Distributed battery dispatch for uncertainty mitigation in renewable microgrids

In this work, we propose and analyze battery-level approaches to mitigate system welfare losses due to total energy forecast errors in energy-constrained microgrids. We present a receding horizon approach, develop a data-based method and heuristic approaches that leverage the receding horizon environment, and then analyze performance on a system based on real data. We find that the data-based method offers modest mitigation of the effects of total energy error, and a heuristic approach based on l2 regularization offers similar mitigation. We also find that the choice of load utility model affects mitigation strategies' profitability. Finally, we show how these methods can be used in a distributed framework.