Multi-Gaussian Window-Based Blind Deconvolution Method for Separating Terahertz Overlapped Echoes of Thermal Barrier Coatings

Accurate thickness measurement of thermal barrier coatings (TBCs) is critical for reliable and safe operation of aero-engines. However, the seriously overlapped terahertz (THz) signals of TBCs with unrecognizable echoes resulted in unsatisfactory accuracy of thickness measurement. To address this issue, the multi-Gaussian window-based blind deconvolution method (MGWBDM) is proposed to fully separate the overlapped THz echoes for accurate thickness measurement. First, the characteristic of THz echo is explored to form the template of the window. Second, the multi-Gaussian window is presented, and the similarity with the THz echo is improved for reconstructing the high-fidelity signal. Then, constraints are proposed to fine-tune the window for eliminating the partially overlapped phenomenon of THz echoes. After that, the multi-Gaussian window-based deconvolution filter is designed to fully separate the overlapped THz echoes, and the thickness of TBCs is calculated by the time-of-flight (TOF) principle. Finally, simulation and experiments were carried out. The results show that the MGWBDM has the fully separating ability for overlapped THz echoes and achieves satisfactory accuracy for thickness measurement.