Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
CONVOLUTIONAL NEURAL NETWORKS FOR PATTERN-BASED FAULT DIAGNOSIS IN LOW-ROTATION EQUIPMENT
Submission Author:
Elton Prestes de Souza , PA
Co-Authors:
Elton Prestes de Souza , Lis Silva de Moura, Thiago Barroso Costa, João Lucas Lobato Soares, Walter dos Santos Sousa, Alexandre Mesquita, André Luiz Amarante Mesquita, Danilo Braga
Presenter: João Lucas Lobato Soares
doi://10.26678/ABCM.COBEM2023.COB2023-2299
Abstract
Continuous monitoring of critical industrial equipment is crucial to proactively prevent unexpected failures, minimizing downtime and production delays. Vibration analysis is a valuable diagnostic tool for effective predictive maintenance planning. Machine learning techniques have successfully monitored operational conditions and diagnosed failures in industrial components. However, identifying failures in equipment operating at low rotations is challenging due to low energy levels in vibration signals and interference from external noises. This study proposes an effective fault diagnosis system for equipment operating at low rotations using a Convolutional Neural Network (CNN). The methodology applies Wavelet Transform to obtain two-dimensional images from vibration signals and classifies the bearing's state. The neural network, comprising four convolutional layers, pooling layers, and a dense layer, extracts relevant features for fault detection. Training the network with signals from faulty and healthy bearings distinguishes patterns associated with different conditions. The proposed methodology achieves 96% accuracy, demonstrating its high precision in identifying changes in vibration patterns. These results highlight the efficacy of the proposed artificial intelligence-based approach in identifying faults in equipment operating at low rotations.
Keywords
Convolutional neural networks, machine learning, Vibration Analysis, Low-Rotation, Wavelet Transform, Predictive maintenance
