In 2019, every inhabitant of Berlin was without electricity for an average of 34 minutes, which was a new negative record. The economic loss caused by production stops during the outage easily surpasses millions of Euros and may lead to legal proceedings on the liability.
Power outages are often connected to weather conditions, outdated infrastructure components, congested networks or on a change of vegetation and landscape. Thus, the information about the outage has to be combined with weather forecasting, infrastructure and earth observation data to generate a data set that covers all relevant features and weights them with respect to their influence on outages.
The image data of the visible and radar spectrum from earth observation data sets can be processed by using convolutional neural networks (CNN). These can detect and classify objects for identifying change of vegetation and landscape, such as trees ranging into electricity grids.
The outage data combined with the features from vegetation and landscape evaluation and weather forecast data - which can be processed by decision tree or regression tree models, such as Bayesian Additive Regression Tree (BART) - can be trained. This combined trained model is then capable of predicting where the probability of outages is highest.