Faults diagnosis of stochastic dynamic systems based on neural network probability density function estimation

Author

Grishin, Yuri ; Konopko, Krzysztof

Author_Institution

Fac. of Electr. Eng., Bialystok Tech. Univ., Poland

Volume

1

fYear

2004

Firstpage

335

Abstract

The paper presents a fault diagnosis algorithm based on multidimensional probability density function (pdf) estimation which is suitable for stochastic nonlinear systems. The pdf of symptom vector is estimated with use of the Radial-Basis Function (RBF) and Hyperradial-Basis Function (HRBF) artificial neural networks (NN). The numerical example of diagnosis of a nonlinear system is presented. The influences of the NN parameters and learning on the algorithm performance are discussed.

Keywords

control system synthesis; fault diagnosis; nonlinear dynamical systems; probability; radial basis function networks; stochastic systems; artificial neural networks; fault diagnosis algorithm; hyperradial-basis function; multidimensional probability density function estimation; neural network; nonlinear system diagnosis; radial-basis function; stochastic dynamic systems; symptom vector; Artificial neural networks; Data mining; Fault detection; Fault diagnosis; Multidimensional systems; Neural networks; Neurons; Nonlinear systems; Probability density function; Stochastic systems;

fLanguage

English

Publisher

ieee

Conference_Titel

Electrotechnical Conference, 2004. MELECON 2004. Proceedings of the 12th IEEE Mediterranean

Print_ISBN

0-7803-8271-4

Type

conf

DOI

10.1109/MELCON.2004.1346863

Filename

1346863