Dear M.E. Community,
You are invited to attend the PhD Dissertation Defense of Linfeng Lyu on Thursday, October 31, beginning at 3pm, in the Mechanical Engineering Office Conference Room (ENGR210-I). You may also wish to join remotely, using this Webex link.
Advisor: Dr. Weidong Zhu
Title: Novel Image-based Tracking Continuously Scanning Laser Doppler Vibrometer Systems for Vibration Measurement of Rotating Wind Turbine Blades
Abstract:
The objective of the proposed research is to develop efficient vibration monitoring and structural health monitoring (SHM) methods for a horizontal-axis rotating wind turbine blade using a novel image-based tracking continuous scanning laser Doppler vibrometer (CSLDV) and new signal processing methods associated with tracking continuous scanning laser vibrometry. The image-based tracking CSLDV consists of a camera, a scanner with a set of orthogonal mirrors, and a single-point laser head. Currently, there are no suitable non-contact in-situ vibration monitoring and SHM methods for rotating wind turbine blades, and vibration monitoring and SHM of wind turbine blades are mainly accomplished by visual inspection of stationary blades or by using a limited number of embedded sensors. The image-based tracking CSLDV is capable of rapidly obtaining spatially dense vibration shapes (VSs), such as mode shapes and operating deflection shapes (ODSs) of a rotating structure under random excitation, by continuously sweeping its laser spot over the structure surface. VSs measured by the image-based tracking CSLDV can have more measurement points than those by a commercial scanning laser Doppler vibrometer in a point-by-point manner, and the former needs much less measurement time. Different image processing methods are developed for the image-based tracking CSLDV to extract the rotating structure from its background in captured images of the camera. One image processing method is to determine the difference between two neighboring frames captured by the camera to find the position of the rotating structure. The other image processing method is to average multiple captured images of the camera to obtain the background of the rotating structure and subtract the averaged image from captured images of the camera to determine the position of the rotating structure. Edge detection algorithms can be applied to processed images by the two methods to determine positions of the rotating structure.
Estimated real-time positions of the rotating structure can be used for estimating its rotation speed. When the distance between the image-based tracking CSLDV and the rotating structure is long enough, rotation angles of mirrors of the scanner in the image-based tracking CSLDV are small and the position of the laser spot is basically linearly related to rotation angles of mirrors of the scanner. Pixel coordinates of the laser spot in captured images of the camera are linearly related to the position of the laser spot when the camera is close to the scanner; thus pixel coordinates of the laser spot in captured images are basically linearly related to rotation angles of mirrors of the scanner. Once the position of the rotating structure is determined in captured images, a scan path can be generated on it and rotation angles of mirrors of the scanner can be controlled so that the laser spot of the image-based tracking CSLDV can be swept along the path. One-dimensional and two-dimensional scan schemes are developed for the image-based tracking CSLDV to generate straight and zigzag scan paths on the rotating structure, respectively. New operational modal analysis (OMA) methods are developed based on rigorous rotating beam and plate theories, which can estimate modal parameters, such as damped natural frequencies, modal damping ratios, undamped mode shapes, and ODSs, of the rotating structure under random excitation. One OMA method is the lifting method that can convert a measured response of the rotating structure into responses on multiple virtual measurement points on the scan path. Correlation functions among converted responses can be calculated and used for determining modal parameters of the rotating structure. The other OMA method is the improved demodulation method that filters a measured response by a band-pass filter that only contains one damped natural frequency of the rotating structure. The filtered response is multiplied by a sinusoidal signal whose frequency is the damped natural frequency and a low-pass filter is applied to the resulting response to obtain undamped mode shapes of the rotating structure. A novel demodulation method with a reference signal is developed to identify positions of damages in a beam without its base-line information. A cross-correlation function between a measurement of a CSLDV and a measurement of a single-point laser Doppler vibrometer is calculated and processed by the demodulation method to estimate its undamped mode shapes. Curvatures of estimated undamped mode shapes are calculated to obtain curvature damage indices and determine the position of a damage in the beam. Prototypes of an image-based short-range tracking CSLDV and an image-based long-range tracking CSLDV are developed for experimental validation of the proposed methods. Rotation speeds and modal parameters of rotating blades with different speeds that are excited by air flow that is considered as random excitation are estimated. Both undamped mode shapes of rotating blades on a straight scan path and their full-field undamped mode shapes are estimated and compared with each other. The proposed methods can address major challenges to monitor the vibration of a rotating horizontal-axis wind turbine blade and detect its potential damage in operational conditions.
