Challenge
Reduce impact of turbines on local ecosystems
A key challenge facing the wind industry is the potential for turbines to adversely affect wildlife and local ecosystems both directly, via collisions, and indirectly due to noise pollution, habitat loss, and reduced survival or reproduction. This may be a problem through the construction phase, but it can also persist through daily operations.
When assessing a new site for a wind farm, a developer might be required to map out the wildlife—especially flying animals such as birds and bats. In the case of an offshore wind farm, this requirement might include marine life.
These assessments are important but costly processes; in one case, a developer needed to hire the services of 20 full-time ornithologists for a year to complete the assessment.
Once the wind farm is in operation, operators seek to both track local wildlife and protect it from harm. This can be done for example by turning off or slowing down turbines when birds are migrating through the site.
Solution
Bird detection using computer vision
Mainstream Renewable Power, a pure-play developer of wind and solar power, is working with Cognite to lower the bird fatality rate at the company’s onshore wind farms.
Mainstream set up three stationary cameras at one of its wind farms in South Africa to analyze bird activity using computer vision. The information—nearly 10,000 images a day—is ingested into Cognite Data Fusion® using a custom-made image extractor and then visualized in an analytics dashboard built using low-code technology from Plotly.
In Cognite Data Fusion®, users can find and review images and add manual annotations. The dashboard adds additional functionality. Users can select a specific camera and see the number of bird detections over time, and see activity pattern analytics based on the number of detections.
Impact
The wind power industry is experiencing major growth around the world. As more wind farms begin to generate power, operators will be challenged to prove how they are safeguarding the ecosystems that their installations are located in.
With automated bird detection enabled by Cognite Data Fusion®, Mainstream’s experts can easily monitor bird activity at its sites, and the company’s development team can use the information about bird activity patterns to optimize wind farm layouts.
Continuous data capture enables a cost-effective way to manage turbine shutdowns with reduced risks to health and safety. Storing that data in a central location makes the information easy to access and share, improving Mainstream’s collaboration with authorities and external experts.
The benefits of using computer vision to detect birds will further improve over time. The more images the system captures, the more the model will improve, resulting in better detection in the future. For example, while the current model detects any kind of bird, a future version could distinguish between different species.
