Adaptive filtering and Fractional Fourier transform for ultrasonic signal processsing and flaw detection

In this investigation, an adaptive filter algorithm using Fractional Fourier transform (FrFT) has been developed for enhancement of targets in microstructure noise. In particular, the FrFT is introduced as a transformation tool to prepare ultrasonic signals for a more compact support. It also reduces the non-stationarity of signals. Furthermore, a normalized least mean square (NLMS) filter is applied on the transformed signal to suppress the grain noise. In order to evaluate the performance of the algorithm, various experimental ultrasonic data sets are acquired from a large grained specimen with embedded defects. A 5-MHz broadband transducer is utilized for data acquisition. The proposed algorithm is applied to discriminate defects in presence of high grain scattering noise. The experimental study shows that the algorithm greatly reduces grain noise and improves the signal-to-grain-noise ratio by about 12 dB. With the help of FrFT and the NLMS filter, the algorithm successfully identifies defects embedded in grain noise. This type of study can broaden the scope of applying adaptive filter in material evaluation and target detection.