Original scientific paper

Comparative Study of Response Surface Methodology, Artificial Neural Network and Genetic Algorithms for Optimization of Soybean Hydration

Tushar Gulati ; Microbial Biotechnology and Downstream Processing Laboratory, Agricultural and Food Engineering Department, Indian Institute of Technology, IN-721302 Kharagpur, India
Mainak Chakrabarti ; Microbial Biotechnology and Downstream Processing Laboratory, Agricultural and Food Engineering Department, Indian Institute of Technology, IN-721302 Kharagpur, India
Anshu Sing ; Microbial Biotechnology and Downstream Processing Laboratory, Agricultural and Food Engineering Department, Indian Institute of Technology, IN-721302 Kharagpur, India
Anshu Sing ; Microbial Biotechnology and Downstream Processing Laboratory, Agricultural and Food Engineering Department, Indian Institute of Technology, IN-721302 Kharagpur, India
Muralidhar Duvuuri ; Microbial Biotechnology and Downstream Processing Laboratory, Agricultural and Food Engineering Department, Indian Institute of Technology, IN-721302 Kharagpur, India
Rintu Banerjee ; Microbial Biotechnology and Downstream Processing Laboratory, Agricultural and Food Engineering Department, Indian Institute of Technology, IN-721302 Kharagpur, India

Abstract

The present investigation deals with the modelling and optimization of soybean hydration for facilitating soybean processing and it focuses on maximization of mass gain, water uptake and protein retention in the bean. Process variables considered for optimization were: soybean to water ratio (1:2.48 obtained with response surface methodology, RSM, and 1.19 obtained with artificial neural network and genetic algorithm, ANN/GA), time (2.0 h using RSM and 8.0 h using ANN/GA) and temperature (40.0 °C using RSM and 45.1 °C using ANN/GA). The findings in this first report on optimization of soaking conditions for soybean hydration employing response surface methodology, hybrid artificial neural network and genetic algorithms reveal a substantially better alternative to the time-consuming soaking process, extensively practiced in industries, in terms of process time economy. Reasonably accurate neural network model (regression coefficient of 0.9443) was obtained based on the experimental data. The optimized set of process conditions was predicted through genetic algorithm, and the effectiveness of the ANN/GA model, validated through experiments, was indicated by significant correlations (R2 and mean squared error (MSE) being 0.9380 and 5.9299, respectively). RSM also resulted in accurate models for predicting percentage mass gain, percentage water uptake and percentage protein retention (R2 and MSE in the range of 0.889–0.9297 and 0.80–4.94, respectively).

Keywords

response surface methodology (RSM); artificial neural network (ANN); genetic algorithms (GA); soybean soaking

Hrčak ID:

48430

URI

https://hrcak.srce.hr/48430

Publication date:

5.3.2010.

Article data in other languages: croatian

Visits: 2.379 *