Ultrasonic non-destructive evaluation (NDE) is employed extensively across many industries to ensure the integrity of safety critical infrastructures. Where the structure is formed of coarse-grained materials, such as austenitic steel and Inconel, ultrasonic NDE presents a significant challenge arising from their heterogeneous structures and elastic anisotropy. The Thesis addresses two longstanding problems encountered in ultrasonic NDE of coarse-grained materials: phase aberration and backscattering noise.Phase aberration denotes that the wavefronts from elements of a phased array undergo phases shifts. A phase aberration correction approach based on microstructural characterisation and finite element modelling is evaluated in the third chapter. The validation of an emerging microstructural characterisation technique is presented. This embodies two approaches to simplifying measured crystallographic orientation data to construct finite element models, a reduction of computational overhead by 20 times is achieved whilst maintaining model fidelity.The split-spectrum processing (SSP) technique has been widely used to suppress backscattering noise by employing a bank of bandpass filters followed by a combination operator. However, conventional combination algorithms are either ineffective or sensitive to the variations of material characteristics. The use of two artificial neural network (ANN) techniques and the best linear unbiased estimator (BLUE), as the combination algorithms of SSP is investigated in the fourth and fifth chapters, in order to improve its robustness and performance. The performance of two ANN techniques in terms of effectiveness in improving SNR and computational efficiency are compared to instruct the selection between the two techniques in various cases. The BLUE algorithm can improve image contrast by an average of 80% for combining three sub-band images.Another algorithm based on the statistical analysis of frequency components is also proposed in the sixth chapter to reduce speckle level. This algorithm is observed to reduce speckle level by an average of 15 dB.
|Date of Award||19 May 2016|
- University Of Strathclyde
|Sponsors||EPSRC (Engineering and Physical Sciences Research Council) & University of Strathclyde|
|Supervisor||Richard O'Leary (Supervisor) & Anthony Gachagan (Supervisor)|