Come meet the poster presenters to ask them questions and discuss their work
Check the programme for our poster viewing moments. For more details on each poster, click on the poster titles to read the abstract.
PO007: Preventing Bird Collisions using Artificial Intelligence, Machine Learning, radars and cameras while maximizing wind turbine availability.
Tassos Alefantos, CEO, NVISIONIST SA
As more and more wind parks are built around the world, low-conflict areas for wind turbines are becoming scarce, the global need for bird protection is rising and there is also a clear need for more accurate and reliable solutions. For the first time, an innovative system combines radars and cameras to achieve the required accuracy and reliability and at the same time to maximize wind turbine availability. The system can be installed to offshore wind parks and to large flat areas where onshore wind parks are located. The innovative solution uses radars and Ultra High Deﬁnition cameras of 12 megapixels in combination with Thermal vision technology to achieve 24 hours, all-weather detection and operation. The detection range of the radars is up to 10km. Then cameras are enabled to classify the targeted birds once the distance is less than 1km. The classification of the birds depends on the quality of the dataset the system has been fed. The more data available on a specific bird of interest, the better. When the birds are flying in a collision route (direction and height towards the rotor swept area) with the wind turbines the system sends an automatic command via the OPC server to slow down the rotor speed in order to further analyze the flight data. Such systems should be able to monitor the operation of the wind turbine by receiving input data from the SCADA system in order to improve their functionality. Furthermore, based on the process of detection and classiﬁcation, state-of-the-art acoustic driver modules with adjustable volume are used -if allowed- to deter birds from entering the wind turbine risk zone, making them change the route. In the extreme scenario that a bird enters the critical zone, the turbine (or group of WTGs) can receive automatic signals in various formats, in order to stop its/their operation and prevent the collision. The system is a big data application that communicates directly with wind turbines. Access to all this data is crucial both for the park owner and the environmental authorities; thus a report generator is utilized to produce a user-friendly dashboard that can be further adapted to the needs with graphs that are exported in various formats. Furthermore, all the information get uploaded in the cloud and are accessible to the ornithologists to remotely classify the birds that have not been classified automatically. Then, the system algorithms get “retrained” and the quality of detections and recognitions improves. After excessive R&D studies, the best performance achieved when the following criteria are met. * AI to maximize accuracy, with minimum false positive or false negative events * Able to estimate distance and height of the detected birds in different background contrast, especially in offshore environments via use of radars * Classification of bird species according to EIA * robust and able to withstand harsh weather conditions onshore. It should withstand ice falling from the nacelle or from sliding on the tower. * Avoid multiple rows of speakers. * Advance reporting capabilities to meet authorities' demands.