A Spatial Load Forecasting Method Based on the Theory of Clustering Analysis

The spatial load forecasting method based on time series analysis usually divides predictive zone into cellules, and then forecast the load of each cell based on time series analysis, to predict the whole spatial load. Now there are many prediction models based on time series analysis, but it is hard to realize the minimum of the prediction error for each cell, namely not guarantee the accuracy of results of spatial load forecasting. Thus, the spatial load forecasting method based on the theory of clustering analysis is proposed there. Firstly this method analyzes the results and the relative prediction error of each cell with different load forecasting model. Second, according to the best forecast models, cluster all cellules, which cellules in the same cluster use the same prediction model to forecast the load of cellules of the target year. The results of a practical example show that our method is correct and effective.