The use of artificial neural network for modelling of phycoremediation of toxic elements As(III) and As(V) from wastewater using Botryococcus braunii

Abstract

In the present study, a thorough investigation has been done on the removal efficiency of both As(III) and As (V) from synthetic wastewater by phycoremediation of Botryococcus braunii algal biomass. Artificial neural networks (ANNs) are practised for predicting % phycoremediation efficiency of both As(III) and As(V) ions. The influence of several parameters for example initial pH, inoculum size, contact time and initial arsenic concentration (either As(III) or As(V)) was examined systematically. The maximum phycoremediation of As(III) and As(V) was found to be 85.22% and 88.15% at pH9.0, equilibrium time of 144h by using algal inoculum size of 10% (v/v) and initial arsenic concentration of 50mg/L. The data acquired from laboratory scale experimental set up was utilized for training a three-layer feed-forward back propagation (BP) with Levenberg–Marquardt (LM) training algorithm having 4:5:1 architecture. A comparison between the experimental data and model outputs provided a high correlation coefficient (R2all_ANN equal to 0.9998) and exhibited that the model was capable for predicting the phycoremediation of both As(III) and As(V) from wastewater. The network topology was optimized by changing number of neurons in hidden layers. ANNs are efficient to model and simulate highly non–liner multivariable relationships. Absolute error and Standard deviation (SD) with respect to experimental output were calculated for ANN model outputs. The comparison of phycoremediation efficiencies of both As(III) and As(V) between experimental results and ANN model outputs exhibited that ANN model can determine the behaviour of As(III) and As(V) elimination process under various circumstances.