粗粒材料における欠陥特性評価向上のためのマルチステージチャネル最適化による適応型TFMイメージング

Defect detection and characterization are among the primary objectives of ultrasonic non-destructive testing (NDT). However, due to ultrasonic attenuation and scattering noise caused by interactions of the ultrasonic waves with the grains, the inspection of small subwavelength cracks and steeply inclined cracks in polycrystalline materials remains highly challenging. This work proposes an adaptive total focusing method (TFM) that significantly enhances the signal-to-noise ratio (SNR) and improves the robustness of defect characterization. The proposed approach first localizes potential defects using an optimized baseline subtraction scheme, followed by a hierarchical channel optimization process - termed Multi-Step Channel Optimization TFM (MSCO-TFM) - conducted sequentially over receivers, transmitters, and individual A-scan channels. Both numerical simulations and experimental evaluations, based on the SNR metric and the 6-dB sizing method, demonstrate the superior detection and characterization performance of MSCO-TFM compared to existing techniques such as the receiver-optimized TFM (ROTFM). Experimental results on five representative defects (one subwavelength crack and four steeply inclined cracks) show that MSCO-TFM yields an average SNR improvement of 9.78 dB over conventional TFM and achieves excellent characterization accuracy for a 5 mm, 45° crack. More importantly, unlike inversion-based or deep learning approaches, which rely heavily on high-quality training data and prior knowledge of the inspection environment, MSCO-TFM directly optimizes the full matrix capture dataset acquired by the ultrasonic array. Furthermore, it can effectively compensate for spatiotemporal misalignments between defect and baseline data, without increasing computational complexity or complicating experimental procedures, making it highly suitable for practical engineering applications.