Improved load and generation forecasting for extended day-ahead estimates in the Nordic grid

This paper presents a novel method for forecasting the load and wind generation of the Nordic power system within a 36-hour window, using a MV-LSTM network that incorporates weather data from six regions in Norway. The model utilizes the best estimates of wind generation in Norway, extracted from wind production data in Denmark and Sweden, to anticipate load and wind generation 36 hours in advance. The difference between projected load usage and wind generation approximates the grid's hydropower production, allowing for the optimization of hydropower generation for power producers and system operators. This supports a robust energy market with increasing renewables integration. Additionally, the proposed load and generation estimates are further combined with a 21-bus Norwegian transmission network test system representing the NO2 region for grid analysis, resulting in an enhanced day-ahead scheduling approach for input to hydropower optimization.