Prediction of Water Demand in Pump Station Based on GM(1,1)-MA Model

In order to solve the problem of poor prediction performance in traditional models due to various factors such as linearity, nonlinearity, and time variation affecting the water demand of pump stations, a novel combinatorial model, GM(1,1)-MA is proposed based on the Grey Model (GM) and Moving Average (MA) methods. The water consumption of pump stations are quantitatively predicted and compared under the conditions of traditional prediction models GM (1,1), moving average model MA, and GM (1,1)-MA combination model, which based on variance–covariance method The prediction results demonstrate that the proposed combinatorial model provides more accurate predictions of the actual water demand compared to the individual GM(1,1) and MA models, with average absolute errors reduced by 25.0% and 7.6%, and average percentage relative errors reduced by 19.5% and 12.6%, respectively. The developed combinatorial model effectively enhances the prediction accuracy of pump station water demand, providing a more reliable basis for forecasting and thereby offering valuable insights for optimizing the operation of pump stations.