Title :
A Recurrent Neural Network Based Fault Diagnosis Scheme for a Satellite
Author :
Zhao, Shu Ping ; Khorasani, K.
Author_Institution :
Concordia Univ., Montreal
Abstract :
In this paper, a new fault detection and isolation (FDI) scheme using recurrent adaptive time delay neural networks(ATDNN) is proposed and investigated for satellite´s attitude control subsystem (ACS). Results provided illustrate the excellent properties of our proposed detection scheme (both in identifying fault occurrence and clearance times). The faults considered have occurred in reaction wheels which are commonly used in the ACS as actuators. Simulations for the detection results for multiple faults, simultaneous faults, as well as influences caused by dynamic coupling effects of one axis on others in the ACS have also been provided.
Keywords :
adaptive control; artificial satellites; attitude control; delays; fault diagnosis; neurocontrollers; recurrent neural nets; actuators; adaptive time delay neural networks; dynamic coupling effects; fault detection and isolation; fault diagnosis; recurrent neural network; satellite attitude control subsystem; Actuators; Adaptive control; Delay effects; Fault detection; Fault diagnosis; Neural networks; Programmable control; Recurrent neural networks; Satellites; Wheels;
Conference_Titel :
Industrial Electronics Society, 2007. IECON 2007. 33rd Annual Conference of the IEEE
Conference_Location :
Taipei
Print_ISBN :
1-4244-0783-4
DOI :
10.1109/IECON.2007.4459995
