طراحي روشي نوين براي پيش بيني تقاضاي كوتاه مدت گاز طبيعي در بخش خانگي

Designing a New Model to Forecast Natural Gas Short Run Demand of the Residential Sector

In this paper we design a new model that integrates linear and non linear models to forecast the trend natural gas daily demand in the residential sector of the country and isolate the factors that impact daily demand. We use the model to forecast the natural gas daily demand of the residential sector for the next 10 days in a "step by step" manner, using the ARIMA, Feed forward Artificial Neural Network, and Neural Network-Wavelet Transform models and the new proposed model. The results indicate that the proposed model has fewer forecasting errors and higher accuracy in forecasting the natural gas daily demand of the residential sector of the country. The Neural Network-Wavelet Transform, Feed forward Artificial Neural Network and ARIMA models rank lower than the proposed model in the given order. JEL Classification: C53, E37, Q30