Implementation of machine learning strategies in resonant ultrasound spectroscopy

The elastic reaction of a material is dictated by its tensor of elastic constants. These elastic constants are a measurement of the material's interatomic forces, i.e., they are defined as second derivatives of the free energy with respect to strain in different symmetries, which makes them a useful tool to study the atomic environment of a crystal lattice. Experimental techniques such as resonant ultrasound spectroscopy (RUS) have proven to be very valuable in the determination of the elastic moduli of a material. RUS uses the resonance modes of elastic bodies to infer different material properties such as their elastic moduli. This experimental technique is divided into two different procedures known as the forward problem, which deals with the determination of the resonance modes from the elastic moduli of the body being studied, and the inverse problem which determines the elastic moduli from the experimental measurements of the resonance modes. In this project, a new approach to the inverse problem in RUS is presented through the use of regression trees in the context of machine learning. The results obtained show that the implementation of machine learning strategies in RUS can reduce the variance of the results achieved (when compared to traditional solutions) and can eliminate the need to have initial guesses for the elastic moduli when solving the inverse problem.

• Autores: Giraldo Grueso, Felipe
• Tipo de recurso: Trabajo de grado de pregrado
• Fecha de publicación: 2021
• Institución: Universidad de los Andes
• Repositorio: Séneca: repositorio Uniandes
• Idioma: eng
• OAI Identifier: oai:repositorio.uniandes.edu.co:1992/53548
• Acceso en línea: http://hdl.handle.net/1992/53548
• Palabra clave: Espectroscopía de resonancia ultrasónica; Cuerpos deformables (Física); Aprendizaje automático (Inteligencia artificial); Física
• Rights: openAccess
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/