NONDESTRUCTIVE DETECTION OF TURBINE BLADE DEFECTS IN AERO-ENGINES USING THERMAL IMAGING TECHNIQUES

Abstract

Accurate nondestructive inspection of aero-engine turbine blades is crucial for maintaining engine stability and safety. This paper briefly overviews thermal imaging technology and turbine blades in jet engines. The thermal imaging technology was applied to the nondestructive inspection of thermal barrier coatings on turbine blades. The Faster-regional convolutional neural network (RCNN) algorithm was employed to detect defects in the thermal images, which were preprocessed using the adaptive carrier algorithm. Then experimental analyses were conducted using prepared thermal barrier coatings with three types of defect. Moreover, the Faster-RCNN algorithm combined with adaptive carrier preprocessing was compared with the convolutional neural network and Faster-RCNN algorithms combined with Gaussian filter preprocessing. The results demonstrated that the adaptive carrier preprocessing combined with Faster-RCNN method accurately identified defect types and located defects with higher precision.