Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
494994 | Applied Soft Computing | 2015 | 10 Pages |
•This paper proposes a new network named dHS-ARTMAP based on ARTMAP.•The method reduces proliferation problems, retains stable memories, and enjoys noise tolerance.•dHS-ARTMAP model achieves excellent results compared with other models.
Intelligence technology develops quickly to predict and respond to the actions of electric power users to maintain a reliable and secure electricity infrastructure. This paper proposed a new on-line training network called distributed hyper-spherical ARTMAP (dHS-ARTMAP) to forecast the electricity load. The new model constructs a more compact network structure and largely decreases the proliferation problem that Fuzzy ARTMAP models usually encounter. Experiments of short-term electricity load forecasting are made with the data from Queensland, Australia. Results are compared with other methods. The effectiveness of the dHS-ARTMAP network proves itself a promising alternative to put into practical use.
Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slide