Quality Evaluation Method Of Cantonese-style Baked Goods Based On Electronic Nose

Cantonese-style baked food is a kind of pastry production in the Pearl River Valley and southern coastal areas of China. It is widely known at home and abroad. As the region's climate and environment, Cantonese-style baked food easy to oxidative rancidity, the old means of quality detection, and the fake and shoddy products, the quality and safety of Cantonese-style baked food become especially important. Based on the detailed analysis of the development trend for the identify, classification and quality control research of Cantonese-style baked food at home and abroad, this paper is aimed to use an electronic nose to identify and assess the quality of Cantonese-style baked food using the comprehensive information of the Cantonese-style baked goods volatile components through appropriate data processing method.Research method:firstly, select Cantonese-style almond cake as the research object, and use electronic nose PEN3 to get different types and storage periods of the odor information; Then get the original data set by the baseline processing and normalization transform of sensor signal; Choose all the representative characteristics of odor information to constitute a complete feature set; then obtain the most effective feature subset by optimizing the feedback of the extract features. Using principal component analysis (PCA) on the feature subset for further reduced dimensionality; Followed by Fisher linear discriminant analysis (LDA), Mahalanobis distance discriminant analysis and the amount of items driving the three-layer adaptive learning rate of BP neural network identification results obtained; Finally, these graphics, index, conclusions And the physical and chemical indicators (peroxide value, acid value) were compared for differences and contact analysis.According to the research target, choosing the 6 different types of almond cake,12 batches of different storage periods almond cake with meat and 12 different batches of different storage periods Li-li almond cake for the study. The conclusions are as follows:(1) Classification and identification almond cake with different types Take the following two methods:PCA+LDA method and PCA+the amount of items driving the three-layer adaptive learning rate of BP neural network. There is a very poor classification through response curves and polar plots; and PCA method based on feature subset classification performance has been greatly improved, but classification surface still rely on the close side; PCA+LDA classification maps can be concentrated within the class, and between-class spread clearly; By contrast, PCA+BP neural network method (accuracy rate 100%) is superior to PCA+LDA method (accuracy rate 96%).(2) Judge the quality of almond cake with different storage periodsFirstly, To obtain the physical and chemical indicators as an expert data of almond cake with meat and Li-li almond cake with different storage periods. Data analysis methods using PCA+LDA, and physical and chemical indicators for comparison. Using response curves and polar plots can get the trend of almond cake with different storage periods. Characteristic spectrum by the mean and variance are quite satisfactory and in accordance with the corresponding values of the second sensor is given to verify whether or not the two metamorphic indicators; And through the PCA map under feature subset can get the trends of transformation with different storage periods clearly, but the spacing of categories is closely. Through the PCA+LDA classification map under feature subset can distinguish expired and expiring almond cake. Finally, using mahalanobis distance discriminant analysis, and the data is divided into two categories of expired and expiring on the different storage periods, almond cake with meat and Li-li almond cake to get the right rate of 96% and 97% classification rate.The results show that it's possible to classify and identify the quality of Cantonese-style baked goods accurately by a portable electronic nose PEN3.