An artificial neural network approach to estimating the enzymatic hydrolysis of Chinese yam (Dioscorea opposita Thunb.) starch. Issue 5 (9th January 2017)

• Record Type: Journal Article
• Authors: Ding, Baomiao; Li, Li; Yang, Hualin
• Is Part Of: Journal of food processing and preservation. Volume 41:Issue 5(2017)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2017-01-09
• Subjects: starch hydrolysate -- Chinese yam (Dioscorea opposita Thunb.) -- dextrose equivalent value -- artificial neural network -- sensitivity analysis; Food -- Preservation -- Periodicals; Food industry and trade -- Periodicals; 664.005
• DOI: 10.1111/jfpp.13176
• Languages: English
• ISSNs: 0145-8892

Abstract

Abstract: Reasonable hydrolysis of starches could be used to partly replace fat in food. The Chinese yam starch hydrolysates were investigated based on DE value and inverted microscope. Photomicrographs clearly showed that the sizes of starch hydrolysate granules had been reduced effectively. A three‐layer back‐propagation neural network (BPNN) network was designed based on the results of orthogonal design experiment with a L18 (3 5 ) array, and the BPNN network architecture contained 5, 4, 1 neuron in the input layer, the hidden layer, and the output layer, respectively. The hydrolysis conditions and DE values were considered as input and output variables, respectively, and they were utilized to train the BPNN network. The network configuration could be used to predict the DE values of starch hydrolysates with correlation coefficient of 0.9550. Moreover, variable importance analysis indicated that the descending sequence of the five variables was: enzyme concentration, reaction temperature, pH, starch concentration, and reaction time. Practical applications: Excessive intake of fats would cause some health problems. It is very necessary to reduce the amount of fat in consumers' diet and to produce the low‐fat food. Starch hydrolysates are considered to be a kind of potential fat replacer. It can provide fewer calories. However, starch hydrolysis is highly nonlinear because of the complexity and variability of the processing conditions. Traditional experiment data analysis methods, such as the "pick the winner" approach, are impotent to find optima. In the current study, Chinese yam starch hydrolysates were obtained by enzymatic hydrolysis. The relationship between hydrolysis variables and DE values of starch hydrolysates was described by artificial neural network model. The model could be used to guide the production of the starch‐based fat replacer. Furthermore, Chinese yam could be adequately developed as a good source of starch in East Asia, and its added value was also enhanced.