Source-linked AI summary
Vibration-Based Damage Detection in Wind Turbine Blades using Phase-Based Motion Estimation and Motion Magnification
Aral Sarrafi, Zhu Mao, Christopher Niezrecki, Peyman Poozesh
TL;DR
Attached sensors used in vibration-based SHM can cause mass loading and are labor-intensive for large structures, motivating non-contact measurement. This paper applies PME and phase-based motion magnification to video of a 2.3-meter wind turbine blade, extracting modal information for baseline and damaged cases. The method demonstrates non-contact damage detection without mass loading, while its single-camera setup is limited to planar motion and manual image cleanup can be time-consuming.
Problem
Vibration-based SHM needs dynamic measurements for damage identification, but contact sensors can induce mass loading and large-scale instrumentation can be labor-intensive and spatially limited.
Method
The paper uses single-camera PME and motion magnification to extract resonant frequencies and operating deflection shapes from video of a 2.3-m wind turbine blade.
Results
The proposed approach identifies baseline and damage-related dynamic changes in a complex wind turbine blade without instrumented sensors, surface treatment, mass loading, or altered structural stiffness.
Takeaways & Limitations
Video-based PME and motion magnification can support routine non-contact inspection of wind turbine blades within the demonstrated non-rotating test setting.
Takeaways & Limitations
Single-camera PME captures only planar motion parallel to the image plane, so edgewise modes normal to that plane cannot be detected; edge detection also requires time-consuming manual cleanup.
Abstract
from arXiv · showhide
Vibration-based Structural Health Monitoring (SHM) techniques are among the most common approaches for structural damage identification. The presence of damage in structures may be identified by monitoring the changes in dynamic behavior subject to external loading, and is typically performed by using experimental modal analysis (EMA) or operational modal analysis (OMA). These tools for SHM normally require a limited number of physically attached transducers (e.g. accelerometers) in order to record the response of the structure for further analysis. Signal conditioners, wires, wireless receivers and a data acquisition system (DAQ) are also typical components of traditional sensing systems used in vibration-based SHM. However, instrumentation of lightweight structures with contact sensors such as accelerometers may induce mass-loading effects, and for large-scale structures, the instrumentation is labor intensive and time consuming. Achieving high spatial measurement resolution for a large-scale structure is not always feasible while working with traditional contact sensors, and there is also the potential for a lack of reliability associated with fixed contact sensors in outliving the life-span of the host structure. Among the state-of-the-art non-contact measurements, digital video cameras are able to rapidly collect high-density spatial information from structures remotely. In this paper, the subtle motions from recorded video (i.e. a sequence of images) are extracted by means of Phase-based Motion Estimation (PME) and the extracted information is used to conduct damage identification on a 2.3-meter long Skystream wind turbine blade (WTB). The PME and phased-based motion magnification approach estimates the structural motion from the captured sequence of images for both a baseline and damaged test cases on a wind turbine blade.
1 Introduction
The paper motivates non-contact vibration-based SHM for wind turbine blades by addressing the limitations of attached sensors and introduces PME with video magnification as an alternative. It applies the approach to baseline and damaged blade conditions to identify changes in dynamic behavior.
- Vibration-based SHM identifies damage through changes in structural dynamic behavior, commonly using EMA or OMA.
- Attached accelerometers can add mass loading to lightweight structures, potentially changing their dynamics and reducing measurement fidelity.
- Laser vibrometers avoid mass loading but are relatively expensive and measure responses sequentially, increasing measurement time for high spatial resolution.
- Digital video cameras provide low-cost, simultaneous full-field measurements without physical contact, while stereo methods such as 3D DIC and 3DPT can require substantial computation.
- The study applies PME and video magnification to a 2.3-m wind turbine blade, extracting resonant frequencies and operating deflection shapes for baseline and damaged conditions.Damage is simulated by attaching an external mass, and deviations from the undamaged dynamic behavior are examined.
- The proposed workflow processes image sequences to extract modal information for condition decisions, including simulated icing represented by added mass at the blade tip.Histogram manipulation is used to address poor lighting and improve operating-deflection-shape visualization.
2 Theoretical Background on Phase-Based Motion Estimation (PME) and Video Magnification
PME estimates local motion by tracking phase changes in complex Gabor-wavelet coefficients, while phase-based magnification filters and amplifies selected motion frequencies. A noise-free Gaussian example shows estimated motion closely following the actual motion.
