基于变量选择的蚕茧茧层量可见-近红外光谱无损检测

Non-destructive detection of cocoon shell weight based on variable selection by visible and near infrared spectroscopy

以蚕茧茧层量为研究对象,研究了基于可见-近红外光谱技术的蚕茧茧层量无损检测方法。采用最小二乘支持向量机(least square-support vector machine,LS-SVM)建立可见-近红外光谱模型。采用无信息变量消除算法(uninformative variable elimination,UVE)与连续投影算法(successive projections algorithm,SPA)相结合选取光谱有效波长。结果表明,基于UVE-SPA法进行变量选择,最终将原始光谱的600个光谱变量减少到了8个(673,937,963,982,989,992,995和1008nm)。基于此8个变量建立的LS-SVM模型得到了预测集的确定系数(Rp2)为0.5354,误差均方根(RMSEP)为0.0373的预测结果。表明可见-近红外光谱可以用于对蚕茧的茧层量进行无损检测,同时UVE-SPA是一种有效的光谱变量选择方法。

Visible and near-infrared reflectance spectroscopy（Vis-NIRS）was applied to measure cocoon shell weight. Least square-support vector machine（LS-SVM）was used to establish the Vis-NIR model.Uninformative variable elimination and successive projections algorithm were combined to select wavelength from Vis-NIR spectroscopy.Eight wavelength variables,namely 673,937,963,982,989,992,995 and 1 008 nm,were selected.The UVE-SPA-LS-SVM model was established based on these eight wavelength variables.The results showed that the determination coefficient for prediction set（Rp^2 ）was 0.5354,and the root mean square error for prediction（RMSEP）was 0.0373.It is concluded that Vis-NIRS can be used in the cocoon shell weight measurement,and UVE-SPA is a feasible and efficient algorithm for the spectral variable selection.