Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
DECISION TREE FOR FEATURE SELECTION TO DIAGNOSIS OF BELT CONVEYOR IDLER
Submission Author:
João Lucas Lobato Soares , PA , Brazil
Co-Authors:
João Lucas Lobato Soares, Thiago Barroso Costa, Lis Silva de Moura, Walter dos Santos Sousa, Alexandre Mesquita, André Luiz Amarante Mesquita, Danilo Braga
Presenter: João Lucas Lobato Soares
doi://10.26678/ABCM.COBEM2023.COB2023-2261
Abstract
Belt conveyors, as an essential equipment in the mining industry, requires constant monitoring to maintain good reliability. In order to support the belt and the transported material, idlers are components that, constantly, fail during operation, in which they present bearing defects as the most common failure modes. These defects can propagate more severe failures on the conveyor such as wear or tear on the belt, if idler becomes stuck. Thus, monitoring based on predictive maintenance is essential, and machine learning techniques can be used as an alternative for detecting equipment failures. In diagnostics using machine learning, the feature selection step is important to avoid overfitting and loss of accuracy in the classification of the equipment's condition. The present study analyzes the performance of decision tree algorithm as alternative method for dimensionality reduction. Initially, vibration signals were collected on the belt conveyor bench idler and the Wavelet Packet Decomposition (WPD) was applied to the signals to obtain the energy bands, which were used as features for classification. After the features were determined, three methods were analyzed for feature selection: application of no dimensionality reduction method, application of traditional Principal Component Analysis (PCA), and application of decision tree. Additionally, different classification algorithms were employed: Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN) and Artificial Neural Network (ANN). As results, it was found a superior performance of diagnosis accuracy in all techniques with a dimensionality reduction of features selected by decision tree. Moreover, SVM, KNN and ANN presented accuracy above 98%, indicating high efficiency of the proposed decision tree application as a feature dimensionality reduction for classification.
Keywords
machine learning, Fault Diagnosis, Feature Selection, Decision tree, Belt conveyor idler
