Using machine learning to speed up grid analysis

Automating fault detection for DTEK Grids

Background

DTEK

DTEK Grids is the largest grid operator in Ukraine and it places its focus on electricity distribution and is also responsible for reliable electricity supply and development of the power grid. The company has built an effective model – its electricity losses are 5.5%, the lowest in Ukraine. They are constantly upgrading the power grids to integrate modern technologies.

DTEK Grids distributes every third kilowatt-hour of electricity in Ukraine, operating 188,463km of lines and serving 5.6 mln clients as of May 2019.

[1] DTEK Grids distributes every third kilowatt•hour of electricity in Ukraine

Hepta

The success story of Hepta has become known due to the mix of unique skill-set and experience in the team:

  • 15+ years of experience in on-ground grid inspection
  • 7+ years of experience in helicopter grid inspection
  • 8+ years of experience in drone grid inspection
  • Including data scientists working with satellite data, software engineers deploying machine learning models, aerospace engineers having previously built manned helicopters

The aforementioned combination of unique skill-sets enabled us to approach DTEK as a partner in increasing the reliability of the grid.

Previous situation

DTEK Grids’ asset management procedures were very traditional for the industry – involving ground patrol and motorised vehicles.

Having acquired many smaller grids, the condition of the infrastructure across the board was uneven and inspecting over 180,000km of grid rapidly is a daunting challenge.

Hepta approached DTEK Grids, who are inclined towards innovation by nature and offered to increase the inspection efficiency by introducing drones to the process.

In 2019 Hepta signed a contract for a pilot project including 280km of inspections to prove the automated analysis capability.

The pilot project yielded promising results and the two companies entered a full-scale partnership to fine-tune the automated defect analysis for DTEK Grids’ specifics with the ultimate goal to reduce costs and increase the quality of the service.

First benefits

During the pilot project, it was evident that considering the state of the grid it was sensible to pick the lowest hanging fruit first – increase the efficiency of manual inspections by utilising drones and automated software-based analysis.

DTEK Grids created drone operator teams, who gathered the data in the field. The data was then ingested into uBird platform – a cloud-based power line infrastructure platform developed by Hepta and powered by artificial intelligence.

Naturally, not all of the defects are found by the algorithms, so a combination of manual and automatic processes was used. Artificial Intelligence was used to speed up the analysis by eliminating the need for searching for certain defects.

Hepta and DTEK are continuing with the projects and are expecting to quadruple current drone-based inspection speed by the end of the year 2020.

DTEK Grids Roadmap for Increasing Analysis Efficiency

  • Equip more inspection crews with drones
  • Increase the level of automation
  • Train the artificial intelligence to recognise all of the most critical defects

Contact us to learn more about digitizing infrastructures and conducting utility inspection with drones and AI

Get in touch