主成分——逐步回归——BP算法在近红外光谱定量分析中应用的研究

Study of Near-Infrared Spectra for Quantitative AnalysisUsing Principal Component-stepwise Regression-BP Algorithm

本文提出一种新的Ｂ－Ｐ算法：主成分—逐步回归—ＢＰ算法．用于近红外光谱定量分析中，其所建模型的预测精度明显高于以往的逐步回归—ＢＰ算法和主成分—ＢＰ算法．

Back Propagation network (B-P) has been used in Near-Infrared spectra (NIR) for many years.But the important thing is how to improve its prediction precision.In this study,we put forward Principal component-stepwise regression-BP algorithm(PCR-SRA-BP).It is superior to SRA-BP and PCR-BP algorithm in prediction and the speed of learning.