基于近红外光谱的羊绒羊毛鉴别降维方法比较OA

There are differences in fiber characteristics,warmth retention,softness,and appear-ance among different varieties of cashmere and wool,which directly affect the quality and market value of the product.Therefore,rapid and accurate identification of cashmere and wool varieties is of great significance for improving product quality and enhancing market competitiveness.To o-vercome the limitations of traditional detection methods in terms of efficiency and cost,as well as the challenges of high dimensionality,non-linearity,and spectral overlap of near infrared spectros-copy data,a fast and non-destructive identification method for cashmere and wool varieties based on near infrared spectroscopy technology was proposed.The experimental samples cover 14 varie-ties and a total of 420 cashmere and wool samples.Firstly,the high-dimensional spectral data was subjected to dimensionality reduction using linear discriminant analysis(LDA).Then,the param-eters of the support vector machine(SVM)model were optimized using grey wolf optimizer(GWO)algorithm,resulting in the construction of an LDA-GWO-SVM joint discriminant model.Meanwhile,the performance of various dimensionality reduction algorithms such as principal com-ponent analysis(PCA)for linear methods and locally linear embedding(LLE)and isometric fea-ture mapping(ISOMAP)for nonlinear methods were compared.The results show that linear methods perform well in dimensionality reduction,and the LDA-GWO-SVM model achieves a classification accuracy of 99.21%under 11 principal factors.The combination of near infrared spectroscopy technology and machine learning provides a new method for identifying cashmere and wool varieties,providing effective support for rapid and accurate variety identification and quality control in practical applications.