Abstract
The fault diagnosis of the gearbox is a complex and important work. In this paper, a multilayer gated recurrent unit (MGRU) method is proposed for spur gear fault diagnosis, that is, three-layer gated recurrent unit (GRU). The vibration signals are firstly monitored on the test bench, and then extracted in both time domain and time-frequency domain. Finally, MGRU is used to learn representation and classification. The MGRU can improve the representation of information and identify the features of fault types more precisely with the increasing number of layers. The proposed method was tested by two spur gears with 10 state modes. To evaluate the method's classification accuracy, four methods were utilized for comparison, i.e., the GRU, long short-term memory (LSTM), multilayer LSTM (MLSTM), and support vector machine (SVM), respectively. In addition, the separability and robustness analysis are also discussed for the proposed MGRU performance. All of the results exhibited that the proposed MGRU approach is effective for spur gear fault diagnosis.
Highlights
With the improvement of automation in modern production, the demand for equipment is increasing sharply, and the maintenance of equipment is paid more attention [1]
In order to test the performance of the designed multilayer gated recurrent unit (MGRU) spur gear fault diagnosis, three capabilities of accuracy, separability and robustness are studied respectively
The MGRU model is improved by 3.72% on the basis of gated recurrent unit (GRU) and multilayer LSTM (MLSTM) improved by 0.46% on the basis of long short-term memory (LSTM)
Summary
With the improvement of automation in modern production, the demand for equipment is increasing sharply, and the maintenance of equipment is paid more attention [1]. As a branch of machine learning model, deep learning has great advantages in this respect [18], [19] It mainly extracts hierarchical representation from input data by establishing a depth neural network of multi-layer non-linear transformation. For fault diagnosis, extracting common time, frequency and time-frequency domain analysis features and using them as input of the model can achieve better data fusion [29]. Follow these principles, the structural steps of this paper are divided into the following three phases: (1) signal acquisition. 2) For gear fault diagnosis, this paper compared the LSTM, the multilayer long short-term memory (MLSTM), the GRU, the MGRU and the SVM models respectively.
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.
