Comparative investigation of RSM and ANN for multi-response modeling and optimization studies of derived chitosan from Archachatina marginata shell

The design of this paper was to investigate comparatively the optimization techniques of response surface methodology (RSM) and artificial neural network (ANN) when applied to the conditions for chitosan production from Archachatina marginata shell and the % removal of methylene blue, MB, from synthetic textile wastewater. The proposed RSM and ANN models are optimized using genetic algorithm (GA). The optimum conditions for the extraction processes of chitosan and % removal of MB are determined and the derived chitosan at optimized conditions is characterized using analytical techniques. The ANN portrays better modeling abilities than RSM for the responses. The predicted values of % yield of chitosan, % DD and % removal of MB are obtained as 51.56, 98.68 and 94.71 respectively using the RSM-GA technique while the ANN-GA technique predicted 45.32%, 91.96% and 95.96% respectively. The experimental values of the responses are in excellent agreement with the ANN-GA predicted values with % errors being 1.8, 1.2 and 1.19 respectively. Hence, the conditions of chitosan production from Archachatina marginata shell and its bioremediation capacity of synthetic wastewater from textile industry can be adequately and accurately optimized and modeled using ANN-GA for routine seafood applications and treatment of industrial wastewater effluents.