Enhancing BOD5 forecasting accuracy with the ANN-Enhanced Runge Kutta model

This study enhances the prediction of biochemical oxygen demand (BOD5), a vital water quality parameter, by developing hybrid artificial neural network models integrated with advanced optimization algorithms. Data from two monitoring stations in South Korea were used to create five models, including the innovative ANN-Enhanced Runge Kutta (ANN-ERUN) model. ANN-ERUN achieved the highest accuracy, significantly outperforming other models. At Gong station, it reduced prediction error (root mean square error: 1.24 mg/L; mean absolute error: 0.83 mg/L) and achieved a determination coefficient of 0.857. Models using eight water quality parameters, including dissolved oxygen and chemical oxygen demand, exhibited superior performance. These findings confirm the effectiveness of ANN-ERUN in precise BOD5 prediction, offering a robust tool for environmental monitoring and sustainable water quality management.