Short Term Load Forecasting Using Computational Intelligence Methods

Abstract

Load forecasting is very essential to the operation of electricity companies. It enhances the energy-efficient and reliable operation of a power system. This dissertation focuses on study of short term load forecasting using different types of computational intelligence methods. It
uses evolutionary algorithms (i.e. Genetic Algorithm, Particle Swarm Optimization, Artificial Immune System), neural networks (i.e. MLPNN, RBFNN, FLANN, ADALIN, MFLNN, WNN, Recurrent NN, Wilcoxon NN), and fuzzy systems (i.e. ANFIS). The developed methods give load forecasts of one hour upto 24 hours in advance. The algorithms and networks were have been demonstrated using simulation studies. The power sector in Orissa has undergone various structural and organizational changes in recent past. The main focus of all the changes initiated is to make the power system more efficient, economically viable and better service oriented. All these can happen if, among other vital factors, there is a good and accurate system in place for forecasting the load that would be in demand by electricity customers. Such forecasts will be highly useful in proper system planning & operations. The techniques proposed in this thesis have been simulated using data obtained from State Load Dispatch Centre, Orissa for the duration September – 2006 to August – 2007.