Engineering-Oriented Ultrasonic Decoding: An End-to-End Deep Learning Framework for Metal Grain Size Distribution Characterization

Highlights: What are the main findings? Multimodal ultrasonic features with time–frequency encoding and an encoder–decoder model, aided by elliptic spatial fusion, enable grain size distribution prediction for GH4099. The method achieves MAEs of 1.08 μm (mean) and 0.84 μm (standard deviation) with a KL divergence of 0.0031, outperforming attenuation- and velocity-based approaches. What are the implications of the main findings? Transfer learning calibration rapidly restores accuracy under new input conditions, improving adaptability for practical ultrasonic inspection. The framework provides a scalable, low-cost path for accurate, cross-scenario grain size characterization in non-destructive evaluation. Grain size is critical for metallic material performance, yet conventional ultrasonic methods rely on strong model assumptions and exhibit limited adaptability. We propose a deep learning architecture that uses multimodal ultrasonic features with spatial coding to predict the grain size distribution of GH4099. A-scan signals from C-scan measurements are converted to time–frequency representations and fed to an encoder–decoder model that combines a dual convolutional compression network with a fully connected decoder. A thickness-encoding branch enables feature decoupling under physical constraints, and an elliptic spatial fusion strategy refines predictions. Experiments show mean and standard deviation MAEs of 1.08 and 0.84 μm, respectively, with a KL divergence of 0.0031, outperforming attenuation- and velocity-based methods. Input-specificity experiments further indicate that transfer learning calibration quickly restores performance under new conditions. These results demonstrate a practical path for integrating deep learning with ultrasonic inspection for accurate, adaptable grain-size characterization.