Fault-Tolerant Observers
Fault-Tolerant
Observers
Supervisor: Dr Imad M. Jaimoukha
Designing of a robust fault-tolerant observer that guarantees a certain level of performance for real-time applications.
By: YEO Zhi-Wei, Laurence 00566245
Overview
For any system such as a factory plant, an aircraft jet engine system or a water pumping system, it can be replicatedi by a mathematical model which includes the affecting dynamics based on observations and assumptions. However, it would be practically challenging to find an accurate model to describe the system entirely and adding more dynamic parameters will further complicate the simulation, especially when a system structure or initial state is not fully known in the first instance. Uncertainty also worsens when unexpected disturbances are introduced into the system; thus distorting the output to vary differently from the expected actual signal.
Additional consideration to take note of is the presence of faults occurring at the output of the actual system. This could be due to the failure of the system’s sensors or actuators and the number of outputs will yield different combination of faults which may possibly affect the tracking efficiency of the observing system. In order to mitigate these modelling errors and to combat perturbations, an observer is introduced to improve system performance.
An observer is imperative in any system, which can be mathematically modelled, in order to reduce the effects of uncertainties so as to achieve less tracking error. However, faults are out of the system designer’s control and therefore, the measured system requires an observer, L that is robust enough to meet certain level of reliable performance by measuring its maximum fault detectability and the ability to decouple disturbances from the system’s output. Significant hurdle arises when the number of measured outputs of the system for the observer to measure becomes vast; thus increasing the processing time exponentially
Observers
Supervisor: Dr Imad M. Jaimoukha
Designing of a robust fault-tolerant observer that guarantees a certain level of performance for real-time applications.
By: YEO Zhi-Wei, Laurence 00566245
Overview
For any system such as a factory plant, an aircraft jet engine system or a water pumping system, it can be replicatedi by a mathematical model which includes the affecting dynamics based on observations and assumptions. However, it would be practically challenging to find an accurate model to describe the system entirely and adding more dynamic parameters will further complicate the simulation, especially when a system structure or initial state is not fully known in the first instance. Uncertainty also worsens when unexpected disturbances are introduced into the system; thus distorting the output to vary differently from the expected actual signal.
Additional consideration to take note of is the presence of faults occurring at the output of the actual system. This could be due to the failure of the system’s sensors or actuators and the number of outputs will yield different combination of faults which may possibly affect the tracking efficiency of the observing system. In order to mitigate these modelling errors and to combat perturbations, an observer is introduced to improve system performance.
An observer is imperative in any system, which can be mathematically modelled, in order to reduce the effects of uncertainties so as to achieve less tracking error. However, faults are out of the system designer’s control and therefore, the measured system requires an observer, L that is robust enough to meet certain level of reliable performance by measuring its maximum fault detectability and the ability to decouple disturbances from the system’s output. Significant hurdle arises when the number of measured outputs of the system for the observer to measure becomes vast; thus increasing the processing time exponentially