DocumentCode :
653054
Title :
Identification of a solid-core magnetic bearing using incommensurate fractional-order models
Author :
Jianpeng Zhong ; Lichuan Li
Author_Institution :
Sch. of Electr. Eng., Xian Jiaotong Univ., Xian, China
fYear :
2013
fDate :
25-27 Sept. 2013
Firstpage :
262
Lastpage :
267
Abstract :
Due to the effects of eddy current, solid-core magnetic bearings (MBs) show fractional-order characteristics. In this paper, a practical strategy for incommensurate fractional-order system identification is adopted to model a solid-core MB. The model structures are firstly determined by selecting the number of terms of the numerator and denominator in sequence, and then the derivative orders are derived based on a passive search within a given range. By using a weighed least squares method, the coefficients of derivative operators are calculated. Besides, a modified error criterion is proposed by combining the mean square error with a maximal phase error. The results show that the new error criterion is better for deriving a model with smaller phase errors, and the identified fractional-order models are more precise and simpler than the integer-order models.
Keywords :
eddy currents; least mean squares methods; magnetic bearings; search problems; MB; denominator; derivative orders; eddy current; incommensurate fractional-order models; integer-order models; maximal phase error; mean square error; model structures; numerator; passive search; solid-core magnetic bearing identification; weighed least squares method; Accuracy; Data models; Equations; Integrated circuit modeling; Mathematical model; Solid modeling; System identification; error criterion; fractional-order systems; least squares methods; magnetic bearings; system identification;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Advanced Mechatronic Systems (ICAMechS), 2013 International Conference on
Conference_Location :
Luoyang
Print_ISBN :
978-1-4799-2518-6
Type :
conf
DOI :
10.1109/ICAMechS.2013.6681791
Filename :
6681791
Link To Document :
بازگشت