Controlling biochemical oxygen demand in the multi-soil-layering using neural network tool

The effect of a set of environmental parameters on the pollution indicators bod5,in multi-soil-layering (msl) has been studied using neural networks (nns) and multiple regression analysis (mra). the obtained results show that,nns may be used as a practical tool to control the wastewater treatment. the process of rural wastewater treatment by msl,which is an innovative system used for the first time in morocco,was studied by modelling the relationships between a set of environmental factors and bod5 based upon 163 sampling. the models of interaction between variables were tested by using both mra and nns methods. considering the relevant factors obtained from the mra,a correlation coefficient (r2) of 0.58 with a standard deviation (s) of 0.76,approve that,there is no significant linear connection between different environmental factors and bod5. the non-linearity of the relationship has led us to use a nns methods. the architecture of the best model obtained by nns is mlp 5-1-1 with a r2 of 0.82 and a root mean square (rms) of 0.80. the best models established were found to be nonlinear. obtained results suggest that nns may be used as a practical tool to control the wastewater treatment in the msl. © 2017,university of mohammed premier oujda morocco.