Banca de QUALIFICAÇÃO: TIAGO NASCIMENTO DE FREITAS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : TIAGO NASCIMENTO DE FREITAS
DATA : 30/11/2021
HORA: 15:00
LOCAL: UFABC - Sistema Remoto
TÍTULO:

Methodology for predictive maintenance of commercial vehicle turbochargers using bi-directional recurrent neural networks

PÁGINAS: 63
GRANDE ÁREA: Outra
ÁREA: Multidisciplinar
RESUMO:

New technologies and sustainable transport solutions are no longer a differential, but the basis of the mission and values of large companies, which seek to remain in the market with profitability sustainability and social responsibility. Providing a personalized service solution to customers through a predictive maintenance system, increasing vehicle uptime, and reducing operational costs make companies more competitive and attractive in addition to ensuring the best performance of the vehicle. This research aims to propose a methodology for a predictive maintenance system applied to commercial vehicles, identify, and implement current technologies in machine learning and data science to find patterns and correlations in the available data. The proposed prognostic model uses a recurrent neural network called Bi-Directional Long Short-Term Memory (BLSTM) which in comparison with traditional unidirectional models, can predict the current state of the component using at the same time both past and future information to achieve hight accuracy in the failure prognosis. The model is intended to predict the need for repair of commercial vehicle engine turbochargers, assist in decision making, and structuring of vehicle data, enabling more sophisticated and efficient maintenance management while minimizing costs with break downs and the costs with maintenance actions. Prediction models will be developed through the integration of vehicle specification data, embedded data recorded on its current state, data saved when the vehicle visits a workshop, historical data and maintenance records performed for fault identification. The final architecture aims to estimate the degradation levels and the remaining useful life of the turbocharger. The methodology could be applied to other databases and other commercial vehicle components.

MEMBROS DA BANCA:
Presidente - Interno ao Programa - 1603909 - RICARDO GASPAR
Membro Titular - Examinador(a) Interno ao Programa - 2269065 - ROMULO GONCALVES LINS
Membro Titular - Examinador(a) Externo ao Programa - 2078059 - LUIZ ANTONIO CELIBERTO JUNIOR
Membro Suplente - Examinador(a) Externo à Instituição - CALEBE PAIVA GOMES DE SOUZA - UFPI