欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
建立了一种基于独立成分分析的局部建模新方法，该方法首先将独立成分分析（ICA）用于近红外光谱的特征提取，然后，根据所提取的独立成分选择校正集中与预测样本相邻近的样本构成校正子集，建立局部偏最小二乘（PLS）回归模型并对预测样本进行预测。将所提出的方法应用于烟草样品中尼古丁含量的测定，所得结果优于常用的全局建模方法。
英文摘要：
A local regression method was proposed based on independent component analysis （ICA）. The method performs feature extraction from the measured near infrared （NIR） spectra by using ICA at first, and then carries out a subset selection for local regression according to the distance between a prediction sample and the calibration samples in the independent component （IC） space. Finally, by using the selected subset, a PLS model is built and used to predict the property of the prediction sample. An application of the proposed method in prediction of nicotine content in tobacco samples from NIR spectra was investigated, it was found that the results are better than both the local regression method based on principal component analysis （ PCA ） and the conventionally used global regression method.
同期刊论文项目
同项目期刊论文
期刊信息
位置：