Investigation of the Environmental Quality of Watershed Prediction System Based on an Artificial Intelligence Algorithm

Abstract

Monitoring and predicting the environmental quality of watersheds is an important method for studying water pollution. However, current prediction models have several disadvantages, including excessive information, complex models, and extensive computation. This paper proposes a water pollution prediction system using Artificial Neural Network (ANN) trained by the Back Propagation (BP) algorithm with a 2-6-2 structure. Chemical oxygen demand (COD) and NH4+ data from the catchment areas of Kaihua and Anji counties in Zhejiang Province were used as the basis for the training of the BP model, using water quality indicator concentration data from November 2020 to October 2021. The effectiveness of training the BP neural network is confirmed by the goodness of fit of -4.8% and -4.48% for COD and NH4+, respectively. The mean relative errors between the predicted and observed values of the neural network for COD and NH4+ were -2.08% and 2.13%, respectively, indicating that the results of predicting water quality by the BP neural network were basically consistent with reality.

Keywords: Water quality prediction, NH4+, COD, BP neural network