Banca de QUALIFICAÇÃO: DANILSON KAIO DE MACEDO FREITAS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : DANILSON KAIO DE MACEDO FREITAS
DATE: 26/05/2025
TIME: 10:00
LOCAL: On-line
TITLE:

Analyzing Subtle Mechanical Vibrations Through Machine Learning-Enhanced Robotic Vision Techniques

KEY WORDS:

Micromotion Amplification, Computer Vision, Vibration Mechanics, Video Processing

PAGES: 53
BIG AREA: Engenharias
AREA: Engenharia Mecânica
SUBÁREA: Projetos de Máquinas
SPECIALTY: Estática e Dinâmica Aplicada
SUMMARY:

This work bridges techniques from Robotic Vision with principles of vibration mechanics to detect subtle movements originating from equipment operation. In relation to the state-of-the-art, this is an innovative proposal whose subject of study are movements imperceptible to the naked human eye. In this way, a simple video camera, used for data acquisition, functions as an array of millions of vibration sensors distributed across a portion of an industrial plant. This tool, when combined with specialized algorithms, is capable of capturing specific frequencies with satisfactory levels of precision, not just at a single point, but across the entire machine or system, based on variations in the vicinity of unitary elements of the image, known as pixels. The consulted literature in this research area shows that for many years, researchers have been attempting to solve challenges associated with this process, such as artifacts, blurring, noise amplification, and the amplification of unwanted movements, achieving processed videos with significant qualities. An evident extrapolation of such research lies in comparing the current state-of-the-art with conventional vibration analysis techniques. To this end, a setup consisting of a wheel balancing machine and unbalanced wheel-tire assemblies was used. The machine, camera, and algorithms were tested, and the experimental results endorse the significant degree of similarity between the current state-of-the-art and conventional technology, indicating a trend towards supplementation or even replacement by the new methodology in certain cases.

COMMITTEE MEMBERS:
Presidente - 1510735 - DANILO ALVES PINTO NAGEM
Externo ao Programa - 2177445 - BRUNO MOTTA DE CARVALHO - UFRNExterno ao Programa - 1422699 - HERTZ WILTON DE CASTRO LINS - UFRN