Modeling hydrogen production using green algae Chlorella vulgaris utilizing Neural Networks

Abstract

The production of hydrogen via biophotolysis using algal strain Chlorella vulgaris within an anaerobic batch
reactor has been studied.This paper presents the development of a model used to predict the production of hydrogen as
function of time with Artificial Neural Network (ANN). The model reported is based on a multi-layer perceptron function
neural network (MLP-NN) with a configuration of 3-6-4-1 combined with sigmoid transfer functions tansig,tansig,
purline and trainlm respectively. The architecture of the model has been designed in order to mimic the interrelationship between three input parameters: substrate concentration, medium pH and the media contents of nitrogen
and phosphate. The ANN model was refined and tested with the use of 48 experiments. The correlation coefficient
between the experimental data and the model prediction was R2= 0.985 for training and testing. The results showed that
the ANN model successfully predicted the production of hydrogen from Chlorella vulgarisalgal strain and provided a
high level of accuracy for the training and testing stages with a maximum error of 6% and 2% respectively.