Parallel neural network-fuzzy expert system strategy for short-term load forecasting: System implementation and performance evaluation

The on-line implementation and results from a hybrid short-term electrical load forecaster that is being evaluated by a power utility are documented i n this paper. This forecaster employs a new approach involving a parallel n eural-fuzzy expert system, whereby Kohonen's self-organizing feature map wi th unsupervised learning, is used to classify daily load patterns. Post-pro cessing of the neural network outputs is performed with a fuzzy expert syst em which successfully corrects the load deviations caused by the effects of weather and holiday activity. Being highly automated, little human interfe rence is required during the process of load forecasting. A comparison made between this model and a regression-based model currently being used in th e Control Centre has shown a marked improvement in load forecasting results .