Rosin Tackifying Resin PCA-clustering Analysis Method Based On ATR Infrared Spectrum-wavelet Transform Method

With the rapid development of computer technology,chemometric methods such as pattern recognition have come into the attention of researchers again,and have been widely used in various fields.How to combine the new computer technology with the traditional instrumental analysis technology and develop practical application is particularly important.Rosin tackifier has been widely used in various pressure sensitive adhesives,and it is another important component of pressure sensitive adhesives besides the elasticity of the main rubber.At present,there are no reports on clustering identification by combining pattern recognition method and ATR-IR method.In this paper,ATR infrared spectroscopy was used to analyze rosin tackifier.The sample was prepared by the simplified tablet pressing method,and the infrared spectrum of rosin tackifier was collected by ATR-FTIR.In this paper,the classification method of rosin tackifier by wavelet decomposition principal component cluster analysis is established,and the data preprocessing method,wavelet decomposition method,principal component analysis and cluster analysis parameters are confirmed respectively.The optimal data preprocessing method is normalization method,the optimal wavelet decomposition method is to use the D2 detail signal of Sym8 and coif5 wavelet decomposition as the spectral variable,and the clustering result is the best when the main component fraction is selected as six.The cluster analysis model of rosin tackifier was established.The accuracy of Sym8 wavelet decomposition Manhattan distance dispersion square sum method,Sym8 wavelet decomposition Euclidean distance class average method and coif5 wavelet decomposition Euclidean distance dispersion square sum method was the highest,reaching 86.96%.This method can not only distinguish the types of rosin tackifier,but also the tackifier from different manufacturers.Compared with the untreated data,the accuracy of clustering analysis is 60.87%.The accuracy of clustering analysis after the model processing is 86.96%.The method described in this paper can greatly improve the accuracy of clustering analysis of rosin tackifier resin.