CUT SPECTRUM PROCESSING - A NOVEL SIGNAL-PROCESSING ALGORITHM FOR ULTRASONIC FLAW DETECTION

Multiple scattering from material microstructure (material noise) is one of the major limitations during ultrasonic nondestructive evaluation of coarse-structured materials. The well known ultrasonic signal processing technique referred to as split spectrum processing (SSP) has proven successful in suppressing material noise. However, the performance of conventional SSP algorithms depends strongly on the choice of frequency range utilized during expansion of the ultrasonic signal into a number of narrow-band split signals. If any of the split signals does not carry coherent target information there is a considerable risk of losing the target. This issue is discussed in detail in the paper and new robust SSP algorithms that reduce the above mentioned risk are proposed. A novel signal processing technique, referred to as cut spectrum processing (CSP), which eliminates the risk of losing the target echoes during expansion, is proposed. The CSP, which performs the expansion by excluding consecutive small parts of the signal spectrum, does not require a priori knowledge of the frequency range of the ultrasonic signal. Experimental results are presented, where SSP and CSP have been applied to ultrasonic signals originating from flaws in stainless steel.