Title: Novel Image-based Tracking Continuously Scanning Laser Doppler Vibrometer Systems for Vibration Measurement of Rotating Wind Turbine Blades
Abstract:
The objective of the proposed research is to develop efficient vibration monitoring and structural health monitoring (SHM) methods for a horizontal-axis rotating wind turbine blade using a novel image-based tracking continuous scanning laser Doppler vibrometer (CSLDV) and new signal processing methods associated with tracking continuous scanning laser vibrometry. The image-based tracking CSLDV consists of a camera, a scanner with a set of orthogonal mirrors, and a single-point laser head. Currently, there are no suitable non-contact in-situ vibration monitoring and SHM methods for rotating wind turbine blades, and vibration monitoring and SHM of wind turbine blades are mainly accomplished by visual inspection of stationary blades or by using a limited number of embedded sensors. The image-based tracking CSLDV is capable of rapidly obtaining spatially dense vibration shapes (VSs), such as mode shapes and operating deflection shapes (ODSs) of a rotating structure under random excitation, by continuously sweeping its laser spot over the structure surface. VSs measured by the image-based tracking CSLDV can have more measurement points than those by a commercial scanning laser Doppler vibrometer in a point-by-point manner, and the former needs much less measurement time. Different image processing methods are developed for the image-based tracking CSLDV to extract the rotating structure from its background in captured images of the camera. One image processing method is to determine the difference between two neighboring frames captured by the camera to find the position of the rotating structure. The other image processing method is to average multiple captured images of the camera to obtain the background of the rotating structure and subtract the averaged image from captured images of the camera to determine the position of the rotating structure. Edge detection algorithms can be applied to processed images by the two methods to determine positions of the rotating structure.
Estimated real-time positions of the rotating structure can be used for estimating its rotation speed. When the distance between the image-based tracking CSLDV and the rotating structure is long enough, rotation angles of mirrors of the scanner in the image-based tracking CSLDV are small and the position of the laser spot is basically linearly related to rotation angles of mirrors of the scanner. Pixel coordinates of the laser spot in captured images of the camera are linearly related to the position of the laser spot when the camera is close to the scanner; thus pixel coordinates of the laser spot in captured images are basically linearly related to rotation angles of mirrors of the scanner. Once the position of the rotating structure is determined in captured images, a scan path can be generated on it and rotation angles of mirrors of the scanner can be controlled so that the laser spot of the image-based tracking CSLDV can be swept along the path. One-dimensional and two-dimensional scan schemes are developed for the image-based tracking CSLDV to generate straight and zigzag scan paths on the rotating structure, respectively. New operational modal analysis (OMA) methods are developed based on rigorous rotating beam and plate theories, which can estimate modal parameters, such as damped natural frequencies, modal damping ratios, undamped mode shapes, and ODSs, of the rotating structure under random excitation. One OMA method is the lifting method that can convert a measured response of the rotating structure into responses on multiple virtual measurement points on the scan path. Correlation functions among converted responses can be calculated and used for determining modal parameters of the rotating structure. The other OMA method is the improved demodulation method that filters a measured response by a band-pass filter that only contains one damped natural frequency of the rotating structure. The filtered response is multiplied by a sinusoidal signal whose frequency is the damped natural frequency and a low-pass filter is applied to the resulting response to obtain undamped mode shapes of the rotating structure. A novel demodulation method with a reference signal is developed to identify positions of damages in a beam without its base-line information. A cross-correlation function between a measurement of a CSLDV and a measurement of a single-point laser Doppler vibrometer is calculated and processed by the demodulation method to estimate its undamped mode shapes. Curvatures of estimated undamped mode shapes are calculated to obtain curvature damage indices and determine the position of a damage in the beam. Prototypes of an image-based short-range tracking CSLDV and an image-based long-range tracking CSLDV are developed for experimental validation of the proposed methods. Rotation speeds and modal parameters of rotating blades with different speeds that are excited by air flow that is considered as random excitation are estimated. Both undamped mode shapes of rotating blades on a straight scan path and their full-field undamped mode shapes are estimated and compared with each other. The proposed methods can address major challenges to monitor the vibration of a rotating horizontal-axis wind turbine blade and detect its potential damage in operational conditions.