Keywords: back propagation modelling, river water, water quality, simulation, predicting, error analysis, testing, dissolved oxygen, BOD, biochemical oxygen demand, ANNs, artificial neural networks, forecasting
Error analysis for river water quality prediction using back propagation modelling
The purpose of this research is to describe the errors of Back Propagation (BP) technique that could be applied in river water quality prediction. ANN was trained by using the upriver data to forecast the downriver quality in two approaches of long-distance forecasting and short-distance forecasting respectively. It showed that the predicted results from short-distance approach were more accurate than those of long-distance forecasting. It is indicated that the proper numbers of river segments could be chosen according to the predicted errors in the practical river water quality forecasting.