Fault Detection in Rotating Machines Using Convolutional Neural Networks and Vibration Images

• Grant number: 24/00737-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: July 01, 2024
• End date: June 30, 2025
• Field of knowledge: Engineering - Mechanical Engineering - Mechanics of Solids
• Principal Investigator: Tiago Henrique Machado
• Grantee: João Pedro Daltro Santos
• Host Institution: Faculdade de Engenharia Mecânica (FEM). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

In the context of industrial operations, it is crucial to ensure the continuous availability of machines and equipment, avoiding unplanned interruptions. To achieve this objective, advanced fault prediction and identification techniques have proven to be indispensable. By adopting such approaches, it is possible to increase the reliability of systems and minimize the need for corrective interventions. This aspect becomes even more relevant when it comes to rotating machines, widely present in different industrial sectors. In this sense, exploring fault prediction and detection methods becomes a fundamental strategy for optimizing operational performance and ensuring operational efficiency. In this context, the objective of the present work is to investigate machine learning methods, in particular neural networks, for application in detecting faults in different types of vibratory behavior of rotating systems. For this, a database containing several signals captured from accelerometers in real physical rotating systems will be used, together with the application of the Vibration Imaging technique for data pre-processing.