Authors: Ahmad Hussain Al-Bayati; Hong Wang
Addresses: Computer Science Department, University of Kirkuk, Iraq ' Control Systems Centre, University of Manchester, Manchester, M1 9PL, UK
Abstract: The paper presents the design for new three linear observers based on three optimal theorems to detect and diagnose an actuator fault and disturbance. The diagnosed fault for two observers AOA have been considered as an actuator fault while the fault for the third observer AOAD has been assumed as an additive fault. The gain matrices satisfied the proposed optimal conditions based on Lyapunov functions to diagnose fault and disturbance. The proposed faults and disturbance are white noise, coloured noise and non-Gaussian fault. Conditions for the observers AOA based on Theorem 1 and AOAD have been optimised according to the gain and actuator matrices while the observer AOA based on Theorem 2 has been designed according to the optimisation of the gain, actuator and fault matrices. As the results show, the observer AOA based on Theorem 2 is superior in terms of fault detection and diagnosis FDD when compared with the AOA based on Theorem 1 and AOAD observers. Finally the AOAD observer is the least efficient.
Keywords: fault detection; fault diagnosis; FDD; actuator faults; additive faults; linear observers; white noise; coloured noise; non-Gaussian faults; optimisation.
International Journal of Systems, Control and Communications, 2016 Vol.7 No.1, pp.1 - 15
Received: 06 Dec 2014
Accepted: 26 Feb 2015
Published online: 02 Mar 2016 *