Artificial Intelligence-Supported Disaster Satellite Project from BEDAŞ

Category: ELECTRICAL ENERGY , ENERGY AGENDA - Date: October 22, 2025

With the support of the Energy Market Regulatory Authority (EMRA), electricity distribution companies in the Marmara Region, including BEDAŞ, have developed next-generation, artificial intelligence-powered software that quickly identifies damage to electrical infrastructure after major disasters and instantly reports the status of transportation routes to field teams. The Disaster Satellite Project, featuring an AI model trained on a dataset of high-resolution satellite imagery from the Kahramanmaraş earthquake and other earthquakes in other regions around the world, has been successfully tested in pilot areas in the Marmara Region.

The electricity distribution sector is preparing to launch a next-generation project to strengthen its energy infrastructure against major disasters, particularly earthquakes, and to ensure the most effective field response in the event of a disaster. As part of an R&D project conducted by distribution companies serving the Marmara Region, including Boğaziçi Elektrik Dağıtım A.Ş. (BEDAŞ), with the support of the Energy Market Regulatory Authority (EPDK), next-generation AI-powered software has been developed that can quickly detect damage to the electrical infrastructure after major disasters.

The Disaster Satellite Project, featuring an artificial intelligence model trained using a dataset of high-resolution satellite imagery from the aftermath of the 2023 Kahramanmaraş earthquake and other earthquakes experienced in various regions around the world, has been successfully tested in the Marmara Region. The Disaster Satellite Project, which analyzes the damage to power poles, distribution centers, transformer centers, and critical energy points, particularly after an earthquake, using satellite imagery, enables rapid response. It instantly notifies field teams about which areas require more urgent action and the status of transportation routes.

Road conditions are being determined in the Marmara region.

With the R&D project "Development of a Next Generation Artificial Intelligence Model-Based Electrical Structural Elements Damage Detection System", which was developed using satellite images and Synthetic Aperture Radar (SAR) data immediately after natural disasters, it will be possible to quickly extract information from satellite images after major disasters, especially earthquakes.

The project, which allows teams to be quickly deployed to critical locations to ensure energy continuity after a disaster, has successfully proven its applicability in selected pilot areas within the Marmara Region's distribution companies. Following major disasters, the project will allow field teams to quickly determine the extent of damage to power poles, distribution centers, transformers, and step-down stations in the Marmara Region, as well as the status of roads providing access. Artificial intelligence-powered software, developed in partnership with other distribution companies serving the Marmara Region, will also be able to assess damage to electrical equipment after a disaster using satellite imagery.

TRAINED WITH A DATASET CREATED FROM HIGH-RESOLUTION SATELLITE IMAGERY FROM THE AFTERMATH OF THE KAHRAMANMARAS EARTHQUAKES

The AI ​​model, developed over a long period of industry development, was trained using the KATE-PD dataset compiled after the 2023 Kahramanmaraş Earthquake. This specialized dataset consists of 14 high-resolution images obtained from various satellite systems, covering a total area of ​​3,293 km². The images were meticulously labeled in 512x512 pixels and subjected to expert review, resulting in the generation of 1,952 high-quality damage labels. In this respect, KATE-PD has become one of the few examples in the world for reliable damage assessment from single-time satellite imagery after a disaster.

Istanbul Technical University UHUZAM (Satellite Communications and Remote Sensing Center) served as the contractor for the project, which analyzes the damage status of critical energy points after earthquakes through satellite images and provides rapid intervention.

Published by: Alperen ERTAŞ- [email protected]