In operating wind energy converters (WEC), maintenance and repair represent a significant cost factor. Downtimes resulting from the need for this work must be minimized and failures caused by unnoticed faults prevented. A key contributor to the load on the installation and to material fatigue is the plant’s unavoidable vibration. Changes in the vibration characteristics can provide evidence of possible hidden damage. In productive operation, condition monitoring systems with sensors on the powertrain and, increasingly, also in tower and rotor blades are therefore used to collect vibration data. Especially in the blades, sensors can be installed only in selected, fixed positions during the blades’ production. If these fail in operation, they cannot normally be replaced.
Fraunhofer IOSB is developing a measuring system with which the vibrations of a wind turbine can be measured at a distance of 200 to 300 meters with high spatial resolution. The special focus of the project is on surveying the vibration of the spinning rotor blades during operation.
The actual measurements are taken using laser vibrometry: A laser beam is projected onto the point to be measured. If the illuminated surface moves parallel to the laser’s direction of propagation, the refracted light experiences a frequency shift due to the Doppler effect. The extent of this shift can be measured by comparing the refracted with the emitted light. Surface vibrations result in a time modulation of the frequency shift. To measure a rotor blade of a wind turbine in operation by this method, the laser spot must track a point on the surface of the rotating blade for several seconds with an accuracy of just a few centimeters. For this purpose the vibrometer is mounted on a pan–tilt head – a carrier platform that can rotate horizontally and vertically at high angular precision.
The challenges in the project consist in the development of a vibrometer suitable for measuring a moving object, and a tracking method capable of detecting the rotor’s motion and controlling the pan–tilt head.
The tracking system is based on a stationary camera that captures the entire rotor. The first step is to isolate the rotor from the background in the camera image. This segmentation is problematic in the visual spectrum due to the strong variations in contrast over both time and space. For this reason, an infrared camera is used. The second step consists of detecting the blade tips to determine the rotor’s motion in space. Image processing must take place entirely in real time. Due to the latency inherent in data processing and in the pan–tilt head’s control system, the blade position must be predicted for a brief period in the future. To this end, a dynamic 3D model of the blade is generated and continually updated from the detected rotor blade motion. The projection of the 3D model onto the detector level yields the required azimuth and elevation angles for controlling the pan–tilt head. In addition, the laser spot from the vibrometer is detected in the camera image and its actual position compared to the desired target position. The deviation is fed back to the pan–tilt head controller as correction factor to compensate the slight distortions caused by the camera optics, parallax error, hard-to-compensate inertia and torque effects of the substructure, and other factors.
The Tracker and Track Evaluation group of the OBJ department is contributing the image pro-cessing methods. From these, the Heterogeneous Hardware Structures group develops the real-time control of the pan–tilt head.
Due to eye safety considerations, commercial laser vibrometers do not have sufficient laser power for the measuring distance required here. Furthermore, these vibrometers are not capable of scanning moving rotor blades. The blades’ movement would result in an additional macroscopic Doppler shift, which would be superimposed on the signal of the vibration measurement, thereby exceeding the receiver’s bandwidth. A vibrometer that works at a laser wavelength of 1.5 µm has therefore been developed to allow a higher but still eye-safe output power in order to cover measuring distances of several hundred meters. Camera, optics and filters that are sensitive to this short-wave infrared wavelength and capable of representing wind turbine and laser spot in a balanced contrast ratio have been chosen for the tracking. Due to the motion of the measured object – the rotating rotor blade – the vibrations are overlaid with a macroscopic velocity component. The resulting, relatively strong frequency shift of the laser line must already be compensated in the receiver, and this compensation matched to the constantly changing frequency offset. Finally, the impact of the vibrometer and pan–tilt head’s own movement on the measuring signal must be neutralized.
The Laser Sensors group of the OPT department is developing the laser vibrometer, and the Optronic Sensor Systems group is responsible for project coordination and for the interaction of the components.
A scale model of a wind turbine that meets the real-life requirements regarding angles and dynamics was built in the laboratory. It was used in particular as test arrangement for implementing the tracking and pan–tilt head control algorithms. The resulting prototype was demonstrated at CeBIT in 2014. It allows stabilized positioning of the vibrometer laser spot on an arbitrary point on the rotating blades of a model wind turbine (rotor diameter about 2 m, measuring distance about 10 m). The vibrometer supplies the signal characteristic and the performance spectrum of the rotor blade vibrations at the selected point. An additional, green pilot laser was used for visualization.
Video: CeBIT 2014 (german)
The next step consists of adjusting the system for recording vibration data of a real wind turbine, since laser and camera had to be modified for the much shorter distances in the laboratory to develop the methodology. The control system will then be extended to allow automatic scanning of spatial vibration patterns of the spinning rotor blades. To date, measuring points could be approached only individually by hand.
In the long term, an extension of the tracking system for offshore applications is also conceivable. By further developing the active laser alignment tracking, the inevitable relative movements between sensor platform (such as a ship) and the measured installation could conceivably be compensated.
R. Ebert, "Laser vibrometry for wind turbines inspection". Keynote presentation at SPIE Smart Structures/NDE 2016.Proc. SPIE 9804, Nondestructive Characterization and Monitoring of Advanced Materials, Aerospace, and Civil Infrastructure 2016, 98040H (April 8, 2016); doi:10.1117/12.2219559