Search

A significant challenge in non-destructive evaluation is the ability to discern signals, originating from a crack and a geometric feature in a part, that are either closely spaced or superimposed in time. An example problem is the ultrasonic inspection of aircraft holes in vertical riser structures with limited accessibility for a transducer from an external wing surface. Characteristics of the path of the ultrasonic signal and the contact condition between the fastener and hole can hinder the use of traditional amplitude dynamics for signal classification and crack detection. A local correlation method was developed to detect the relative shift of signals in time for adjacent transducer locations due to the varying echo dynamics from crack and part geometries. This approach analyses a series of signals from a moving transducer by first accurately aligning the signals to the primary part signal feature and subsequently measuring the shifting of secondary signals of interest within multiple time windows. A methodology is proposed that supplements amplitude dynamic measures with both the change in time and signal variation measures. Feature maps of these dynamic measures are presented to demonstrate the methodology and provide a practical means for classification. Experimental and simulated studies are presented that validate this methodology for several ultrasonic NDE applications. Lastly, an alternative time delay estimation approach is presented for the local correlation method to improve computation time with comparable performance. [ABSTRACT FROM AUTHOR]