期刊
JOURNAL OF FOOD SCIENCE AND TECHNOLOGY-MYSORE
卷 48, 期 2, 页码 204-210出版社
SPRINGER INDIA
DOI: 10.1007/s13197-010-0158-2
关键词
Kinetics; Connectionist; ANN models; Basundi; Dairy products; Shelf life prediction
A ready-to-reconstitute formulation of Basundi, a popular Indian dairy dessert was subjected to storage at various temperatures (10, 25 and 40 degrees C) and deteriorative changes in the Basundi mix were monitored using quality indices like pH, hydroxyl methyl furfural (HMF), bulk density (BD) and insolubility index (II). The multiple regression equations and the Arrhenius functions that describe the parameters' dependence on temperature for the four physico-chemical parameters were integrated to develop mathematical models for predicting sensory quality of Basundi mix. Connectionist model using multilayer feed forward neural network with back propagation algorithm was also developed for predicting the storage life of the product employing artificial neural network (ANN) tool box of MATLAB software. The quality indices served as the input parameters whereas the output parameters were the sensorily evaluated flavour and total sensory score. A total of 140 observations were used and the prediction performance was judged on the basis of per cent root mean square error. The results obtained from the two approaches were compared. Relatively lower magnitudes of percent root mean square error for both the sensory parameters indicated that the connectionist models were better fitted than kinetic models for predicting storage life.
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