An adaptive Subgrid Identification Strategy for Future Medium Voltage Smart Grids based on Machine Learning

This paper proposes a novel strategy for adaptive autonomous identification of subgrids in medium voltage (MV) smart grids. The algorithm is based on machine learning. In contrast to a intended islanding of grids, the term subgrid describes a subordinate energy system of indeterminate size, which is physically isolated from the interconnected grid and continues to operate after disconnection. This process takes place outside the control of the distribution system operator. Due to the risks for people and equipment, subgrids must be identified reliably and quickly in order to disconnect them from the power supply. With the purpose of executing the subgrid identification strategy (SIS) in a scalable manner in future MV smart grids, a novel engineering process based on a virtualisation of smart grid services has been developed.