腐乳感官和理化品质的核主成分分析

为解决PCA不适合多指标综合分析中非线性主成分分析的问题，采用核主成分分析(KPCA)方法，对我国不同地区16种腐乳的品质进行了综合评价。使用核函数将原空间映射到高维特征空间。在高维空间进行了线性主成分分析。结果表明。通过对核参数的适当选取。可使最大特征值的贡献率达到或接近85％。避免了多个主成分的不同组合而导致的评价结果的不一致。应用KPCA得到的腐乳品质的评价结果具有一定的客观性。

Kernel principal composition analysis of sensory and physiochemical quality of fermented bean curd

Through kernel principal composition analysis, a new method of comprehensive evaluation was made for the sensory and physiochemical quality of sixteen kinds of fermented bean curd from markets. By using the kernel functions, one can efficiently calculate principal compositions in high dimensional feature spaces, related in input space by some nonlinear map. The results showed that the maximum eigenvalue contributes nearly 85 % by choosing appropriate parameters, avoiding the different array as a result of many principal composition. KPCA should have a good performing in comprehensive analysis of the quality of fermented bean curd.