Relative fault vulnerability prediction for energy distribution networks

Aging infrastructure in energy distribution networks, e.g., electrical systems and district heating increases components’ fault vulnerability. This decreases the security of supply infers large costs on consumers utilities alike. In this paper, we explore application geospatial data processing machine learning as a means grasping complex relationship between pipes’ cables’ physical working environment their relative The paper presents vulnerability prediction network an network. A part work revolves around treatment data. regard, demonstrate method to combine from different sources, namely environmental data, georeferenced observations, GIS case how further augment with spatial clustering expert knowledge, injected form rule-based reconstruction. Additionally, propose new level imbalanced technique handle scarcity observations. Our results confirm, that our outperforms traditional techniques i.e., methods change opposed changing model or it is trained. also show for effectively identifies failure-prone pipes cables age-based ranking, which current industry practice.