Mill Gear Box of Intelligent Diagnosis Based on Support Vector Machine Parameters Optimization

Abstract:

For mill gearbox fault detection problems, and puts forward combining support vector machine (SVM) and genetic algorithm, is applied to rolling mill gear box fault intelligent diagnosis methods. The choice of parameters of support vector machine (SVM) is a very important for the SVM performance evaluation factors. For the selection of structural parameters of support vector machine (SVM) with no theoretical support, select and difficult cases, in order to reduce the SVM in this respect, puts forward the genetic algorithm to optimize parameters, and the algorithm of the model is applied to rolling mill gear box in intelligent diagnosis, using the global searching property of genetic algorithm and support vector machine (SVM) of the optimal parameter values. Results showed that the suitable avoided into local solution optimization, the method to improve the diagnostic accuracy and is a very effective method of parameter optimization, and intelligent diagnosis for rolling mill gear box provides an effective method.

• Periodical: Applied Mechanics and Materials (Volume 697)
• Pages: 239-243
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.697.239
• Online since: November 2014
• Authors: Xiao Hui Liu, Yong Gang Xu, De Ying Guo, Fei Liu
• Keywords: Genetic Algorithm (GA), Intelligent Diagnosis, Mill Gear Box, Parameter Optimize, Parameter Selection, Support Vector Machine (SVM)

References

[1] Wu Lijuan, Zhang Jianyu, Gao Lixin. The intelligent diagnosis system based on neural networks and case-based reasoning [J]. Mechanical design and manufacturing, 2009. 03: 261-263.

[2] Cherkassky V. Ma Y. Practical Selection of SVM Par-ameters and Noise Estimation for SVM regression[J]. Neural Networks(S0893-6080), 2004. 17(1): 113-126.

DOI: 10.1016/s0893-6080(03)00169-2

[3] Deng Naiyang, Tian Yingjie. Support ve-ctor machine (SVM) theory, algorithms and mechanical design and manufacturin[M]. Beijing: Science press. (2009).

[4] Deng Naiyang, Tian Yingjie. The new method, the support vector machine (SVM) in data mining[M]. Beijing: Science press. (2009).

[5] Mangasarian O L. A finite Newton method for classification [J]. Optimization Methods and Software, 2002, 17: 913-929.

DOI: 10.1080/1055678021000028375

[6] Wang Xiaoping, Cao Liming. Genetic al-gorithm(ga)-theory, application and software implementation [M]. Xi an: xi , an jiaotong university press, (2002).

[7] LIU YuHsin. Global maximum likelihood estimation procedure for multinomial probit model parameters[J]. Transportation Resear-ch Part B, 2000, 34(6): 20-23.

DOI: 10.1016/s0191-2615(99)00033-8

[8] Wang Qunjing, Bao Xiaohua, Ni Youyu-an ect. Based on support vector machine (SM) and genetic algorithm of claw pole generator modeling and parameter optimization [J]. Journal of electrical engineering technology, 2006, 21(4): 57-61.

DOI: 10.1109/icems.2005.202912

[9] Wang Chunlin, Zhou hao, Li Guoneng ect. Based on support vector machine and g-enetic algorithm ash melting point predictio-n [J]. Proceedings of the csee, 2007, 27(8): 11-15.

[10] Ren Zhiqiang. Based on support vector machine (SVM) of rolling mill gear box fa-ult intelligent diagnosis technology research [D]. Beijing university of technology, 2010: 10-11.