基于近红外漫反射光谱无损检测梨可溶性固形物的光强影响探究

Study on the Influence of Light Intensity on Near-Infrared Diffuse Reflectance Spectra of Pear Soluble Solids Content

通过设置四种不同的光源强度研究光强对近红外漫反射无损检测梨可溶性固形物的影响,对四种类别光强的光谱定性分析显示四类光谱差异微小,肉眼几乎无法辨别。在进一步的定量分析中,通过主成分分析、逐步线性回归分析以及偏最小二乘法分析的比较,主成分分析(r值跨度:0.253～0.606;RMSEC值跨度:0.549～0.614;RMSEP值跨度:0.455～0.752)与逐步线性回归分析(r值跨度:0.249～0.551;RMSEC值跨度:0.536～0.624;RMSEP值跨度:0.646～0.734)得到的模型较差。通过对光谱进行一阶求导和二阶求导预处理,主成分分析与逐步线性回归分析建模结果仍不理想。通过二阶求导预处理,偏最小二乘法所建的模型得到优化,其中相关系数r值跨度为0.947～0.970,混合模型的相关系数r值达到了0.95 7,分析结果表明光强对梨的近红外漫反射光谱无损检测可溶性固形物的影响差异不大,为光谱仪的田间作业奠定了基础。

The objective of the present research was to study the influence of four kinds of light intensity of near-infrared diffuse reflectance spectra on the determination of pear soluble solids content. Qualitative analysis of the four spectra showed that the differences between them were too slight to find. After further analysis by principal component regression （PCR）, stepwise mul- tiple linear regression （SMLR） and partial least squares regression （PLSR） as well as comparisons among them, the results of PCR model and SMLR models which had low correlation coefficients （r） were worse than those of PLSR model. The model based on PLSR was further optimized by the second order derivative pretreatments of the spectra, and its correlation coefficients ranged from 0. 947 to 0. 970. The relative coefficients in PLSR model based on spectra from four sets were 0. 958. The results indicate that light intensity has little influence on the determination of pear soluble solids content based on near-infrared diffuse reflection spectra.