Fault Diagnosis of Bearing Based on Conjugate Gradient BP Algorithm

Abstract:

Largely used in industry field, bearing is one of the most vulnerable components in an equipment. Owing to the complicated and nonlinear relationship between features and corresponding specific fault, it is less efficient to diagnosis the faults in tradition ways ,especially to deal with the fault of a mega machine. BP neural network whose strength is to solve the nonlinear problems makes it more precise and efficient to determine the fault of bearing. Conjugate gradient algorithm is proposed as the training method of the BP neural network. Compared with standard training method, conjugate gradient has the advantage of training speed and generalization ability, which is confirmed by the results of neural network model whose inputs are common bearing fault samples.

• Periodical: Advanced Materials Research (Volume 1039)
• Pages: 191-196
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.1039.191
• Online since: October 2014
• Authors: Li Xin Lu*, Yan Zhao, Gui Qin Li, Zheng Li, Xiao Yuan, Hong Bo Li
• Keywords: Bearing Fault Diagnosis, BP Neural Network (BPNN), Conjugate Gradient Algorithm

References

[1] L. M. R. BACCARINI, et al. SVM practical industrial application for mechanical faults diagnostic,J. Expert Systems with Applications, 2011. 38(6): 6980-6984.

DOI: 10.1016/j.eswa.2010.12.017

[2] P. JAYASWAL, S. VERMA,A. WADHWANI. Development of EBP-Artificial neural network expert system for rolling element bearing fault diagnosis,J. Journal of Vibration and Control, 2011. 17(8): 1131-1148.

DOI: 10.1177/1077546310361858

[3] T. -C. LIN. Analog circuit fault diagnosis under parameter variations based on type-2 fuzzy logic systems,J. International Journal of Innovative Computing, Information and Control, 2010. 6(5): 2137-2158.

[4] L. MAO, L. XIAOQUAN,S. YANG. Research on faults diagnosis of distribution network based on adaptive immune genetic algorithm,J. Application of Electronic Technique, 2012. 8: 031.

[5] Z. Z. LIU, et al. Measure and analysis about vibration characteristic of roller bearing fault,J. Machinery Design ＆ Manufacure, 2009(3): 103-105.

[6] Y. P. CAI, et al. Roller bearing fault detection using improved envelope spectrum analysis based on EMD and spectrum kurtosis J. Journal of Vibration and Shock, 2011. 30(2): 167-172.

[7] D. W. PATTERSON, Artificial neural networks: theory and applications. 1998: Prentice Hall PTR.

[8] T. JAYALAKSHMI,A. SANTHAKUMARAN. Statistical normalization and back propagation for classification,J. International Journal of Computer Theory and Engineering, 2011. 3(1): 1793-8201.

DOI: 10.7763/ijcte.2011.v3.288

[9] H. AZAMI, S. SANEI,K. MOHAMMADI. Improving the Neural Network Training for Face Recognition using Adaptive Learning Rate, Resilient Back Propagation and Conjugate Gradient Algorithm,J. International Journal of Computer Applications, 2011. 34.

[10] J. WANG, W. WU,J. M. ZURADA. Deterministic convergence of conjugate gradient method for feedforward neural networks,J. Neurocomputing, 2011. 74(14): 2368-2376.

DOI: 10.1016/j.neucom.2011.03.016

[11] X. -S. XING, et al. Research on BP algorithm based on conjugate gradient. in Information Science and Engineering (ICISE), 2010 2nd International Conference on. 2010. IEEE.

[12] R. WAKISAKA, et al. Speech kurtosis estimation from observed noisy signal based on generalized Gaussian distribution prior and additivity of cumulants. in Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on. 2012. IEEE.

DOI: 10.1109/icassp.2012.6288807

[13] T. T. YU,C. SHAO, Fault Diagnosis of Bearing Based on BP Neural Network, 2008, Dalian University of Technology.