How Does a Bird Detection System Work?
IdentiFlight’s technology combines high performance optical systems with the latest in machine vision and AI software. When installed as a network, IdentiFlight towers operate as an autonomous system with overlapping aerial coverage for the most detailed viewing available. Proprietary software and neural network technologies process the images to determine 3D position, velocity, trajectory, and protected species of interest, all within seconds of detection. By detecting a bird as far as one kilometer out and classifying it as a protected species such as an eagle (or not) in real time, the IdentiFlight system arms wind farm operators with critical visual and quantitative data needed to reduce or avoid collisions. Protected birds are conserved, and energy production loss is minimized.
IdentiFlight’s technology combines cutting edge technology in both image sensors and software.
Watch how it works in the video displayed to the left.
Image Sensors
An IdentiFlight imaging head includes eight wide field of view (WFOV) cameras, fixed in location. The job of the WFOV cameras is to detect moving objects in the environment and begin to track them. Bird detection can be complicated by other moving objects in the area, such as wind turbine blades, clouds, waving trees, cows, cars, people, airplanes, etc. The IdentiFlight machine vision technology automatically determines motion that is of interest and ignores the rest through “masking” algorithms.
Other technologies, such as radar, often claim to provide masking algorithms. However, the IdentiFlight masking algorithms are industry-leading with the ability to track a bird, even as it is flying within the rotor disk of a wind turbine. Radar technologies must mask out the entire rotor disk and, therefore lose the bird if it is in front of the wind turbine.
Once an object is detected, IdentiFlight points two movable high resolution stereoscopic cameras (HRSC) at the object. The job of the HRSC is to determine the distance to the object and gather the data necessary to determine if the object is the species of interest or not. Stereoscopic measurements derive distance.
Technology
Machine vision techniques consist of analyzing images against a set of rules to determine the degree to which an image matches the features sought. For example, is the bird of the right size? Does it have color characteristics matching the target species? Are the wing profiles in flight similar to the target species—for example, is it a flat glide profile of an eagle or the dihedral shape of a vulture?
Each frame gathered by the high-resolution stereo cameras (HRSC) is analyzed for the degree it matches a predetermined set of rules. Pattern recognition technology has developed catalogs of rules that analyze the images obtained by the HRSC.
In addition to machine vision techniques that analyze specific pre-determined attributes of an image, IdentiFlight also utilizes convolutional neural network technology. In this technology, millions of images gathered from IdentiFlight’s data set are used to train a neural network. The neural network is trained with images known to be the target species and tuned to reduce errors in predicting the species against a massive validation set of known bird images. IdentiFlight finds that with appropriate training, the neural network classification is superior even to machine vision techniques. Using the convolutional neural network data allows the system to continuously learn and improve as the data set grows. Learn more below.
Curtailment
IdentiFlight protects species via targeted curtailment of turbines. IdentiFlight detects and identifies protected birds at sufficient distance from specific wind turbines to initiate a curtailment of one or more turbines in order to reduce the risk of a bird colliding with that turbine. The curtailment logic resides on the IdentiFlight Base Station where signals can be sent automatically to the customer’s SCADA system for immediate action.
When determining a wind farm’s curtailment parameters, there are multiple variables that can be adjusted to achieve the desired cost/benefit level. There is not a “one size fits all” approach. Wind Farm A may be able to tolerate a set number of takes because of a take permit, but Wind Farm B may not be able to tolerate any takes and therefore would set a very conservative curtailment program. Curtailment criteria are ultimately defined by the operator and are highly configurable.
IdentiFlight’s highly effective detection capabilities for curtailment of wind turbines are driving the need to develop smarter ways of thinking about curtailment.
Dynamic data driven strategies are made possible by simulating various strategies to arrive at a balance for the need for avian conservation and revenue generation.
IdentiFlight data utilizes AI techniques to optimize the curtailment prescription. Studies suggest that bird behavior near some turbines may be inherently less risky than near others, signifying an ability to further optimize the balance between risk and conservation.
Our Journey to Curtailment
Detection
A bird is flying and comes within range as the IdentiFlight constantly analyzes wide-field images with its eight camera array, detects movement and points the stereo cameras at the bird, locking onto it.
Classification
IdentiFlight determines the distance of the bird with the stereo cameras then classifies it based on our neural network and the bird’s physical characteristics. The system then sends the data to a local database and the Base Station.
These first two steps occur in milliseconds.
Data Delivery
The base station sends the data to the graphical display on site, the curtailment module for analysis, and to the IdentiFlight Dashboard for real-time operational and data analysis.
Flight Analysis
The curtailment module analyzes bird flight path in relation to wind turbines and determines whether to curtail the turbine. Curtailment may be triggered by proximity, trajectory, or flight characteristics.
Curtailment
If the curtailment module detects a bird is in danger it will send off a curtailment signal to customer SCADA Network via OPC or other means. Additionally, the IdentiFlight Dashboard reports turbine curtailments as well as imagery that can be analyzed by qualified individuals to determine if the correct decision was made.
IdentiFlight is here to serve
wind farm operators. Like you.
Our system architecture & cybersecurity keep you in command.
An IdentiFlight system in a wind farm consists of one or more IdentiFlight towers that provide visual coverage for selected wind turbines. Each IdentiFlight tower detects, locates and identifies the species of interest. The IdentiFlight tower relays key information to a centrally located Base Station. At the Base Station, decisions are made according to a predetermined curtailment parameters regarding which turbines pose a risk to a protected bird and should be directed to shut down until the bird clears the danger.
Protecting data and networks from unauthorized intrusion is a key part of the IdentiFlight network design. Customers have their own cybersecurity policies and IdentiFlight will work with customers to tailor a network to meet customer policies and procedures.
These days, the modern world is mobile. Luckily, so is IdentiFlight
In addition to wind farm installations, IdentiFlight is also available as a mobile unit for use for field surveys by engineers, biologists, statisticians and academic stakeholders to address broader species of interest. The mobile IdentiFlight unit includes an imaging head with wide field of view and stereo vision sensors mounted on a telescoping tower and powered by a hybrid power system. It is installed on a ruggedized trailer for use in remote and challenging locations.
Because it is easy to mobilize, install and power, mobile systems provide a cost effective tool for monitoring and collecting data real-time in areas where a permanent installation would not be needed (studies, seasonal observations, etc.). With ruggedization, the technology can operate in environments that would be challenging for around-the-clock human observation, providing a valuable tool to enhance the quality of data and biologists’ reporting.
Challenging terrain?
Not a problem for IdentiFlight.
One of the first steps in integrating an IdentiFlight system into a wind farm is determining the siting of IdentiFlight towers. IdentiFlight towers are high precision instruments, and to maximize effectiveness, turbine layout and terrain must be carefully considered when determining placement.
Our installations are operating effectively in a variety of sites from cornfields, to mountains, to locations with heavy tree cover.
