Distant vibration measurement of wind turbines in operation (WEALyR)

The tracking system is based on a stationary camera that captures the rotor in full format. As a first step, the wind turbine must be separated from the background in the image of this camera. This is essentially done by creating a difference between successive images in the video stream. Stationary elements, and thus elements that are identical in two images, cancel each other out, leaving moving objects such as the rotor blades.

In the second step, the rotor blade tips are detected and the motion sequence of the rotor in space is determined from this. The entire image processing must run in real time. Due to the unavoidable latency of the data processing and the control system for the SNK control, the rotor blade position must be predicted for a small time interval into the future. For this purpose, a dynamic 3D model of the wind turbine is built up in the computer based on the detected rotor blade movements and continuously updated. The projection of the 3D model onto the detector plane yields the required azimuth and elevation angles for controlling the SNK.

In addition, the laser spot of the vibrometer is detected in the camera image and its actual position is compared with the desired target position. The deviation flows as a correction into the control of the SNK and compensates, among other things, for slight distortions caused by the camera optics, parallax errors and inertia and torques of the setup that are difficult to compensate for.

To enable detection of the measuring laser spot of the vibrometer, a SWIR camera (SWIR = Short Wave Infra Red) is used for the tracking procedure. Suitable optical filters (bandpass) ensure a balanced contrast ratio between laser and ambient brightness in the image, so that the laser spot and rotor blades can be detected equally.

The so-called Doppler effect describes the velocity-dependent frequency shift of a light wave when it is emitted or reflected by an object moving relative to the observer. A laser Doppler vibrometer uses this effect by comparing the frequency of its emitted laser beam with the light backscattered by an object to be measured. From the frequency shift, the relative motion between the measuring device and the object can be calculated.

A vibration is a periodic motion and thus causes a temporal modulation of the measured frequency shift. Similar to the music in FM radio reception, vibration information is obtained by demodulating the measured signal waveform.

In the interest of eye safety, commercial laser Doppler vibrometers do not have sufficient laser power for the measurement distance of several hundred meters required here. In addition, rotating rotor blades cannot be scanned with such vibrometers, because the inherent motion of the blades leads to an additional macroscopic Doppler shift. Not only is its magnitude many times greater than that of the vibrations to be measured, but its value and sign also change constantly with the rotation of the rotor at a high rate. On the one hand, this means that the signal exceeds the bandwidth of conventional receivers, and on the other, it cannot be easily demodulated due to the rapidly changing frequency offset.

The laser Doppler vibrometer built in the project operates at a laser wavelength of 1.5 µm. Thus, even a higher output power is still eye-safe, allowing larger measurement distances. Different concepts are used to handle the macroscopic Doppler shift due to the rotor's own motion. In the original development on the laboratory system, an electronic control loop continuously tracked the frequency offset of the receiver to the frequency shift caused by the target motion. In field measurements on large wind turbines, A/D converters with sufficient bandwidth were used to record even the macroscopic frequency shift. Digital signal evaluation subsequently enables the separation and analysis of proper motion and vibration information.