Model coupling can improve the prediction accuracy of urban water demand: A case study of ARIMA-LSTM

Authors

  • Yuncheng Dong

Abstract

Predictive modeling of water supply volumes stands as a pivotal tool in informing the strategic scheduling of urban water supply systems. It not only furnishes essential data for the system’s operational optimization but also facilitates water utility companies in maintaining a delicate equilibrium between supply and consumer demand, thereby advancing the dual objectives of energy conservation and augmenting the efficiency of water resource utilization. To further refine the precision of such forecasts and thereby enhance the overall management of urban water resources, this study embarks on an investigation centered on the development of a hybrid predictive model. Specifically, it leverages real-world datasets obtained from two water treatment facilities of disparate sizes within a selected metropolitan area as the core of its analytical framework.

The research adopts a daily temporal resolution, harnessing the combined strengths of Autoregressive Integrated Moving Average (ARIMA) models and Long Short-Term Memory (LSTM) networks to construct a sophisticated forecasting model tailored to urban water demand. By integrating ARIMA’s prowess in capturing temporal patterns and trends with LSTM’s capacity for learning intricate, long-term sequential dependencies, this study aims to push the boundaries of forecasting accuracy. Consequently, this endeavor contributes to the enhancement of decision-making processes within water utilities, ensuring a sustainable balance between water provision and demand, and ultimately fostering the judicious use of this invaluable natural resource.

Downloads

Published

2024-06-14

How to Cite

Yuncheng Dong. (2024). Model coupling can improve the prediction accuracy of urban water demand: A case study of ARIMA-LSTM. Onomázein, (64 (2024): June), 78–84. Retrieved from http://www.onomazein.com/index.php/onom/article/view/684

Issue

Section

Articles