Comparison and evaluation of different processing algorithms for the nondestructive testing of fiber-reinforced plastics with pulse thermography

Abstract

Abstract When making components out of fiber-reinforced plastics, the composite material itself is formed during the manufacturing of the component, so that suitable methods for process monitoring and quality assurance of each manufactured component regarding its structural integrity are necessary. An alternative approach to the commonly used ultrasonic testing is pulse thermography, which is suitable for the inspection of large areas. But this method is restricted regarding inspection depth, on one side, and thermal contrast and depth resolution with increasing depth, on the other side. In this paper we compare several data processing methods which can be used to overcome the limitations of this method and to optimize the contrast for detection of deeply buried defects. (1) differential absolute contrast, (2) thermographic signal reconstruction, (3) pulse phase thermography, (4) principle component thermography and (5) high order statistics are compared regarding their capability of detecting defects, e. g. delaminations, in different depths. As a measure for the achievable contrast of the different methods designed to detect various defects we compared the respective signal-to-noise-ratio. Additionally, we discuss further approaches for improving the detection capability by applying different evaluation methods in sequence or combining them with image processing methods based on geometric filters. For this purpose a CFRP specimen with 18 artificial defects was examined.