A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features

This paper presents an investigation process in building a bearing fault classifier based on wavelet coefficients and statistical parameter features. The building process starts by processing raw vibration data that was acquired from a bearing test rig. The data acquisition process was carried out f...

Full description

Bibliographic Details
Main Authors: Latuny, Jonny, Entwistle, Rodney
Other Authors: Andrei Kotousov
Format: Conference Paper
Published: Engineers Australia 2012
Subjects:
Online Access:http://hdl.handle.net/20.500.11937/34796
_version_ 1848754320846094336
author Latuny, Jonny
Entwistle, Rodney
author2 Andrei Kotousov
author_facet Andrei Kotousov
Latuny, Jonny
Entwistle, Rodney
author_sort Latuny, Jonny
building Curtin Institutional Repository
collection Online Access
description This paper presents an investigation process in building a bearing fault classifier based on wavelet coefficients and statistical parameter features. The building process starts by processing raw vibration data that was acquired from a bearing test rig. The data acquisition process was carried out for both normal (fault-free) and fault operation of a double row self-aligning ball bearing. Two accelerometers were used to collect the vibration data. One was attached near the bearing under investigation and the other was attached at one of the shaft support bearing of the test rig. The raw data was then processed to extract the statistical parameters (i.e., kurtosis, RMS, variance, standard deviation). Further, the same data was processed using a wavelet transform employing Daubechies wavelet filter to produce wavelet coefficients and their energy levels. The features generated from statistical parameters and wavelet transform scheme were then used to train an Artificial Neuro-Fuzzy Inference System (ANFIS). In order to reduce the number of rules generated during the training process, only two inputs were used for the purpose of building the classifier. The selection of the most influential inputs for the training process of the ANFIS is achieved through the use of the ANFIS built in capability of selecting the best correlation of two inputs towards one target output which best represents the bearing operating condition.The process of selecting the most influential inputs-output combination was carried out using an extensive computation to obtain the best related two inputs, out of the six inputs available. The number of input-output combinations tested was 720, equal to the total of six input permutations. In the search for the best combination of inputs-output, the possible application between the combination of statistical parameters, wavelet coefficients and wavelet’s level of energy were investigated extensively in order to obtain the best classifier for bearing fault diagnosis. The ANFIS was then implemented to capture the input-output relation of the selected inputs to generate a suitable classifier that could be used to classify bearing operating condition. The classifiers generated were then tested to evaluate their ability and accuracy in predicting faulty bearing operating conditions. The result showed that a bearing fault classifier produced by using ANFIS through the proposed combined features of statistical parameters and Daubechies wavelet transform is promising as a bearing fault classifier.
first_indexed 2025-11-14T08:38:33Z
format Conference Paper
id curtin-20.500.11937-34796
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T08:38:33Z
publishDate 2012
publisher Engineers Australia
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-347962017-03-08T13:13:19Z A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features Latuny, Jonny Entwistle, Rodney Andrei Kotousov ANFIS bearing fault classifier diagnosis Daubechies wavelet wavelet energy level statistical parameters This paper presents an investigation process in building a bearing fault classifier based on wavelet coefficients and statistical parameter features. The building process starts by processing raw vibration data that was acquired from a bearing test rig. The data acquisition process was carried out for both normal (fault-free) and fault operation of a double row self-aligning ball bearing. Two accelerometers were used to collect the vibration data. One was attached near the bearing under investigation and the other was attached at one of the shaft support bearing of the test rig. The raw data was then processed to extract the statistical parameters (i.e., kurtosis, RMS, variance, standard deviation). Further, the same data was processed using a wavelet transform employing Daubechies wavelet filter to produce wavelet coefficients and their energy levels. The features generated from statistical parameters and wavelet transform scheme were then used to train an Artificial Neuro-Fuzzy Inference System (ANFIS). In order to reduce the number of rules generated during the training process, only two inputs were used for the purpose of building the classifier. The selection of the most influential inputs for the training process of the ANFIS is achieved through the use of the ANFIS built in capability of selecting the best correlation of two inputs towards one target output which best represents the bearing operating condition.The process of selecting the most influential inputs-output combination was carried out using an extensive computation to obtain the best related two inputs, out of the six inputs available. The number of input-output combinations tested was 720, equal to the total of six input permutations. In the search for the best combination of inputs-output, the possible application between the combination of statistical parameters, wavelet coefficients and wavelet’s level of energy were investigated extensively in order to obtain the best classifier for bearing fault diagnosis. The ANFIS was then implemented to capture the input-output relation of the selected inputs to generate a suitable classifier that could be used to classify bearing operating condition. The classifiers generated were then tested to evaluate their ability and accuracy in predicting faulty bearing operating conditions. The result showed that a bearing fault classifier produced by using ANFIS through the proposed combined features of statistical parameters and Daubechies wavelet transform is promising as a bearing fault classifier. 2012 Conference Paper http://hdl.handle.net/20.500.11937/34796 Engineers Australia restricted
spellingShingle ANFIS
bearing fault classifier
diagnosis
Daubechies wavelet
wavelet energy level
statistical parameters
Latuny, Jonny
Entwistle, Rodney
A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features
title A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features
title_full A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features
title_fullStr A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features
title_full_unstemmed A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features
title_short A bearing fault classifier using Artificial Neuro-Fuzzy Inference System (ANFIS) based on statistical parameters and Daubechies wavelet transform features
title_sort bearing fault classifier using artificial neuro-fuzzy inference system (anfis) based on statistical parameters and daubechies wavelet transform features
topic ANFIS
bearing fault classifier
diagnosis
Daubechies wavelet
wavelet energy level
statistical parameters
url http://hdl.handle.net/20.500.11937/34796