Simulation of BOD-DO Modeling in Mahanadi River System lying in Odisha using ANN, India

Abstract

Most of the water resources have been utilized for the disposal of municipal and industrial wastes. To address the non-linearity, subjectivity, transfer and transformation rule of the pollutants and complexity of the cause-effect relationships between water quality variables and water quality status, development and use of water quality model is of utmost importance. Owing to the random discharge of point pollution from various sources has not only rendered such water bodies eutrophic but also their advantageous uses such as water supply, irrigation, recharge of ground water, recreation and habitat for flora and fauna have been adversely affected. Oxygen- demanding substances are major contaminants in domestic and municipal wastewater. The main indicators of river pollution which deals with the oxygen domestic conditions of the river are Biochemical oxygen demand (BOD) and dissolved oxygen (DO). The Multi-layer Perceptron (MLP) neural network techniques was used to estimate for the analysis of point source pollution in terms of BOD and DO concentration and the neural network model is developed using the data collected from the upstream and downstream stations on Mahanadi river system lying in Odisha. The accuracy performance of training, validation and prediction of seasonal water quality parameters has been tested. To test the validity of ANN