Rule-based decision-making unit for eigenstructure assignment via parallel genetic algorithm and LQR designs

This paper proposes a decision-making unit rule-based genetic algorithm for eigenstructure assignment via LQR designs. The proposed decision-making framework is formulated in terms of schema theorem, multi-armed bandit problem and rule-based knowledge; the purpose of these three elements is to guide, in an expert manner, the parallel genetic search to find feedback controllers families that perform the specified eigenstructure assignment. The design techniques are based on multiobjective optimization and independent-runs parallel genetic algorithm. An aircraft state space model is used to study the strategy performance.