期刊
JOURNAL OF FOOD PROCESS ENGINEERING
卷 37, 期 5, 页码 474-481出版社
WILEY
DOI: 10.1111/jfpe.12101
关键词
-
Two intelligent tools, fuzzy expert engine (FEE) and artificial neural network (ANN), were used to predict momentarily the moisture content of zucchini slices during dehydration process. For this purpose, the fuzzy logic toolbox in MATLAB was used to predict moisture ratio (MR) of dried zucchini. Drying time, temperature and product thickness were selected as input variables and MR was selected as output variable. Linguistic variables were defined for both input and output variables. Generally, 81 if-then rules with AND operator was obtained for thin layer drying of zucchini. It was found that ANN with one and two hidden layers, topology of 3-20-20-1 had the best result to predict MR. This network was able to predict MR with R-2=0.998. Finally, comparison of these results revealed that ANN model had greater accuracy than FEE to predict MR of dried zucchini. Practical ApplicationsDifferent intelligent tools including fuzzy and neural networks were used in this study to control the drying parameters and produce dehydrated zucchini with minimal nutritional changes. Zucchini is a rich source of different vitamins, minerals, dietary fibers and omega 3 fatty acids. The suggested nondestructive approach is capable to predict kinetic parameters during the zucchini drying more accurately than the conventional regression method. In other words, it is possible to check the moisture ratio in every minute of drying process and determine the drying time for reaching to specific moisture content in the final product. Since the moisture ratio in the regression model is a function of drying time (with the assumption of different constant coefficients), it is not able to control various complex and nonlinear relations between the independent and dependent variables during drying process. This is the reason that zucchini drying monitored with the suggestive system had a higher quality and most probably used less energy than the similar product controlled with regression model.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据