In pulse thermography, pulsed flash energy is applied to the surface and the temperature of the surface
is recorded and analysed. Generally the the flash duration is short and the heating could be taken as
impulse function. After the surface is heated instantaneously, heat goes down by conduction. If an area
has defect below, the temprature of this area will be different from the temprature of defect free area.
Analytic solution indicates that the time at which the temperature descending curve of the area with
defect below separate from the curve of the defect free area is proportional to the square of the defect
depth. Thus, if the deviation time is determined, the defect depth could be calculated.
In real inspection, different from theoretical model, the temperature decay curve may be noisy and
sometimes fluctuating. And due to the effect of three-dimentional conduction and different boundary
conditions the temperature decay curve is different from the theoretical solution under ideal conditions.
All these affect the identification of the deviation time and then affect the accurary of the depth
measurement. Peak temperature contrast and peak slope of temperature contrast methods are popularly
used in depth measurement, but all these two methods require the prior determination of a reference
point that is known to be on sound material. Peak second derivative method in log scale is a reference
free method which can somehow decrease the influence of noisy data and three-dimentioanl conduction.
To reduce the noise induced by derivation, fitted data instead of raw data is offten used. However, the
global data fitting is not suitable in some situation. In this paper, peak second derivative method based
on patial data fitting is proposed and results are discussed. The results show that this method could
improve the accuracy of depth measurement for CFRP specimen.
|