Abstract:
This paper presents a study on applying machine learning (ML) techniques to analyze a dataset comprising anonymized technical inspections of approximately 4,000 rainwater galleries within the drainage systems alongside Brazilian highways. Periodic inspections to evaluate structural safety while also identifying any existing anomalies for subsequent activities such as instrument monitoring, maintenance, and structural reinforcement are recommended by Brazilian technical standards. This research leverages ML algorithms to complement expert judgment, focusing on identifying and prioritizing rainwater galleries in critical conditions requiring urgent maintenance. Selecting features for ML was demonstrated to be essential for achieving accurate and reliable classification and identifying the gallery’s critical condition results. Efficiency in applying ML algorithms is achieved by the rapid and effective identification of galleries requiring intervention with more priority after visual inspection and by directing inspection or training efforts toward these significant features. This targeted strategy allows for prioritizing the most urgent cases, minimizing the risk of severe incidents due to delayed or inadequate maintenance.
Referência:
SILVA, Cristina Maria Ferreira da; LEAL, Adriano Galindo(DR13) MORAES, Luciana Andrea Mori Faria de;(DR13) SILLES, Felipe Silva(DR13); SILVA, Celso Luciano Alves da(DR13); GAVA, Vagner Luiz(DR13). Application of machine learning techniques for identifying emergency conditions in Rainwater galleries. In: IBERO-LATIN AMERICAN CONGRESS ON COMPUTATIONAL METHODS IN ENGINEERING, ABMEC, 44., 2023, Porto. Portugal. Proceedings… 7 p.
Documento com acesso restrito. Logar na BiblioInfo, Biblioteca GITEB/IPT para acessar o trabalho em PDF:
