Deep reinforcement learning-based network for optimized power flow in islanded DC microgrid

This paper presents an optimum power flow control for islanded microgrid employing deep reinforcement learning. During abnormal grid conditions, the stability of the microgrids is very important to avoid grid outages. In abnormal grid condition, the microgrid operates in the islanded mode for providing uninterrupted supply to loads and stability improvement with power resilience. This islanded operation depends on the effective operation of connected Distributed Renewable Energy Sources (DRES). This paper aims to provide optimum power dispatch, and accurate control of connected DRES enables the grid to restore service. Advanced data-driven control method could provide a solution to grid outages and DRES service support. In this research work, a Deep Reinforcement Learning Network (DRLN) was proposed to identify the State of Charge (SoC) for optimum power flow during source outages, communication link failure, and communication bus failure. A finite horizon policy update provides updated DRES SoC and makes the grid to operate on the islanded mode of operation with service. This provides an optimized power flow and accurate performance. By analysing the voltage, current, and weighted average power trajectories, the performance of the proposed DRLN was validated. The simulation was studied with a constant ZIP load, and also the proposed algorithm was implemented in IEEE 15 bus system to demonstrate its effectiveness. Finally, Hardware in Loop simulation is implemented to validate the results of the proposed DRLN.