Strategies for Fault Estimation in Actuators and Sensors in Non-Linear Processes with Uncertainties.
Faul-tolerant control, fault estimation, unknown input observer, Linear Matrix Inequalities (LMI).
Over the years, control processes have become more complex, containing a large number
of components that work in an integrated manner. Any of these components is subject
to defects or malfunctions. All these factors are defined as faults, which are unexpected
variations of the properties of a given component to its nominal operating condition. The
consequences of failures may cause economic losses and risk the life of the workers in
the enclosure. A fault tolerant control system is able to keep the control process running
with proper performance even in the presence of failures. In its active approach, the nominal
control strategy is reconfigured so that the effect of the fault is accommodated. This
reconfiguration is usually based on the estimate of the fault, which is obtained by means
of an observer. Generally the effectiveness of an observer is related to the degree of knowledge
about the process by the designer. An unforeseen change in system parameters
or the presence of uncertainties may adversely affect the performance of the observer.
This work proposes two state observer design techniques which are able to perform the
simultaneous estimation of states and faults in actuators and sensors in nonlinear systems
in discrete time with uncertainties. The operation of this method is verified by means of
computational simulations based on case studies involving crane, liquid level processes
and control of a flexible articulated robotic manipulator.