Hierarchical Forecasting of Load Demand With Smart Meter Data for Distribution Networks

Load forecasting is an essential task for the use of technologies such as energy storage systems and distributed energy resources in modern distribution networks. However, these technologies can increase the complexity of the operation of the distribution system due to the variability of its operation. Therefore, accurate load forecasting is necessary, and this will require the use of all available data held by the utility at all voltage levels. In this sense, a hierarchical structure is created in distribution systems, where smart meters allow obtaining granular data. In this paper, we present the hierarchical time series approach using different forecasting models to predict the load demand of a primary substation one hour ahead. To evaluate the performance of forecasting models, the Mean Absolute Percentage Error (MAPE) indicator is used. In this case, the bottom-up approach is used to forecast at the top level. The forecast results reveal that the hierarchical structure provides better performance with the forecast models employed.