Modeling and Prediction in the Enzymatic Hydrolysis of Cellulose Using Artificial Neural Networks

Artificial intelligence technique namely artificial neural network (ANN) was used to describe the enzymatic kinetics of cellulose hydrolysis in a heterogeneous system, and compared with response surface methodology (RSM). Three hydrolysis conditions (activity of added cellulase, substrate concentration and time) served as the input of the neural network model, and the glucose content served as the output. The experimental data from Box-Behnken design were used to train the neural network using the back propagation algorithm. The others of 33 design were used to check the performance of the trained network. The ANN modelled and predicted values showed better agreement with the experimentally reported ones than RSM. ANN could mimic the heterogeneous enzymatic hydrolysis of cellulose.