An ensemble learning model for water quality forecast was constructed by using time series models including ARIMA and Prophet model as base learners, combined with BP neural network model. The monitoring data of five water quality indicators, including DO, CODMn, NH₃-N, TP and TN, for a section of the Yangtze River Basin from 2019 to 2020 were selected to test the validity of the model. The results show that the mean absolute percentage errors of the predicted results of the ensemble learning model for the five indicators were lower than those of the time series model by 35.0%, 29.9%, 4.1%, 40.6% and 17.1%, respectively. The prediction accuracy of the ensemble learning model is higher than that of the single model, which can be more accurate for water quality forecast.