Water quality prediction in the Yellow River source area based on the DeepTCN-GRU model

Water quality prediction is critical in water resource management. Accurate water quality prediction can detect potential water quality issues ahead of time and provides an important scientific foundation for achieving sustainable water resource management. To predict the acid-base index (pH) and total nitrogen content (TN) in water quality indicators, this study uses data from the Kaifeng Yellow River water source area to propose a deep learning combination model (DeepTCN-GRU) that combines the benefits of convolutional neural networks (CNN) and recurrent neural networks (RNN). The study analyzed the effects of data processing, different lag values, and different prediction durations on the predictive performance of the model, as well as compared the predictive ability of different deep learning models and explored their predictive performance on water quality data from different water sources. The research results found that data processing can significantly reduce noise in the data and improve the predictive ability of the model; The DeepTCN-GRU model has the best prediction performance for water quality indicators pH and TN when the lag value is 30 days and the prediction duration is 1 day; Compared to other deep learning models, the DeepTCN-GRU model reduces RMSE, MAE, and MSE metrics by at least 29.69 %, 40.21 %, and 36.03 %, R2 There has been a minimum 6.63 % gain in value; In the prediction of water quality data from different water sources using the DeepTCN-GRU model, R2 values are all above 0.9. Overall, the DeepTCN-GRU model provides significant support for Yellow River water quality monitoring and management.