Data driven net load uncertainty quantification for cloud energy storage management in residential microgrid.

Residential communities are increasingly adopting renewable energy sources (RES) to minimize energy consumption costs. However, these RES are weather-dependent and uncertain, posing challenges to ensuring reliable operations. Addressing the uncertainties in power supply management becomes a critical research question. Energy storage systems play a crucial role in providing battery-powered supply for residential loads under uncertain conditions. The operation of microgrids is directly influenced by uncertainties. This paper proposes data-driven-based net load uncertainty quantification fusion mechanisms for cloud-based energy storage management with renewable energy integration. Firstly, a fusion model is developed using SVR, LSTM, and CNN-GRU algorithms to estimate day-ahead load and PV power forecasting errors. After that, two mechanisms are proposed to determine the day-ahead net load error. In the first mechanism, the net load error is directly forecasted, while in the second mechanism, it is derived from the forecast errors of load and PV power. The net error analysis is conducted with a statistical mean and standard deviation, resulting in different uncertainty-bound confidence intervals around the forecasted value. Subsequently, the cloud energy storage system operation cost is calculated with the best uncertainty quantification mechanism for two different case studies. This approach allows for better management of uncertainties in energy storage systems and enables more informed decision-making under varying conditions. • Learning-based fusion algorithm for net load forecasting. • Develop uncertainty mechanism for uncertainty quantification. • Economic assessment of cloud energy storage. [ABSTRACT FROM AUTHOR]