Ultrasonic measurement of interface location between two polycrystalline microstructures using maximum-likelihood estimation

Abstract

This paper presents the implementation of maximum-likelihood estimation methods for determining the interface location between two polycrystalline microstructures by analysing the statistical properties of ultrasonic backscattering signal amplitudes. Both Rayleigh-based and Chi-squared-based statistical models are employed to formulate the maximum-likelihood estimates. Statistical performance is evaluated using simulated ultrasonic datasets generated from modelled microstructures with varying mean grain sizes, comparing the proposed methods' effectiveness. The sensitivity of the measurements and the impact of inconsistent coupling are analysed using simulated datasets based on standard normal distributions. Experimental validation is performed on an additively manufactured two-layer steel specimen. Results demonstrate that the probabilistic framework effectively models complex wave scattering phenomena from polycrystalline microstructures, making the approach particularly suitable for identifying interfaces between them.