Rolling bearing fault convolutional neural network diagnosis method based on casing signal
- PDF / 1,975,496 Bytes
- 10 Pages / 595.22 x 842 pts (A4) Page_size
- 0 Downloads / 241 Views
DOI 10.1007/s12206-020-0506-8
Journal of Mechanical Science and Technology 34 (6) 2020 Original Article DOI 10.1007/s12206-020-0506-8 Keywords: · Aero-engine · Rolling bearing · Casing · Convolutional neural network · Wavelet scale spectrum · Fault diagnosis
Correspondence to: Guo Chen [email protected]
Citation: Zhang, X., Chen, G., Hao, T., He, Z. (2020). Rolling bearing fault convolutional neural network diagnosis method based on casing signal. Journal of Mechanical Science and Technology 34 (6) (2020) ?~?. http://doi.org/10.1007/s12206-020-0506-8
Received June 17th, 2019 Revised
March 25th, 2020
Accepted April 1st, 2020 † Recommended by Editor Chongdu Cho
© The Korean Society of Mechanical Engineers and Springer-Verlag GmbH Germany, part of Springer Nature 2020
Rolling bearing fault convolutional neural network diagnosis method based on casing signal Xiangyang Zhang1, Guo Chen1, Tengfei Hao2 and Zhiyuan He1 1
College of Civil Aviation, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China, School of Automotive and Rail Transit, Nanjing Institute of Technology, School of Automotive and Rail Transit, Nanjing 211167, China 2
Abstract
Affected by the transmission path, it is very difficult to diagnose the vibration signal of the rolling bearing on the aircraft engine casing. A fault diagnosis method based on convolutional neural network is proposed for the weak vibration signal of the casing under the excitation of rolling bearing fault. Firstly, the processing method of vibration signal is studied. Through comparison and analysis, it is found that the fault characteristics of rolling bearing are more easily expressed by continuous wavelet scale spectrum, and a better recognition rate is obtained. Finally, the experiment was carried out with an aero-engine rotor tester with a casing, and the method based on wavelet scale spectrum and convolutional neural network was used for diagnosis. The results were compared with the support vector machine method. The results show that the method has a high recognition rate for the weak fault signals of different fault types collected on the aero engine case, and its fault recognition rate reaches 95.82 %, which verifies the superiority and potential of the method for rolling bearing fault diagnosis.
1. Introduction As a key component in aero-engines, rolling bearing is of a high failure rate due to their high temperature, high speed and large load change range. Once a fault occurs, it will cause abrasion of the rotor, transmission failure, and even cause the engine to stop in the serious case. Therefore, the condition monitoring and fault diagnosis of aircraft engine rolling bearing is of great significance [1, 2]. Machine learning is an effective method for fault diagnosis of rolling bearings. Chen et al. [3] extracted the characteristics of time domain and frequency domain, reduced the feature by principal component analysis (PCA), and then used Gaussian mixture model to diagnose bearing faults. Zhang et al. [4] proposed to use self-organizing neur
Data Loading...
