Search

The automatic identification of thermography defects is one of important targets of infrared nondestructive testing technology. The analysis of thermal images based on the principle of Hough transform not only can reduce the workload of identification through human beings' eyes, but also can improve the accuracy of defect identification and improve the work efficiency. The paper introduces the basic principle of Hough transform, realizes the testing for the linear defects in images through MATLAB simulation experiment, and describes the testing of other geometric figures under Hough transfer.

The rotor blade is an important component in wind turbine generator, it consists of glass fiber reinforced plastics (GFRP) and sandwich areas containing wood or plastic foam, different defects may be generated because of manufacturing problem. To improve the safety of wind turbine blades, pulsed thermography is being investigated in this study. Surface defects, such as air bubbles, edge bonding, etc. were easily and clearly detected on several full scale blades. Several specimens were manufactured to simulate the adhesive conditions between supporting spar and GFRP shells. The direct analysis of the original, 1stor 2ndderivatives was normally applied to judge any possible defects in the measured sample, and also used in quantitatively calculation. This study compared the relative thermal contrast (RTC) of the original and 1stderivative for air bubbles at shallow depth and deficiency adhesive bonding at deep depth, the calculation of the RTC could give us the approximate best time to analysis defects of different depth and types. The comparison also indicated that it is better to use the 1stderivative than the original to analysis the deep defects, however, there is no significant advantage in the analysis of shallow defects between the original and 1stderivative.