非线性动态网络/系统（N/S）模型辨识与故障诊断

journal6 ›› 2004, Vol. 25 ›› Issue (4): 4-9.

非线性动态网络/系统（N/S）模型辨识与故障诊断

(1．湖南大学电气与信息工程学院自动化系，湖南，长沙，410082；2.中南大学信息科学与工程学院，湖南，长沙，410083）

出版日期:2004-12-15 发布日期:2012-09-22

Model Identification and Fault Diagnosis for Nonlinear Dynamic Networks /Systems (N/S)

(1.College of Electical and Information Engineering,Hunan University,Changsha 410082，Hunan China；2.School of Information Science & Engineering,Central South University,Changsha 410083,Hunan China)

Online:2004-12-15 Published:2012-09-22
About author:XIE hong(1964-),male,associate professor,doctor,place of birth:Changsha,hunan;main research fields:theory of nonlinear network and system,fault diagnosis of N/S.
Supported by:
Research supported by National Natural Science Foundation of China,GRANT(50277010,59707002)

摘要/Abstract

摘要：采用方波脉冲函数变换（BPFT）对一类非线性动态N/S混合模型(H1,B)进行了辨识，导出了计算混合模型（H1,B）的相关公式和N/S响应的伏特劳级数解在方波域内的离散递推算式，解决了一类非线性动态N/S模型的数值计算问题.在此基础上，提出了一种基于多重预置模型的非线性N/S的故障诊断方法，该法通过检验各个预设模型与N/S当前状态的匹配程度来判断N/S是否处于某种故障状态，而无须在线估算N/S当前的模型及分析其特征，从而极大地减轻了在线计算工作量，可实现在线故障诊断.给出了故障诊断实例，实验结果表明该法故障诊断的准确率达到80％～90％.

关键词： 方波脉冲函数变换, 模型辨识, 递推计算, 多重预置模型, 故障诊断, 伏特劳级数

Abstract: By applying block-pulse function transform (BPFT) the hybrid model (H1,B) for a class of nonlinear dynamic N/S was identified in the paper.The relative formulas for calculating the hybrid model (H1,B) and discrete recursive formulas for computing Volterra series resolution of model of N/S were derived.Based on above a fault diagnosis method of multiple preset models for N/S was also proposed.This method can judge and classify the faults occurring in N/S by checking the matching degree between each preset model and the current working state of N/S.There is no need for computing the current model of N/S and analyzing the features on line.So the method can largely reduce the computational cost and the fault diagnosis can be realized on line.An example of fault diagnosis was given here.The results show that the accuracy of fault diagnosis reaches 80-90 percent by this means.

Key words: Block-pulse function transform, model identification, recursive calculation, multiple preset models, fault diagnosis, Volterra series

谢宏, 彭卫韶, 何怡刚. 非线性动态网络/系统（N/S）模型辨识与故障诊断[J]. journal6, 2004, 25(4): 4-9.

XIE Hong, PENG Wei-Shao, HE Yi-Gang. Model Identification and Fault Diagnosis for Nonlinear Dynamic Networks /Systems (N/S)[J]. journal6, 2004, 25(4): 4-9.

参考文献

[1] XU Nan-rong.Introduction of System Identification [M].Beijing：Press of Electronic Industry，1990 (in Chinese).

[2] CHANG Cun-qian et al.System Identification and Parameter Estimation [M].Beijing：Press of Machine Industry，1999 (in Chinese).

[3] FANG Chong-zhi et al.Process Identification[M].Beijing：Press of Qinghua University，1998(in Chinese)

[4] ZENG Guan-da.Block-pluse Functions Transform With Application [J].Numerical Maths and Applications,1989,(1):68-81.

[5] ZENG Guan-da.The Recursive Analysis of Volterra Series Solutions for Nonlinear Networks [J].Journal of Wuhan University (Natural Science Edition),1991,(1):50.

[6] CATELANI M，GORI M.On the Application of Neural Networks to Fault Diagnosis of Electronic Analog Circuts[J].Measurement,2001,17:73-80.

[7] BANDLER J W,SALAMA A E.Fault Diagnosis of Analogue Circuits[A].Proc.IEEE on circuits and Systems[C].Florida.U.S.Printed in Florida，2003,72(8):1 279-1 327.

[8] LIN Gui-nian.Nonlinear Networks and Systems[M].Beijing：Press of China Railway,1989:101-104(in Chinese).

[9] BILLINGS S A,TSANG K M.Spectral Analysis for Nonlinear Systems,Part II:Interpretation of Nonlinear Frequency Response Function[J].Mechanical Systems and Signal Processing,2001,3(4):5-16.

[10] BILLINGS S A,TSANG K M.Spectral Analysis for Nonlinear Systems,Part III:case study examples[J].Mechanical Systems and Signal Processing,2001,3(1):3-20.

[11] ZENG Guang-da，et al.Recursive Computation of Transfer Matrix for A Class of linear Systems[J].Journal of Hua-zhong Institute,1985,13(2):23.

[12] WEI Rui-xuan.Fault Diagnosis Method Based on Multiple Preset Models of GFRF[J].Journal of Xi’an Traffic University,2002,36(8):810.

[13] HAN C Z,WANG L Q,TANG X Q，et al.Identification of Nonparametric GFRF Model for A Class of Nonlinear Dynamic Systems[J].Control Theory and Applications,1999,16(6):816-825.

[14] XIE Hong，et al.A method of Deriving Fault Equation for Nonlinear Circuits[A].Proceedings SCI/Volume X VII 2002 International Conference on Information Systems,Analysis and Synthesis [C].New York U.S.Printed in New York.December 2,2002.

[15] XIE Hong，et al.Research of Approach for Fault Diagnosis of Nonlinear Resistive Network Based on BPNN[J].Journal of Hunan University,Natural Science,2002,29(6):87-90.(in Chinese)