Fault Classification of Low-Speed Bearings Based on Support Vector Machine for Regression and Genetic Algorithms Using Acoustic Emission
- 1 Downloads
This work under consideration makes use of support vector machines (SVM) for regression and genetic algorithms (GA) which may be referred to as SVMGA, to classify faults in low-speed bearings over a specified speed range, with sinusoidal loads applied to the bearing along the radial and axial directions.
GA is used as a heuristic tool in finding profound solution to the difficult problem of solving the highly non-linear situation through the application of the principles of evolution by optimizing the statistical features selected for the SVM for regression training solution. It is used to determine the training parameters of SVM for regression which can optimize the model and hence without the forehand knowledge of the probabilistic distribution can form new features from the original dataset. Using SVM for regression, the non-linear regression and fault recognition are achieved. Classification is performed for three classes. In this work, the GA is used to first optimize the statistical features for the best performance before they are used to train the SVM for regression. Experimental studies using acoustic emission caused by bearing faults showed that SVMGA with a Gaussian kernel function better achieves classification on the bearings operated at low speed, regardless of the load type and, under different fault conditions, compared to the exponential kernel function and the other many kernel functions which also can be used for the same conditions.
This study accomplished the effective classification of different bearing fault patterns especially at low speeds and at varying load conditions using support vector machines (SVM) for regression and genetic algorithms (GA) referred to as SVMGA.
KeywordsAcoustic emission Artificial intelligence (AI) Artificial neural networks (ANN) Exponential kernel function Gaussian kernel function Rolling element bearing Support vector machines (SVM) Genetic algorithm (GA)
We profusely thank the staff of the C-AIM laboratory of the University of Pretoria, for their co-operation and help in seeing that a worthwhile experiment was conducted with successful recording of the acoustic signal which was used for analysis purpose in this work. We also thank our supervisor and colleagues for their support in the course of carrying out the experiment and also other well wishers who rendered their help to successfully carry out the experiment.
- 1.Hamadache M, Lee D (2014) Improving signal-to-noise ratio (SNR) for inchoate fault detection based on principal component analysis (PCA). In: 14th international conference on control, automation and systems (ICCAS 2014), Oct. 22–25, in KINTEX, Gyeonggi-do, KoreaGoogle Scholar
- 5.Yu Y, Dejie Y, Junsheng C (2006) A fault diagnosis approach for roller bearing based on IMF envelope spectrum and SVM. Elsevier (ScienceDirect) Meas 40:943–950Google Scholar
- 7.Kan X (2011) ECE 480 team 4, support vector machine, concept and MATLAB buildGoogle Scholar
- 8.Li X, Chen W (2014) Rolling bearing fault diagnosis based on physical model and one-class support vector machine, Hindawi Publishing Corporation. ISRN Mechanical Engineering 2014:160281Google Scholar
- 9.Sloukia FE, Bouarfa R, Medromi H, Wahbi M (2013) Bearings prognostic using mixture of Gaussian hidden Markov model and support vector machine. Int J Netw Secur Appl (IJNSA) 5(3):85–97Google Scholar
- 11.Muhammet U, Mustafa D, Mustafa O, Haluk K (2013) Fault diagnosis of rolling bearing based on feature extraction and neural network algorithm, recent advances in telecommunications, signals and systems. ISBN: 978-1-61804-169-2. www.wseas.us/e-library/confrences/2013/lemesos. Accessed 2017
- 13.Hagan MT, Demuth HB, Beale M (1997) Neural network design. Thomson, BostonGoogle Scholar
- 14.Shuang L, Fujin Y, Jing L (2007) Bearing fault diagnosis based on k-l transform and support vector machine. In: Third international conference on natural computation (ICNC 2007) 0-7695-2875-9/07Google Scholar
- 16.Rao SS (2011) Mechanical vibrations. In: Fifth Edition in SI Units, University of Miami. Prentice Hall. Published in 2011 by Pearson Education South Asia Pte Ltd. 23/25 First Lok Yang Road, Jurong Singapore 629733Google Scholar
- 17.Gun SR (1998) Support vector machines for classification and regression, technical report, Faculty of Engineering, Science and Mathematics, School of Electronics and Computer Science, University of SouthamptonGoogle Scholar
- 18.SmolHamadache AJ, Lee MD (2014) Improving signal-to-noise ratio (SNR) for inchoate fault detection based on principal component analysis (PCA). In: 14th international conference on control, automation and systems (ICCAS 2014). KINTEX, Gyeonggi-do, Korea, pp 561–566Google Scholar
- 21.Kim Y, Tan ACC, Mathew J, Yang B (2006) Condition monitoring of low speed bearings: a comparative study of the ultrasound technique versus vibration measurements. WCEAM Paper 029:1–10Google Scholar
- 23.Lu D, Qiao W (2014) A GA-SVM hybrid classifier for multiclass fault identification of drivetrain greaboxes. In: Energy conversion congress and exposition (ECCE), IEEE pp 3894–3900Google Scholar
- 24.Tong Q, Han B, Lin Y, Zhang W, Cao J, Zhang X (2017) A fault feature detection approach for fault diagnosis of rolling element bearings based on redundant second generation wavelet packet transform and local characteristic-scale decomposition. J Vib Eng Technol 5(1):101–110Google Scholar
- 25.Fatima S, Mohanty AR, Kazmi HF (2016) Fault classification and detection in a rotor bearing rig. J Vib Eng Technol 4(6):491–498Google Scholar