AI-assisted powergrid inspection

Offentlig prosjektsammendrag

Given the aging infrastructure of utilities companies and the gradual expansion of installations, the demands of inspection have been growing significantly. In the general context of inspection procedure, a quite demanding and time consuming requirement constitutes the thorough manual inspection of imagery captured during inspections by manned or unmanned aerial vehicles. To address this condition, an important shift is required, in order to establish automated, robust and highly efficient solution for faults detection on utilities infrastructure, which along with the cost reduction will increase the reliability of defects predictions. This project envisions an intelligent framework that will set the path for a fully automated detection of defected components laid on utilities masts and overall power grid. This new generation tool will be capable to automatically execute infrastructure captured images inspection, give predictions of defects over mast components, recognize severity of faults and suggest maintenance works. It will also provide intuitive and user-friendly interface via cloud services to human inspectors. Hence, inspectors’ job will become less time consuming, but more targeted and important for giving feedback towards the reconfiguration, accuracy improvement and fidelity of this artificial intelligence tool. Furthermore, the scope of this framework is to be deployable and ready-to-use as a supplementary part of infrastructure service operations. Since this project aims for direct exploitation in utilities market, it will be evaluated in large scale real infrastructure environments. The accomplishment of the envisaged scenarios will boost the goal of retaining utilities companies’ competitiveness and will particularly impact our company services on accomplishing more efficient and complete power grids inspections.