An Integrated Model Combining CNN-GRU with ARIMA for pressure prediction of water supply network

In the problem of time series prediction, the accuracy has always been an enduring research hotspot. In this paper, we proposed an integrated model named CNN-GRU-ARIMA to improve the accuracy of pressure prediction for water supply network. The proposed model is mainly comprised of the two parts, namely ARIMA and CNN-GRU. The ARIMA is usually used for the univariate prediction, and the CNN-GRU represents the multivariate prediction model. Also, several metrics including MAE, RRSE, RMSE are exploited to facilitate quantitative analysis of the accuracy differences between different models mentioned. The experiments show that our proposed scheme adopted in integrated model can achieve the better performance than some state-of-the-art approaches.