Precipitation concentration index management by adaptive neuro-fuzzy methodology
| Autor: | Dragoljub Šević, Milan Gocic, Dalibor Petković, Slavisa Trajkovic, Miloš Milovančević |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Atmospheric Science
Global and Planetary Change Adaptive neuro fuzzy inference system 010504 meteorology & atmospheric sciences Neuro-fuzzy Meteorology 0208 environmental biotechnology Feature selection 02 engineering and technology 01 natural sciences Regression 020801 environmental engineering Model predictive control Conventional PCI Environmental science Precipitation 0105 earth and related environmental sciences Extreme learning machine |
| Zdroj: | Climatic Change. 141:655-669 |
| ISSN: | 1573-1480 0165-0009 |
| Popis: | This paper reconsiders the precipitation concentration index (PCI) in Serbia using precipitation measurements such as the mean winter precipitation amount, annual total precipitation, mean summer precipitation amount, mean spring precipitation amount, mean autumn precipitation amount and the mean of precipitation for the vegetation period (April–September). Potentials for further improvement of PCI prediction lie in the improvement of current prediction strategies. One of the options is the introduction of model predictive control. To manage the PCI, it is good to select factors or parameters that are the most important for PCI estimation and prediction, i.e. to conduct variable selection procedure. In the present study, a regression based on the adaptive neuro-fuzzy inference system (ANFIS) is applied for selection of the most influential PCI inputs based on the precipitation measurements. The effectiveness of the proposed strategy is verified according to the simulation results. The results show that the mean autumn precipitation amount is the most influential for PCI prediction and estimation and could be used for the simplification of predictive methods to avoid multiple input variables. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink