三种非线性多元统计分析方法在智舌应用中的比较研究.docVIP

三种非线性多元统计分析方法在智舌应用中的比较研究 A Comparison Study Of Three Nonlinear Multivariate Data Analysis Methods For Smartongue Kernel PCA LLE and 5 10 15 20 25 30 35 40 Sammon mapping Tian Shiyi Chen Min Deng Shaoping School of Food Science and Biotechnology SFSB Zhejiang Gongshang University Abstract Smartongue was a voltammetric electronic tongue developed by the laboratory which was based on multifrequency large amplitude pulse voltammetry MLAPV and a non-specific sensors array of six naked metallic electrodes Three non-linear multivariate data analysis methods such as Kernel principal component analysis Kernel PCA Locally linear Embedding LLE and Sammon mapping were studied about the data processing ability for Smartongue Normal principal component analysis PCA was used as a reference method and discrimination index DI value was used as a quantitative indicator for evaluation the separation ability of different plot The result indicated that non-linear data processing methods had a more powerful ability than PCA for processing the data of Smartongue Among these techniques Sammon mapping successfully classified three bitter solutions six artificial green tea products and five milk power solutions with different storage time well and exhibited the best data processing ability for extracting the useful information from the data of Smartongue Key words Food science Nonlinear multivariate data analysis Smartongue 0 Introduction Electronic tongue[1] was a kind of modern qualitative and quantitative analytical instrument that consists of cross-sensitive sensors array and appropriate pattern recognition From the 1980s where the first electronic tongue was invented many researchers devoted themselves into the research of electronic tongue and several types of electronic tongue were developed For example Toko research group[2] of Kyushu University in Japan and Legin research group[3] of Saint Petersburg University in Russia developed a kind of po