Rotating machinery fault diagnosis using similarity-based models

Abstract

This work proposes an automatic fault classifier that uses similarity-based modeling (SBM) to identify faults on rotating machines. The similarity model can be used either as an auxiliary model to generate features for a classifier or as a standalone classifier. A new approach for training the model using a prototype-selection method is investigated. Experimental results are shown for the MaFaulDa database and for the Case Western Reserve University (CWRU) bearing database. Results indicate that the proposed modifications improve the generalization power of the similarity model and of the associated classifier, achieving accuracies of 96.4% on the MaFaulDa and 98.7% on the CWRU databases.

A Parallel Ensemble Learning Model for Fault Detection and Diagnosis of Industrial Machinery

“…These signals are transformed into a set of features that are relevant for FDD. Specifically, statistical features are extracted from the time and spectral domains of each signal [34]. The extracted features form a new representation that reduces the dimensionality of the original features as well as reduces the computational costs.…”

“…Comparatively, EMDL-FDD yields an accuracy of 98.45%, outperforming all other techniques. Table 9(b) presents a comparison of EMDL-FDD with six state-of-the-art methods on the MAFaulD data set, namely Fourier domain features and ANN (FANN) [44], feature vector (kurtosis and entropy) with ANN (KE-ANN) [45], similarity-based models (SBM) [34], synthetic minority oversampling and DNN (SMOTE-DNN) [46], improved SBM (ISBM) [47], and PdM-CNN [9]. Again, EMDL-FDD yields the highest accuracy rate of 99.79%, outperforming the best compared result of 99.58% from PdM-CNN.…”

Processamento de Sinais de Vibração aplicado à Classificação de Falhas em Rolamentos

“…Pesquisas que utilizaram métodos de similaridades foram realizadas por [Ribeiro et al 2017]. A utilizac ¸ão do modelo de similaridade auxilia na gerac ¸ão de descritores para um classificador e pode ser utilizada também como um classificador independente.…”

Diagnóstico Inteligente de Falhas em Equipamentos Dinâmicos Através da Transformada Wavelet e Dados Estatístic

“…Pesquisas que utilizaram métodos de similaridades foram realizadas por Ribeiro et al (2017). Nessa técnica, a utilização do modelo de similaridade auxilia na geração de descritores para um classificador e pode ser utilizada também como um classificador independente.…”