- Intensity-based optical flow is sensitive to illumination changes and noise, motivating phase-based motion estimation for video motion extraction.
- Phase-Based Motion Estimation: PME uses finite-support 2D Gabor wavelets to transform each image frame into complex coefficients that represent local motion.Gabor wavelets are sinusoidal functions modulated by Gaussian envelopes, with orientation controlled through rotated spatial variables.
- Phase-Based Motion Estimation: The transformation maps image intensities I(x,y,t) into complex-domain coefficients C(u,v,t) through convolution with a Gabor wavelet.
- Phase-Based Motion Estimation: Phase differences between consecutive frames are proportional to motion, with Δϕ(u,v,t) = hδ_x and h = 2π/λ in the described case.
- Phase-Based Motion Estimation: In a noise-free Gaussian-surface simulation, PME estimates align very well with actual motion, supporting application to subtle wind-turbine-blade vibrations.
3 Experimental Test Setup for the Wind Turbine Blade (WTB)
The experiments compared baseline and temporarily damaged wind turbine blades using controlled impulse excitation and high-speed video captured across the blade. Damage was emulated by adding mass at the blade tip, while PME and motion magnification processed the recorded vibration data.
- A Skystream® 4.7 wind turbine blade was clamped at its root and excited with a modal-hammer force impulse.
- A single 4-megapixel PHOTRON® high-speed camera with a 14-mm lens recorded the blade motion, using 8-bit pixel intensities from 0 to 255.
- Both tests used approximately consistent excitation at 43 cm from the clamped root, with the camera positioned 1.8 m above the blade to capture full in-plane bending.
- The camera field of view included both the blade tip and root, and image contrast was enhanced for visualizing testing results.
- Temporary damage was induced by mounting 0.385 kg of additive mass at the blade tip, approximately 5% of total blade mass, to emulate icing.
- The test configuration provided vibration video data from baseline and damaged blades for subsequent PME and motion-magnification analysis.
4 Results and Discussion
The paper applies PME and motion magnification to extract resonant frequencies and ODS from baseline and damaged wind turbine blade videos. The extracted dynamic features are compared with EMA and between conditions to assess damage detection.
- Analysis workflow: PME extracts resonant frequencies and operating deflection shapes from recorded blade vibrations for subsequent damage identification.The workflow transfers estimated motion to the frequency domain, identifies resonant frequencies by peak picking, and reconstructs ODS at selected frequencies.
- Experimental setup: A 2.3-m Skystream wind turbine blade is tested in baseline and damaged conditions, with damage simulated by adding external mass.The experimental setup uses hammer excitation and high-speed video capture; the added mass changes the blade’s mass distribution.
- Resonant frequencies: PME detects only planar flapwise motion because the single camera lacks depth information, excluding the blade’s edgewise modes at 26.22 Hz and 65.78 Hz.Other resonant frequencies agree well with traditional EMA results, but the third and sixth modes cannot be detected in this configuration.
- Damage identification: Adding mass decreases the damaged blade’s resonant frequencies, enabling damage detection through frequency shifts measured by PME.The damaged and baseline frequency-domain motions are compared after PME estimation.
- Operating deflection shapes: Motion magnification isolates selected resonant-frequency content and produces videos representing the corresponding operating deflection shapes.The center frequency is selected from PME-estimated resonances, while the passband is tuned to retain the desired frequency content.
- Damage identification: ODS changes provide spatially informative damage evidence: the second and fourth bending shapes show altered forms and node and antinode locations.MAC is used to compare baseline and damaged ODS vectors, while PME-derived ODS agree with EMA at correlations above 85%.
5 Conclusion
The paper proposes and demonstrates PME with motion magnification for non-contact damage detection in wind turbine blades. Its scope is limited to additive-mass damage, motivating studies of other damage types.
- PME and motion magnification enable non-contact structural damage detection on wind turbine blades.The approach is demonstrated for an industrial, complex-geometry structure using a commercially available camera.
- The approach avoids mass loading, structural-stiffness changes, instrumented sensors, and surface treatment.
- Image enhancement improves video-frame quality under limited lighting and contrast before extracting operating deflection shapes.
- Other damage types beyond additive mass must be studied to evaluate the full potential of PME and motion magnification for SHM.