Statistical Optimization and Neural Modeling of Amylase Production from Banana Peel Using Bacillus subtilis MTCC 441

Abstract

Optimization of process parameters is of critical task in developing an industrial fermentation process for various reasons. Many techniques are available for experimental design and optimization of fermentation process parameters, i.e., fermentation medium composition and conditions, each having its own advantages and disadvantages. In this work response surface methodology (RSM) with central composite design (CCD) of experiments and artificial neural network (ANN) coupled with global optimization technique (TOMLAB’s Direct Alogoritm—gblSolve) are used for optimization of process parameters for the production of alpha amylase using banana peel as the substrate and bacterial source Bacillus subtilis MTCC 441. A 46 run central composite design was used to plan the experiment with six parameters (banana peel concentration, peptone concentration, pH, temperature, incubation time and inoculum size) with five levels. The maximum amylase activity predicted by CCD and ANN is in good agreement with the experimental values at the optimized levels. The present work shows the better optimization and prediction capacity of ANN techniques compared to RSM technique.