Identification Of Almond Cake Deterioration Based On Computer Olfactory

With the sustainable development of national economy, more and more abundant food supply, and more abundant species, a variety of food safety problems have gradually exposed in life, along with the food supply balance at the same time, food quality and safety issues become more and more important. Almond cake is a traditional bakery in Guangdong. It is popular in the Pearl River Basin and southern coastal areas. The main ingredient of almond cake Material is oil,sugar and pork. etc. When preservation is not the time, almond cake is susceptible to microbes, and oxidative rancidity, because of the local hot and humid climate.Therefore, in this paper, mung with meat almond cakes ("almond cake" for short) were used for the research object. The same batch of almond cakes with12different storage periods was selected to do tracking experiment. And their physicochemical index values were used for an expert knowledge. PEN3E-nose was employed to get E-nose signals values of almond cakes. The study preliminarily analyzed the relationship between PENS E-nose signals values and physicochemical index values of almond cake by correlation analysis.(1) Due to lack of integrity, SCA just study the relationship between two variables, which may not fully reflect the intrinsic link between some variables as a whole and some others as a whole; While study the relationship between two sets of variables by CCA. it is more suitable for the study of correlation between response signals of E-nose and physicochemical indicators of almond cake.(2) In this paper canonical correlation coefficients show that the correlation coefficients of the first pair and the second pair of canonical variables are0.96245and0.586784respectively, which account for almost100%of information; From the canonical model and canonical structure analysis,W5S sensor response signal (X2) and W3S sensor response signal (X10) have close relations with the physicochemical indicators, and the first canonical variable V] of response signals set has a better predictive capability of acid value (Y1)and peroxide value (Y2) of almond cake, of which correlation coefficients are0.9558.0.7004respectively; Canonical redundancy analysis show the two canonical variables of V and W can explain19.92%and78.54%of the cumulative percentage of the total variances of another group respectively.(3) The results indicated that the PEN3E-nose response signals of W5S sensor and W2W sensor were highly correlated with physicochemical index. On the basis of CCA. the study analyzed odor fingerprint about mean and variance of E-nose response signal. The results showed that W5S sensor eigenvalue increased with the increase of the physicochemical index, and the study determined almond cake deterioration threshold with eigenvalue of the mean and variance of sensor W5S response signal for1.75and0.13respectively, and with response value of W5S signal for3.1. Finally, maximum scatter difference(MSD) linear discriminant analysis and PCA-LDA were used to classify E-nose response signals of almond cake respectively, further the study identified unknown samples using Mahalanobis Distance and Euclid Distance. The results indicated that LDA based on maximum scatter difference and PCA-LDA could correctly identify unknown samples consistent with physicochemical index. accuracy reached100%.