Gas load forecasting model input factor identification using a genetic algorithm

Abstract

Genetic algorithms (GAs) are used as a tool to identify the input factors for an hourly gas load forecasting model. The proposed model can provide up to 106 hours of load forecasts. Experiences obtained during the application of GA for determination of inputs are discussed. Linear regression based models using the results of this study had an average error 23% less than the existing method at one gas utility over six service areas.