Programme  Poster session 2  abstract 364

Using Neural Network Models to Forecast the Inflow of Doroudzan Dam – Iran

Author(s): Ali Torabi Haghighi, Pouyan Keshtkaran
main author:(Pouyan Keshtkaran) Islamic Azad University - Estahban Branch Estahban,Fars,Iran email: tel:+989173132092 author2:(Ali Torabi Haghighi) Fars Regional Water Authority, Shiraz, Iran email: atorabi_ha@

Keyword(s): Drudzan dam, neural network model, seasonal model, montly model, weekly model

Article: abs364_article.pdf
Poster: abs364_poster.pdf
Get Adobe Reader

Session: Poster session 2
AbstractOne of the methods to have optimum exploitation

of the water resources is forecasting the amount of available water in these resources. Doroudzan dam with 993

MCM reservoir capacity which is located in a dry area in south of Iran is a great water resource in region to use for

irrigation, industry and generation hydroelectric power. Saving water in the reservoir to have the best use of it is a

priority. In case of good forecast of inflow to the reservoir, discharging water through spillway could be minimized by

an appropriate schematization of outflows and amount of excess water could be conducted to generate electricity.

For this purpose 3 neural network models were created to forecast the amount of seasonal inflow, monthly inflow

and weekly inflow to the reservoir. In seasonal model the data of last year rainfall, cumulative rainfall until last season,

atmospheric forecast for specific duration and the amount of inflow in last season and last year were used. In monthly

model neural network model the seasonal forecast, last year inflow amount and the cumulative inflow amount until

last month were used. In weekly model, out puts of monthly model, the inflow in up stream branches, cumulative

rainfall in the year till last week, last week rainfall and the inflow rate in last week were used. The models was used

for simulation of the reservoir during 2000-2005 and during this years 500 MCM was saved in the reservoir and

used to generate more electricity.

  Return up