Optimization of a sensor-fault-detection-filter via genetic algorithms

In this paper the principle of observer-based sensor fault detection and isolation is improved by the use of genetic optimization algorithms. Residual signals are generated by taking linear combinations of the observation errors such that asymptotic decoupling can be achieved. While the residual-generator itself is easy to implement its design in the view of fault-isolation turns out to be a complex problem. It is demonstrated how the observer-eigenstructure can be optimized for transient decoupling of the residuals using genetic optimization algorithms. In order to illustrate its applicability, the method is applied to an industrial turbo-charged combustion engine power plant