Quantitative imaging using ultrasonic tomography based on full waveform inversion

• Main Author: Rao, Jing
• Other Authors: Fan Zheng, David
• Format: Theses and Dissertations
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Mechanical engineering::Mechanics and dynamics
• Online Access: http://hdl.handle.net/10356/75928
• Institution: Nanyang Technological University

Quantitative imaging can provide quantitative estimation of physical properties such as velocity, density and attenuation. The ability to reconstruct physical properties from wave measurements is very valuable in many applications as it could allow diagnostic methods with high sensitivity. For example, numerous defects are reflected in acoustic velocity variations, and thus velocity reconstruction from wave measurements can be used for defect detection. Based on this concept, ultrasonic tomography provides a promising alternative for defects/inclusions diagnosis. The key thing in ultrasonic tomography is the applicability and usefulness of accurate and high resolution algorithms for reconstructions of physical properties. In this thesis, a quantitative ultrasonic imaging approach based on full waveform inversion (FWI) is developed for accurate reconstructions of physical properties in isotropic structures. The forward model is computed in the frequency domain by solving a full-wave equation in a two dimensional (2D) acoustic model, accounting for higher order effects such as diffractions and multiple scattering. The inversion is based on local optimization of a waveform misfit function between modeled and measured data, and is applied iteratively to discrete frequency components from low to high frequencies. The mono-parameter FWI combined with ultrasonic guided waves is used for the application of corrosion mapping. In guided wave tomography (GWT), the resulting wave velocity maps can be converted into thickness maps by the dispersion characteristics of selected guided modes. The results suggest that the FWI algorithm is capable of reconstructing the thickness map of an irregularly shaped defect accurately on a 10 mm thick plate with the thickness error within 1 mm. This thesis discusses the reconstruction accuracy of FWI on plate-like structures by using simulations as well as experiments. FWI can obtain a resolution around 1.5-2 wavelengths from the realistic data compared to 0.7 wavelengths from the idealized case. The resolution loss with the realistic test data indicates the acoustic model cannot accurately capture the scattering of guided waves and the elastic inversion approach can achieve resolution improvements. The thesis also investigates the performance of FWI with limited view configurations, and applies the regularization technique performed by an adaptive threshold approach to synthesize the missing components, improving the image quality. Then, this thesis extends the application of GWT to liquid-loading plates. A significant challenge is the energy of guided waves leaking into the liquid. By considering attenuation effects through complex velocity in frequency-domain FWI, the validation experiment performed on a water-loaded plate with an irregularly shaped defect shows excellent performance of the reconstruction algorithm. A new GWT system based on customized piezoelectric transducers is also presented for corrosion monitoring of plate-like and pipe structures. The efficiency and the accuracy of this system have been demonstrated through continuous forced corrosion experiments. Finally, the multi-parameter FWI is applied to ultrasonic tomography for the application of density mapping, which aims to simultaneously reconstruct density and velocity by using ultrasonic bulk waves. Density is very difficult to reconstruct in multi-parameter FWI due to the trade-off effects between velocity and density, and the difference in amplitude of different parameters in the wave-field. The inverse Hessian has been shown to be able to mitigate these effects and rescale the amplitudes. The results show that good velocity reconstruction is achieved and well-resolved density reconstruction is also obtained.