Long and short term demand forecast, a real application
| Autor: | Grau Torrent, Sergi|||0000-0003-1192-1430, Luis, Anna, Saludes Closa, Jordi|||0000-0002-6666-1982, Pérez Magrané, Ramon|||0000-0002-9216-4234 |
|---|---|
| Přispěvatelé: | Universitat Politècnica de Catalunya. Departament de Matemàtiques, Universitat Politècnica de Catalunya. Departament d'Enginyeria de Sistemes, Automàtica i Informàtica Industrial |
| Jazyk: | angličtina |
| Rok vydání: | 2019 |
| Předmět: | |
| Zdroj: | UPCommons. Portal del coneixement obert de la UPC Universitat Politècnica de Catalunya (UPC) |
| Popis: | Two of the main purposes of a water supply company are the operation of the network and its planning. One of the critical elements for the planning and the operation is the demand forecast. There are multiple methods for the short term. These tools are based on the analysis of historical data (daily, hourly or higher frequency) as an indicator of future flows due to a repetitive and cyclic behaviour of the consumers. The Autoregressive Integrated Moving Average (ARIMA) is one of the most straightforward approaches producing good results. Nevertheless, the demand forecasting is continuously evolving and new models are suggested like the fully adaptive forecasting model or those based on the chaos theory. The ARIMA model for short term, predicts one day demand using 22 features grouped in three types. Water demand of the previous 48 hours. To capture fast changes and weather influence. Water demand of the previous 10 same week days. To capture type day influence and seasonality. Normalized water demand of the previous 10 same week days. To avoid the false seasonality influence. A second short-term water demand forecasting model is used. It is a heuristic model that automatically stores and updates water demand patterns and demand factors for all days of the week and for a configurable number of deviating days like national holidays, vacation periods, and individual deviating days. The model uses this information to adaptively learn the patterns and factors that are used when forecasting the water demand. The two demand forecast algorithms are used and compared for the short and long term prediction. Their results are compared with those of the literature. The results suggest that both methods perform similarly in short term but the ARIMA is more easily generalizable for long term predictions. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|