Abstract
To enhance the performance of deep auto-encoder (AE) under complex working conditions, a novel deep auto-encoder network method for rolling bearing fault diagnosis is proposed in this paper. First, multiscale analysis is adopted to extract the multiscale features from the raw vibration signals of rolling bearing. Second, the sparse penalty term and contractive penalty term are used simultaneously to regularize the loss function of auto-encoder to enhance the feature learning ability of networks. Finally, the cuckoo search algorithm (CS) is used to find the optimal hyperparameters automatically. The proposed method is applied to the experimental data analysis. The results indicate that the proposed method could more effectively distinguish fault categories and severities of rolling bearings under different working conditions than other methods.
Highlights
Rolling bearings are widely used in rotary machines, which play an important role in determining the running states of equipment
It is indicated that the sparse penalty term and the contractive penalty term are applied to the loss function of the stacked AE at the same time, which can obtain more satisfactory diagnosis results
A novel deep AE network is proposed for rolling bearing fault diagnosis
Summary
Rolling bearings are widely used in rotary machines, which play an important role in determining the running states of equipment. Erefore, it has important practical significance for the state monitoring and fault diagnosis of rolling bearings [1]. The vibration analysis method has been widely used in fault diagnosis of rolling bearings, which generally contains three steps: (1) feature extraction, (2) feature selection, and (3) pattern recognition [2]. For (1), the effectiveness of feature extraction is related to the accuracy of fault diagnosis. E common feature extraction method mainly includes time-domain methods, frequency-domain methods, and time-frequency methods. The most popular methods are the statistical analysis [3], such as root mean square, root amplitude, and maximum peak value. For (3), the selected features are classified by Bayesian classification [10], support vector machines (SVM) [11], artificial neural network (ANN) [12], and other machine learning algorithms
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
