Towards Data-Driven Malfunctioning Detection in Public and Industrial Power Grids

Investigation of a concept for remote detection of malfunctioning grid supporting devices using minimal data is proposed in this work. The operation of future electricity grids highly depends on the behaviour of these devices and their support functions such as reactive power dispatch used for voltage control. Synthesising and utilizing operational data of a distribution grid, a functionality is being developed to enable better surveillance of grid connected devices ensuring security of supply and resiliency. In a first step, data driven approaches for anomaly detection are explored. They are applied to the operational data of the device to detect unwanted behaviour. Results show first indicators of applicability and possible obstacles.