845, Australis
Abstract In the near-infrared spectroscopy , the Forward Interval Partial Least Squares (FiPLS ) and Back ward Interval Partial Least Squares (BiPLS ) are commonly used modeling methods , which are based on the wavelength variable selection . These methods are usually of high prediction accuracy , but are strongly charac-teristic of greedy search , which causes that the intervals selected are not good enough to indicate the analyte in -formation . To solve the problem , a spectral characteristic intervals selection strategy (FB-iPLS ) based on the combination of FiPLS and BiPLS is proposed . On the basis of spectral segmentation , both FiPLSs are used to select useful intervals , and BiPLS is used to delete useless intervals , so as to perform the selection and deletion of the characteristic variables alternatively , which conducts a two-way choice of the target characteristic varia -bles , and is used to improve the robustness of the model . The experiments on determining the ethanol concen tration in pure water are conducted by modeling with FiPLS , BiPLS and the proposed method . Since different size of intervals will affect the result of the model , the experiments here will also examine the model results with different intervals of these three models . When the spectrum is divided into 60 segments , the FB-iPLS method obtains the best prediction performance . The correlation coefficients (r) of the calibration set and vali dation set are 0. 967 7 and 0. 967 O respectively , and the cross-validation root mean square errors (RMSECV) are 0. 088 8 and 0. 057 1 , respectively . Compared with FiPLS and BiPLS , the overall prediction performance of the proposed model is better . The experiments show that the proposed method can further improve the predic-tive performance of the model by resolving the greedy search feature against BiPLS and FiPLS , which is more efficient for and representative of the selection of characteristic intervals
KeywordsNear-Infrared Spectroscopy ; FiPLS ; BiPLS ; FB-iPLS ; Greedy search ; Characteristic intervals
中图分类号：0657.3
Introduction
文献标识码：A
DOI : 10. 3964 /j. issn. 1000-0593(2016 )02-0593-06
mation of the tested substance in samples (concentration , cat-egory + etc .) . It will give rise to spectral information overlap-ping and some redundant information including a lot of noises ,
Near-infrared spectroscopy contains a large number of absorption peaks of frequency doubling and frequency synthe-sis groups containing hydrogen , which can reflect the infor-
Received : 2014-11-25 ; aceepted : 2015-04-20
sample background and the like. It is difficult to eliminate them by preprocessingJ . If these data are involved in model building + which not only increases the computational com-
Foundation item: The National Science and Technology Projects in Rural Areas (2014BADO4B05 ), Natural Science Foundation of China
(41371349)
Biography : Q U Fang-fang , (1990), female , Master Degree Candidate in College of Computer and Information Technology , Three Gorges
University
e-mail : quff1128@ 163 .com
Corresponding author
e-mail : rendong5227@ 163 .com