Banca de DEFESA: TIAGO NASCIMENTO DE FREITAS

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : TIAGO NASCIMENTO DE FREITAS
DATE: 07/06/2023
TIME: 14:00
LOCAL: UFABC - remoto
TITLE:

Methodology for predictive maintenance of commercial vehicle turbochargers using bidirectional recurrent neural networks

PAGES: 93
BIG AREA: Outra
AREA: Multidisciplinar
SUMMARY:

New technologies and sustainable transport solutions are no longer a differential, but rather the foundation of the mission and values of large companies that seek to remain in the market with profitability, sustainability, and social responsibility. Offering personalized service solutions to customers through a predictive maintenance system, increasing vehicle uptime, the remaining useful life and reducing operational costs make companies more competitive and attractive, while ensuring optimal vehicle performance. This research proposes a methodology for integrity identification and maintenance prediction in commercial vehicle turbochargers using a Bi-Directional Long Short-Term Memory (BLSTM) recurrent neural network, comparing its performance with Random Forest (RF) and Long Short-Term Memory (LSTM) models. The methodology utilizes real industry data obtained from workshop history and operational data records, employing data mining techniques for detailed analysis of variables, dataset composition, and predictive model construction. Experimental results showed that the proposed predictive models demonstrated superior performance compared to current maintenance strategies in predicting maintenance needs, with the BLSTM model standing out with the best overall performance and lowest specific cost-score. McNemar's test indicated that the models have significantly different performances in the classification task, emphasizing the importance of appropriate model selection to ensure good performance in predictive maintenance of turbochargers. The proposed methodology has potential application in other commercial vehicle databases and components, contributing to cost reduction in maintenance and increased operational reliability of vehicles, with significant implications in the automotive industry in terms of cost savings and operational efficiency.

COMMITTEE MEMBERS:
Presidente - Interno ao Programa - 1603909 - RICARDO GASPAR
Membro Titular - Examinador(a) Interno ao Programa - 884.028.433-87 - CALEBE PAIVA GOMES DE SOUZA - UFPI
Membro Titular - Examinador(a) Externo ao Programa - 2078059 - LUIZ ANTONIO CELIBERTO JUNIOR
Membro Suplente - Examinador(a) Interno ao Programa - 1914234 - ALEXANDRE ACACIO DE ANDRADE
Membro Suplente - Examinador(a) Externo ao Programa - 1604330 - ANDRE FENILI
Membro Suplente - Examinador(a) Externo ao Programa - 3042266 - WALLACE GUSMAO FERREIRA