New methodologies and concepts are developed in the control theory to meet the
ever-increasing demands in industrial applications. Fault detection and diagnosis of technical
processes have become important in the course of progressive automation in the operation of
groups of electric drives. When a group of electric drives is under operation, fault tolerant
control becomes complicated. For multiple motors in operation, fault detection and diagnosis
might prove to be difficult. Estimation of all states and parameters of all drives is necessary
to analyze the actuator and sensor faults. To maintain system reliability, detection and
isolation of failures should be performed quickly and accurately, and hardware should be
properly integrated. Luenberger full order observer can be used for estimation of the entire
states in the system for the detection of actuator and sensor failures. Due to the insensitivity
of the Luenberger observer to the system parameter variations, state estimation becomes
inaccurate under the varying parameter conditions of the drives. Consequently, the estimation
performance deteriorates, resulting in ordinary state observers unsuitable for fault detection
technique. Therefore an adaptive observer, which can estimate the system states and
parameter and detect the faults simultaneously, is DESIGNed in our paper. For a Group of DC
drives, there may be parameter variations for some of the drives, and for other drives, there
may not be parameter variations depending on load torque, friction, etc. So, estimation of all
states and parameters of all drives is carried out using an adaptive observer. If there is any
deviation with the estimated values, it is understood that fault has occurred and the nature of
the fault, whether sensor fault or actuator fault, is determined by neural fuzzy network, and
fault tolerant control is reconfigured. Experimental results with neuro fuzzy system using
adaptive observer-based fault tolerant control are good, so as to confirm the best
characteristics of the proposed approach.