Digital Processing of Ultrasound Signals Backscattered from Coarse Grained Austenitic Stainless Steel.

Available from UMI in association with The British Library. In this thesis, the characteristics of individual pulse-echo ultrasound signals received from a cast block of coarse grained austenitic stainless steel (ASS) are considered. Such signals are renowned for beam skewing effects and being rich in backscattered acoustic noise due to the nature of the material. Cast ASS is used for its mechanical properties in critical applications such as nuclear reactor systems and the acoustic noise is known to be detrimental to the pulse-echo non-destructive testing (NDT) of this form of steel. After a review of these subjects, this thesis describes the application of one-dimensional (1-D) time sequence analysis techniques to the digitally recorded ultrasound data in order to assess their ability to discriminate between data characteristics due to the material and that due to a reflection from the back surface of the block. This is seen as a first step in the evaluation of improving ultrasonic NDT of these materials by use of standard equipment in conjunction with digital signal processing techniques. Two data sets are examined; one from a broad band 5MHz centre frequency compression wave transducer and the other from a similar 1MHz transducer. Numerous 1-D signal processing techniques are applied to the data and their insensitivity to the effect of the material characteristics on the data is noted. Many techniques have not previously been applied to this type of ultrasound data. Cepstrum techniques have shown to be inappropriate when analysing such data and adaptive techniques were generally unsuccessful except for one variant when applied to the 5MHz data. Modern spectrum analysis techniques and a l_4 norm deconvolution technique have been shown to be promising in the analysis of such data but further work in this area is recommended, especially time-space processing in order to improve the statistics of the techniques. Also contained in this thesis is a description of a computer model of a multilayered piezoelectric transducer.