A two-step signal denoising method and feature extraction for grain size assessment using ultrasonic waves

Ultrasonic testing offers an efficient and non-destructive method for assessing grain size in polycrystalline metals. This study investigates the feasibility of using ultrasonic trailing wave attenuation to assess grain size in thick plates with high frequency-thickness products. A novel two-step signal denoising method, incorporating cross-correlation operation and variational mode decomposition, is proposed to process complex trailing wave signals. The average power envelope is used to extract the attenuation coefficients from the band-limited signals. Experiments are conducted on GH4742 nickel-based superalloy specimens. The attenuation coefficients derived from the collected trailing wave signals are used to establish a grain size assessment model. The effect of grain size distribution, morphology, and large grains are discussed via electron backscatter diffraction measurements. Furthermore, two case studies on specimens with varying geometries and inhomogeneous microstructures are conducted. The predicted grain sizes exhibit a relative error of less than 10 %, confirming the validity and robustness of the proposed method.