Prediction of weather-related faults
By using weather observations, archived forecasts, and fault reports (for example registered large-scale faults and FASIT, the goal of this demonstrator is to look for relationships between specific weather conditions and grid faults.
One application of this is that some of the unclassified faults in FASIT now can be classified as weather-related faults, which can again lead to better network planning decisions. Going a step further, another application is that faults which are known to be weather-related and information about the weather conditions in the incident they occurred can be used to train a machine learning model to recognise weather conditions in which there is a higher probability for faults. Thus, given a weather forecast, one can obtain the risk for weather-related faults to occur. This is valuable in the operation of the distribution network, because the operational readiness can be improved in case there is a high risk for weather-related faults in an area. In ENERGYTICS, this second application is further developed.