Abstract:

The purpose of this study was to discriminate among raw milk, pasteurized milk, aged milk, rancid milk and
adulterated milk based on sensor arrays constructed with inert metal electrodes and to investigate identification ability of the
sensor array for different milk samples. The data obtained from the determination of samples by differential pulse voltammetry
(DPV) were analyzed by one-way analysis of variance (ANOVA) and principal component analysis (PCA). Six sensors were
primarily screened and assigned to different groups by response performance with respect to the ANOVA mean square values
as well as P or F values, and the discrimination efficiencies of different combinations of the sensors were tested by PCA. As
a result, optimal sensor combination was establsihed. The results showed that Euclidean distance above 2 between every two
samples and good separation of all the samples in the PCA plots. Detection by DPV with a sensor array consisting of palladium,
platinum and gold electrodes and horizonal summation of acquired data allowed effective disrimination among the six milk
samples by Euclidean distance and PCA.

Key words: raw milk, differential pulse voltammetry, principal components analysis, euclidean distance, sensor array