This paper addresses the detection of thickness reduction defects in thermal barrier coating （TBC） systems. It compares eddy current testing responses obtained by pulse-modulation waves at different frequencies with those from corresponding sinusoidal waves via linear regression analysis， demonstrating the advantages of pulse-modulation eddy current testing. A numerical simulation model for coating thinning defects in TBC systems is established. Based on the simulation results， three regression models are developed to fit the experimental data， and these models effectively characterize the relationship between the measured signals and defect size. Probability of detection （POD） analysis is performed on the regression results. The detection threshold is determined using the receiver operating characteristic curve， and the POD curve is constructed. The POD analysis results indicate that a 90% POD corresponds to a defect size of 50.62 μm， with a 95% lower-confidence bound.
