Genetic robustification of digital model-following flight-control systems

Abstract

The dynamical characteristics of high-performance fighter aircraft vary considerably over typical operational flight envelopes, and digital model-following flight-control systems for such aircraft must accordingly be robust. However, the linearised dynamics of such aircraft are usually modelled as partially irregular plants, i.e. plants having rank-defective first Markov parameters for which no robustness results are currently available in the case of fast-sampling digital PID controllers. Genetic algorithms can, however, be used to tune such PID controllers so as to robustify them over entire flight envelopes. This use of genetic algorithms is illustrated by the design of a robustified digital model-following flight-control system for an open-loop unstable F-16 aircraft operating over a flight envelope involving three different flight conditions.<>