Why the project is important
The energy transition requires a grid infrastructure that is secure and highly available. Traditional inspection processes are time-consuming and costly. Swissgrid is developing a solution that automatically analyses the condition of overhead lines: the Smart Image Database. This will form the basis for optimising maintenance cycles and extending service life in the future. The project also strengthens international cooperation: as part of a cross-industry proof of concept, it shows how shared data pools and AI technologies can accelerate the digitalisation of maintenance.
Project
Challenge
The inspection of overhead lines is a complex, costly process. Highly specialised experts have to manually check thousands of images to identify damage such as corrosion or material fatigue – a time-consuming task that ties up valuable resources. The goal is to ensure efficient grid maintenance.
Dependence on external analyses and lengthy manual evaluation are slowing down the digitalisation of maintenance. This harbours risks: the longer it takes to detect damage, the higher the consequential costs and the greater the risk to security of supply.
Solution
The goal is automation instead of manual control – as far as the state of the art allows. This innovation is centred around the Smart Image Database. Using an explorative approach, the aim is to create a comprehensive image data base for overhead line components and to develop AI models that automatically detect damage and anomalies. The images shown are from drone flights as part of the MIDAS project, which involves systematically recording and documenting overhead line components. The images are used as sample material.
- Shared image data base: Image data showing overhead line components is used as the basis for AI-assisted analysis.
- AI training: AI models are trained and gradually taught to detect damage and anomalies by themselves.
Benefits and results
Faster damage detection reduces costs, while quality is improved thanks to a large, standardised data base. Smart Image Database digitalises maintenance and reduces dependency on external service providers. This collaborative approach results in AI-assisted grid inspection as a service that strengthens security of supply and extends the service life of infrastructure. It is designed for specialists carrying out maintenance, inspection and servicing on behalf of transmission and distribution system operators.
Current project status/outlook
The project will run from June 2023 to August 2026 and is currently in the further development and optimisation phase. Now that initial data base structures have been created and AI models have been developed and enhanced, the focus is on using and expanding synthetic image data to improve damage detection even further. By the end of 2025, all the relevant images had been collected, prioritised and assigned suitable labels. Synthetic images are constantly being added to the datasets. The development, training and validation of the new AI models has been under way since January 2026. The plan is to fully optimise the model by August 2026.
Collaboration with partners and agile project management ensure continuous development and a high level of innovation. With visible success: in 2024, the Smart Image Database AI project was awarded the MAINTAINER – Special Prize for innovation by the MainDays platform for innovation and best practice in maintenance.
