Method to detect the bolt thread defect in the aerospace vehicle with ultrasonic image and its implementation

Abstract

Purpose This study aims to solve the problems of ambiguous localization, large calculation, poor real-time and limited applicability of bolt thread defect detection. Design/methodology/approach First, the acquired ultrasound image is used to acquire the larger area of the image, which is set as the compliant threaded area. Second, based on the determined coordinates of the center point in each selected region, the set of coordinates on the left and right sides of the bolts is acquired by DBSCAN method with parameters eps and MinPts, which is determined by data set dimension D and the k-distance curve. Finally, the defect detection boundary line fitting is completed using the acquired coordinate set, and the relationship between the distance from each detection point to the curve and d, which is obtained from the measurement of the standard bolt sample with known thread defect, is used to locate the bolt thread defect simultaneously. Findings In this paper, the bolt thread defect detection method with ultrasonic image is proposed; meanwhile, the ultrasonic image acquisition system is designed to complete the real-time localization of bolt thread defects. Originality/value The detection results show that the method can effectively detect bolt thread defects and locate the bolt thread defect location with wide applicability, small calculation and good real-time performance.

Automatic Identification of Corrosion in Marine Vessels Using Decision-Tree Imaging Hierarchies

“…The most widespread method for protecting against oxidation is that of cathodic protection with sacrificial electrodes and antifouling pigment paint. Corrosion areas have to be measured using a combination of ultrasound instrumentation (e.g., ultrasound images [1]) and visual RGB images, operated by class surveyors/human operators. Even if protective coatings are used, all merchant marine vessels must complete an inspection of the hull in a dry dock at least three times in a five-year period, with intermediate surveys being performed within 36 months.…”

“…The automation of such a laborious measurement and visual inspection task would save not only person hours but also the time required for a vessel to remain in dry dock. As already mentioned, dry-dock inspection is performed by means of both ultrasonic [1,3] and optical (RGB) imaging. The problem of corrosion surveying via visual inspection in the optical space (RGB images) and its automation can be seen as an extension of semantic segmentation in images and the association of pixels with a specific class/label (corrosion) as a means to recognize (interpret) specific items in a visual scene (vessel hull).…”

“…For a binary decision tree, this means that the average of square distances of all data points (image pixel quantisation levels) from a hyperplane h n sequentially generates leaf nodes of new clusters C 2n , C 2n+1 [31]. The parameters of the new nodes C 2n , C 2n+1 in the n-th leaf with hyperplane h n and param-eters vector w n can be estimated by finding the minimum and maximum values, as per Equation (1). The optimisation problem of Equation ( 1) can be solved using the generalised eigenvalue problem [33].…”