Potential of prosumer real-time scheduling to improve grid area load forecasts in distribution grids

Grid state forecasting is a common practice at the transmission level for optimising the grid operation and implementing redispatch actions. However, transferring these methods to the distribution grids presents significant challenges, primarily due to the limited availability of measurement devices at this level. Previous research developed a modular grid state forecast for medium and low-voltage grids to address this issue. This approach seeks to optimise the use of limited measurement data and effectively predict imminent grid congestions. One of the critical challenges in this context is accurately forecasting load demand in local grid areas, as this must be done for a small number of customers. Their aggregate effect is not negligible and has a stochastic characteristic. Schedules provided by individual customers or prosumers can be utilised to enhance load forecasting accuracy in grid areas. These schedules are provided by software agents that forecast the stochastic load and its domestic energy systems using physical models. The resulting forecasts are integrated into the mathematical framework for determining the grid area load forecast. Simulation studies and field tests have demonstrated that prosumer real-time scheduling enhances grid area load forecasts and helps to improve the load distribution, especially during periods of high peak loads.