Associated clustering and classification method for electric power load forecasting

Abstract

In the process of power load forecasting, electricity experts always divide the forecasting situation into several categories, and the same category uses the same forecasting model. There exists such a situation that some load curve which domain experts consider belonging to the same category has shown the different characteristics, but some load curve which belongs to different category seems very similar, and usually able to gather into a category by clustering. For this problem, the definition of associated matrix was proposed in this paper, and based on this conception the associated clustering- classification algorithm was proposed, We applied this algorithm to data sample classification for power load prediction, the experiment showed that the classification results obtained by our method were more reliable. © 2010 IFIP.