The course compares classical signal processing techniques with emerging artificial intelligence (AI) and deep learning methodologies for flaw detection, material characterization, imaging, and data compression. Through practical case studies, we present established approaches, including chirplet-based signal estimation and order-statistics-based flaw detection for ultrasonic inspection of materials with high microstructural scattering. These physics-based methods are contrasted with modern deep neural network architectures for automated material characterization, defect classification, and decision-making. We also present recent advances in ultrasonic data compression, through-solid data communication, and real-time System-on-Chip (SoC) hardware/software co-design for embedded NDE systems. By integrating signal theory with the ultrasonic physical properties of materials, such as wave propagation, attenuation, and scattering mechanisms, the course provides a framework for selecting and deploying optimal solutions for ultrasonic NDE applications. Experimental results demonstrate how the interpretability and reliability of classical methods can be combined with the powerful pattern-recognition capabilities of AI to advance next-generation automated ultrasonic NDE systems.
