Modified imperialist competitive optimization to high resolution spatial electric load demand forecasting

This paper presents a methodology for a high-resolution urban spatial load demand forecasting. This methodology is meant to improve the visualization, analysis and inference of load density information in the electric distribution systems in the near future. The proposed methodology converts input data into grid maps and then divides the grid map into larger regions, which will have their expected growth according to convolution matrices and weighting factors that search for characteristics in the history of this region. The definition of the characteristics of the region’s growth is obtained by processing the imperialist competitive algorithm that searches the best array of convolution, which will set the expected growth of the region. Thus, it is possible to obtain a spatial growth forecast of high resolution and with great precision, which are important factors for smart-grid planning.