Strategic bidding using reinforcement learning for load shedding in microgrids

A microgrid is a small-scale power system for providing reliable power supply to a small community. The goal of microgrid operation is to balance the amount of power supplied and the amount of power demanded. In the islanded operation mode of a microgrid, if power requirements are larger than the power generation, load shedding is used to solve the power balance problem. Load shedding restricts the use of power by consumers. In a distributed load-shedding approach, a control center allocates the power to power consumers through a bidding process. At that time, power consumers need bidding strategies to maximize their profits. We propose an optimal bidding strategy using a Q-learning algorithm. To evaluate the performance of the proposed bidding strategy, we implemented a microgrid operation system, and an experimental analysis was performed.