CAJ | 학술논문

采用偏最小二乘法(PLS)和光谱Savitzky-Golay(SG)平滑方法,建立甘蔗清糖浆锤度近红外光谱分析的优化模型.基于最优单波长模型预测效果划分定标集和预测集.全谱(400-2500nm)经过SG平滑处理后用PLS方法建模.建立计算机算法平台,把483种SG平滑模式和1-40的PLS因子数任意组合分别建立PLS模型,根据预测效果选出最优模型,最优模型的SG平滑模式为二阶导数平滑、4、5次多项式类型、43平滑点数,PLS因子数为13,预测均方根偏差(RMSEP)、相对预测均方根偏差(RRMSEP)和预测相关系数(rp)分别为0.433%、0.69%和0.978.预测精度很高,并且大幅度优于未做SG平滑处理直接PLS建模的预测效果.从而表明,SG平滑模式和PLS因子数的联合大范围筛选能够有效地应用于近红外光谱分析的模型优选.
채용편최소이승법(PLS)화광보Savitzky-Golay(SG)평활방법,건립감자청당장추도근홍외광보분석적우화모형.기우최우단파장모형예측효과화분정표집화예측집.전보(400-2500nm)경과SG평활처리후용PLS방법건모.건립계산궤산법평태,파483충SG평활모식화1-40적PLS인자수임의조합분별건립PLS모형,근거예측효과선출최우모형,최우모형적SG평활모식위이계도수평활、4、5차다항식류형、43평활점수,PLS인자수위13,예측균방근편차(RMSEP)、상대예측균방근편차(RRMSEP)화예측상관계수(rp)분별위0.433%、0.69%화0.978.예측정도흔고,병차대폭도우우미주SG평활처리직접PLS건모적예측효과.종이표명,SG평활모식화PLS인자수적연합대범위사선능구유효지응용우근홍외광보분석적모형우선.
Using partial least squares (PLS) and Savitzky-Golay (SG) smoothing method, the optimal near infrared spectral analysis model for sugarcane simple syrup brix was established.The calibration set and the prediction set were divided based on the prediction effect of the optimal single wavelength model.The PLS model was established based on the whole region (400-2500nm) after SG smoothing.By building the computer algorithm platform, for any combination of 483 SG smoothing modes and 1-40 PLS factors,the PLS models were constructed respectively.According to the prediction effect to select the optimal model,and the SG smoothing mode of the optimal model is the second derivative smoothing, 4 or 5 degree polynomial, 43 smoothing points, the PLS factor is 13, RMSEP,RRMSEP and rp are 0.433%, 0.69% and 0.978 respectively.The prediction accuracy is very high, and it is substantially superior to the prediction effect of the PLS mode| without SG smoothing, thus demonstrates that large-scale optimization combining SG smoothing modes and PLS factors can be effectively applied to the model optimization of near-infrared spectral analysis.