Tailoring a Hybrid Intelligent Model to Predict Fermentable Sugar Production from Enzyme−catalyzed Hydrolysis of Damaged Wheat Grains

Abstract

Damaged wheat grains (DWGs) are relinquished into the environment and functions as potential powerhouse for pathogenic microorganisms. These substrates are rich in starch that could be used for attractive applications after conversion to useable form. In this work, enzymatic hydrolysis of DWG flour (10–20 % w/v) was done using α-amylase (1–5 % v/v) over 30−120 min to convert DWG starch to fermentable sugars. SEM micrographs and FTIR analysis revealed the breakdown of the spherical and ellipsoidal surface of the starch granules and formation of large pores at α-amylase attachment sites. Genetic algorithm (GA) mediated artificial neural network (ANN) model was applied to predict the sugars production in the enzyme−catalyzed hydrolysis process. ANN-GA model with 12 hidden layer neurons and Levenberg Maquardt training algorithm was defined to predict enzymatic hydrolysis data with 93.2 % accuracy. Random residuals generated by the model confirmed that the influence of extraneous variables was limited thereby, proving the model efficient.