Short-term electric load forecasting in uncertain domain: a fuzzy decision tree approach

The objective of the research reported in this paper is the development of a model for short term load forecasting for use in an environment characterized by uncertainty. The fundamental requirement for the proposed model is the production of robust and accurate performance with minimal computational and data resources. Our solution strategy was developed around a computational intelligence method which exploits knowledge using fuzzy logic and decision tree based techniques. The model was developed and evaluated using three years data (i.e. 2004, 2005 and 2006) on electric loads obtained from the National Control Centre (NCC) Os.ogbo, Nigeria and was implemented using the Fuzzy Decision Tree software (FID 4.2). The data was supported by knowledge elicited from experienced power monitoring staff at NCC. The results showed that the average fractional forecast errors for the proposed model on selected data from the three years was 0.17 while that of the conventional multiple regression model was 0.80.