Comparative Evaluation of Artificial Neural Network Coupled Genetic Algorithm and Response Surface Methodology for Modeling and Optimization of Citric Acid Production by Aspergillus Niger MCBN297

Abstract

A comparative modeling and optimization of citric acid production from Aspergillus niger MCBN297 using Response Surface Methodology (RSM) and Artificial Neural Network (ANN) coupling Genetic Algorithm (GA) was carried out on seven process parameters.
A polynomial model was developed and RSM optimum process setpoints were determined. A multilayer ANN was structured, trained on experimental data, and served as fitness function for GA optimization.
Two ANN optimized media for citric acid production with predicted values of 4.69 g/L each, gave experimental productions of 6.65 and 6.68 g/L respectively, values higher than expected. Similarly, two RSM optimized media with predicted production of 7.19 and 7.04 g/L respectively, gave experimental values of 2.40 and 3.53 g/L respectively, exceedingly below RSM expectation. However, RSM provided good insight on parameters interactions. Both models can be developed using the same data pool.