基于Si-cPLS的小麦种子发芽率近红外模型优化研究

为了提高近红外光谱技术快速测定小麦种子发芽率的准确度和稳健性,比较分析了基于全光谱的单一偏最小二乘(PLS)模型和多模型共识PLS模型(cPLS)性能,并提出了基于特征光谱的多模型共识PLS模型(Si-cPLS)。实验收集84份小麦种子,通过SPXY法将样本集划分为训练集样本66个,预测集样本18个。从训练集中随机抽取50个样本作为校正集建立一系列PLS子模型,选取其中性能较好100个子模型作为成员模型建立cPLS模型,取成员模型预测结果的均值来分析未知样本。在此基础上,采用组合间隔偏最小二乘法(SiPLS)筛选特征谱区建立多模型共识PLS模型(Si-cPLS)。各模型均采用均值中心化预处理方法,PLS模型、cPLS模型以及Si-cPLS模型对预测集的小麦种子发芽率进行50次重复预测的平均相关系数r分别为0.935,0.949和0.967,平均预测均方根误差RMSEP分别为13.735%,12.533%和10.273%,RMSEP的标准偏差分别为1.144%,0.096%和0.08%。实验结果表明cPLS模型较单一PLS模型更加稳定可靠,而基于特征光谱的Si-cPLS模型则进一步提高了cPLS的稳定性与预测精度,为建立性能优秀的小麦种子发芽率近红外模型提供了新思路。

Research on Optimization of Wheat Seed Germination Rate NIR Model Based on Si-cPLS

To improve the detecting accuracy and robustness of wheat seed germination rate with near infrared spectroscopy technique,singl e PLS model and consensus PLS model(cPLS) developed on the full-spectral were c ompared and analyzed,thus,the Si-cPLS model which developed on the characteri stic spectral regions was put forward.There were 84 samples partitioned into 66training samples and 18 prediction samples using SPXY method.By randomly selec ting 50 samples from training set as calibration set,a series of sub PLS models were build.100 PLS sub models satisfying predefined criterion were selected an d combined one cPLS model by averaging all predicted results.With this basis Si-cPLS model were developed on characteristic spectral regions selected with syn ergy interval method.Statistics on 50 repeat prediction of wheat seed germinati on by full-spectral PLS model,full-spectral cPLS model,and Si-cPLS model sh owed that,the mean correlation coefficient(R) were 0.901,0.922 and 0.936 re spectively,the mean RMSEP were 13.735％,12.533％ and 10.273％ respectively wit h standard deviation of RMSEP of 1.144％,0.096％ and 0.080％ respectively.Resu lts showed that cPLS model was more stable and reliable than single PLS model,W hile Si-cPLS could further increase the stability and prediction accuracy of cP LS model.