The detection and diagnosis of bearing health status using vibration signal has been an important subject for extensive research over the past few decades. The objective of this paper is to proposed permutation entropy as a tool to select best wavelet for feature selection for the detection as well as fault classification of ball bearings. The continuous wavelet coefficients of the time domain signal are calculated at real, positive scales using various real and complex wavelets. Best wavelet and corresponding scale is selected based on minimum permutation entropy. Eleven statistical parameters were used for defect classification in outer race, inner race, ball defect and healthy bearing respectively. Proposed methodology for fault classification is compared with two artificial intelligence techniques such as artificial neural network and support vector machine. Results revealed that permutation entropy based feature extraction techniques provide higher classification accuracy even when there is a slight variation in operating condition which is useful for development of online fault diagnosis.
In rotating machinery one of the prominent causes of malfunction is faults generated in ball bearings, therefore, diagnosis and interpretation of these faults is essential before they become severe. Feature extraction methodology has been presented in this paper based on application of lifting wavelet transform. Minimum permutation entropy is considered as decision making for selecting level of lifting wavelet transform. Sixteen features are calculated from measured vibration signals for various bearing conditions like defect in inner race, outer race, ball defect, combined defect and no defect condition. To achieve better fault identification accuracy selection of features carrying useful information is needed. To select highly distinguished features various ranking methodologies such as Fisher score, ReliefF, Wilcoxon rank, Gain ratio and Memetic feature selection are used. The ranked feature sets that are fed to machine learning algorithms support vector machine, learning vector quantization and artificial neural network for identification of bearing conditions. Tenfold cross-validation results show that selected features give enhanced accuracy for detecting faults. Features selected through Fisher score-support vector machine and ReliefF-artificial neural network gives 100 % cross-validation accuracy. Result shows that proposed Communicated by V. Loia.
Condition monitoring and diagnosis of a bearing are very important for any rotating machine as it governs the safety while the machine is in operating condition. To construct a feature vector selection of suitable signal processing techniques is a challenge for vibration-based condition monitoring techniques. In the methodology proposed, Short Time Fourier Transform (STFT), Walsh Hadamard Transform (WHT) and Variable Mode Decomposition (VMD) are used to generate 2-D time-frequency spectrograms from the various fault conditions of bearing. When Deep learning techniques apply for fault diagnosis, a large amount of dataset is required for training of machine learning model. To overcome this issue single image Generative Adversarial Network (SinGAN) as a data augmentation technique, utilized for generating additional 2-D time-frequency spectrograms from various fault conditions of ball bearing. To detect fault severity, four deep learning algorithms, ResNet 34, ResNet50, VGG16, and MobileNetV2 are used as a classifier. Experiments are conducted on a rolling bearing dataset provided by the bearing data center of Case Western Reserve University (CWRU) for validating the utility of methodology proposed. Results show that the proposed methodology enables to detect fault severity level with high classification accuracy.
This paper deals with the approach of using multiscale permutation entropy as a tool for feature selection for fault diagnosis in ball bearings. The coefficients obtained from the wavelet transformation of the vibration signals of the bearings are used for the calculation of statistical parameters. Based on the minimum multiscale permutation entropy criteria, the best scale is selected and statistical parameters such as crest factor, form factor, and permutation entropy are calculated. Finally, the faults are classified by considering the statistical parameters and permutation entropy as features in supervised and unsupervised machine learning methods, such as a support vector machine and self-organizing maps, respectively. Results revealed that the multiscale permutation entropy-based feature extraction techniques provide higher classification accuracy in comparison to the other methodologies that have been proposed in previous published works. The methodology proposed in this paper also gives good results for unsupervised learning methods, i.e. self-organizing maps.
Intelligent fault diagnosis gives timely information about the condition of mechanical components. Since rolling element bearings are often used as rotating equipment parts, it is crucial to identify and detect bearing faults. When there are several defects in components or machines, early fault detection becomes necessary to avoid catastrophic failure. This work suggests a novel approach to reliably identifying compound faults in bearings when the availability of experimental data is limited. Vibration signals are recorded from single ball bearings consisting of compound faults, i.e., faults in the inner race, outer race, and rolling elements with a variation in rotational speed. The measured vibration signals are pre-processed using the Hilbert–Huang transform, and, afterward, a Kurtogram is generated. The multiscale-SinGAN model is adapted to generate additional Kurtogram images to effectively train machine-learning models. To identify the relevant features, metaheuristic optimization algorithms such as teaching–learning-based optimization, and Heat Transfer Search are applied to feature vectors. Finally, selected features are fed into three machine-learning models for compound fault identifications. The results demonstrate that extreme learning machines can detect compound faults with 100% Ten-fold cross-validation accuracy. In contrast, the minimum ten-fold cross-validation accuracy of 98.96% is observed with support vector machines.
The identification of surface texture images from machining surfaces using image processing techniques has been a prominent research area in the recent decades. The aim of this paper is to identify various machined surface texture images using machine learning techniques. Charge coupled device is used to capture images of machined components. Based on captured images, twelve statistical features are extracted and feature vector is formed. Grey Level Co-occurrence Matrix is used to extract statistical features from the machined surface images. Four Machine learning algorithms such as Random Forest, Support Vector Machine, Artificial Neural Network and J48 were utilized to characterize machined surfaces. Training and Tenfold cross validation process is utilized for identification of machined component images. It is found that Artificial Neural Network and Random forest give100 % training accuracy and 99% cross validation accuracy. Results obtained demonstrate the efficiency of proposed methodology, which is useful for identifying texture images.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.