Abstract
Power utilities are continuously set under high pressure to ensure the best performance of their grid. Nevertheless, power outages continue to be periodically observed ? for instance dure to vegetation in proximity of the power lines. The digitalization of our society offers new opportunities to acquire information relevant for the management of such problems. The access to more data sources also increases the pertinence of Dynamic Risk Analysis (DRA) and facilitates its implementation in power grid operations. However, the management of the considered datasets is a challenging task. In particular, it requires the definition of a comparison framework enabling to continuously and dynamically estimate the informative potential of those datasets. We address this problem by using a the Three-Phases Method to manage 17 different datasets originating from 12 different types of data sources. We look at how these datasets can inform eight parameters identified as impacting the probability of tree falls on power lines. For this, we calculate the informative potential of the datasets by applying a degradation factor due to spatiotemporal discrepancies and a degradation factor based on trust on the Default Maximum Potential of Knowledge (DMPK) provided by the data sources these datasets originate from. This enables to rank the datasets and select the one that best informs each of the parameters. We conclude that the approach used in the present work provides multiple opportunities for a more meaningful management of datasets in risk analysis, but also needs some adaptations to be implementable in an industrial context.
