Automatic visual defects inspection of wind turbine blades via YOLO-based small object detection approach

Cavities with different geometries represent the internal volumes of various engineering applications such as cabins of passenger cars, fuselages and wings of aircraft, and internal compartments of wind turbine blades. Transmissibility of acoustic excitation to and from these cavities is affected by material and cross-sectional properties of the structural cavity, as well as potential damage incurred. A new structural damage detection methodology that relies on the detectability of the changes in acoustic transmissibility across the boundaries of structural cavities is proposed. The methodology is described with a specific focus on the passive damage detection approach applied to cavity internal acoustic pressure responses under external flow-induced acoustic excitations. The approach is realized through a test plan that considers a wind turbine blade section subject to various damage types, severity levels, and locations, as well as wind speeds tested in a subsonic wind tunnel. A number of statistics-based metrics, including power spectral density estimates, band power differences from a known baseline, and the sum of absolute difference, were used to detect damage. The results obtained from the test campaign indicated that the passive acoustic damage detection approach was able to detect all considered hole-type damages as small as 0.32 cm in diameter and crack-type damages 1.27 cm in length. In general, the ability to distinguish damage from the baseline state improved as the damage increased in severity. Damage type, damage location, and flow speed influenced the ability to detect damage, but were not significant enough to prevent detection. This article serves as an overall proof of concept of the passive-based damage detection approach using flow-induced acoustic excitations on structural cavities of a wind turbine blade. The laboratory-scale results reveal that acoustic-based monitoring has great potential to be used as a new structural health monitoring technique for utility-scale wind turbine blades.

Download Full-text

Multi-View Object Detection Based on Deep Learning

Applied Sciences ◽

10.3390/app8091423 ◽

2018 ◽

Vol 8 (9) ◽

pp. 1423 ◽

Cited By ~ 15

Author(s):

Cong Tang ◽

Yongshun Ling ◽

Xing Yang ◽

Wei Jin ◽

Chao Zheng

Keyword(s):

Deep Learning ◽

Object Detection ◽

Regression Models ◽

Detection Methods ◽

Detection Accuracy ◽

Single Shot ◽

Small Object ◽

Object Retrieval ◽

Detection Approach ◽

Small Object Detection

A multi-view object detection approach based on deep learning is proposed in this paper. Classical object detection methods based on regression models are introduced, and the reasons for their weak ability to detect small objects are analyzed. To improve the performance of these methods, a multi-view object detection approach is proposed, and the model structure and working principles of this approach are explained. Additionally, the object retrieval ability and object detection accuracy of both the multi-view methods and the corresponding classical methods are evaluated and compared based on a test on a small object dataset. The experimental results show that in terms of object retrieval capability, Multi-view YOLO (You Only Look Once: Unified, Real-Time Object Detection), Multi-view YOLOv2 (based on an updated version of YOLO), and Multi-view SSD (Single Shot Multibox Detector) achieve AF (average F-measure) scores that are higher than those of their classical counterparts by 0.177, 0.06, and 0.169, respectively. Moreover, in terms of the detection accuracy, when difficult objects are not included, the mAP (mean average precision) scores of the multi-view methods are higher than those of the classical methods by 14.3%, 7.4%, and 13.1%, respectively. Thus, the validity of the approach proposed in this paper has been verified. In addition, compared with state-of-the-art methods based on region proposals, multi-view detection methods are faster while achieving mAPs that are approximately the same in small object detection.

Download Full-text