Title :
Model-based detection of actuator faults in the presence of modelling errors
Author :
Juricic, D. ; Zele, M.
Author_Institution :
Dept. of Comput. Autom. & Control, Jozef Stefan Inst., Ljubljana, Slovenia
Abstract :
The problem of detecting actuator faults in the presence of modelling errors is addressed. A likelihood-ratio test is derived such that the effects of undermodelling are properly taken into account via a stochastic embedding technique. The resulting test takes form of the quadratic sum of prediction errors which has to be compared with a particular threshold value. Systematic choice of the threshold amounts to a demanding problem of approximating the distribution of the test statistics with normal distribution. The effectiveness of the test is demonstrated on a simulated motor-generator system.
Keywords :
DC motors; actuators; fault simulation; statistical distributions; stochastic processes; actuator fault detection; likelihood-ratio test; modelling errors; simulated motor-generator system; stochastic embedding technique; test statistics distribution; Actuators; Gaussian distribution; Stochastic processes; Systematics; Model-based fault detection; modelling error; stochastic embedding; test statistics;
Conference_Titel :
Control Conference (ECC), 2001 European
Conference_Location :
Porto
Print_ISBN :
978-3-9524173-6-2
