Research on Short-term Water Demand Forecasting of Cities Based on Machine Learning
The rational utilization of urban water resources is an important foundation to support the rapid development of cities and ensure the livelihood of residents,and effective forecasting of urban short-term water demand is the premise of rational utilization of urban water resources.According to the actual water supply situation of the water supply area of a city,the daily water demand and the water demand at the interval of 30 minutes were predicted respectively.Firstly,the box plot method was used to identify the outliers of the historical data and back-correct the data,and then the random forest feature importance method was used to analyze the features.The artificial neural network model and random forest model were used to predict the daily water demand,and the results showed that the prediction accuracy of the random forest model was high,and the average absolute percentage error of the prediction results was less than 3％.The Prophet model was used to predict the water demand at 30-minute intervals,and it could provide reference for real-time scheduling and operation.
water demand predictionrandom forestANNProphet model
万方数据
维普
