الفهرس 
Introduction
ObjectivesOnly 14 pages are availabe for public view
 |  | from | 192 |  |  |
| | | from | 192 | | |
Abstract
Good forecasting leads to good decision. Therefore one of the major problem facing the electric utility is the unknown future demand of electricity which needs to be forecasted correctly. The electric load forecasting is a challenging problem that requires extensive statistical analysis. The problem formulation as well as modeling may depend on to a great extent on the geographical region where the forecasting is needed. The accuracy of the forecasting models depends on the functional relationship between the weather variables and electric loads that already known. In this thesis statistical and intelligence methods are used to forecast hourly load and daily load for Sana<U+2018>a energy system (YEMEN). A among them multiple linear regression, time series decomposition, BoxJenkins forecasting models and exponential smoothing methods have been used. The intelligence methods as artificial neural networks (feed forward, recurrent and radial basis ) and Adaptive Neuro Fuzzy Inference System (ANFIS). Test the results for all methods indicate that ANFIS is the most suitable for hourly load forecasting and the radial basis neural networks is the most suitable for daily load forecasting for Sana<U+2018>a energy system (YEMEN).