Flight Control Optimization Using Genetic Algorithm (GA) and Adapative Partitioned Random Search (APRS)

The design of flight control systems involves the solution of several difficult optimization problems. Due to nonlinear dynamics, constraints on states and controls and modeling uncertainities, the flight control optimization problem involves local optima and nondifferentiable cost functions. This necessitates the use of global search methods such as Genetic Algorithm (GA) and Adaptive Partitioned Random Search (APRS) methods. We have shown that for a class of signal processing problems involving global maximization of a likelihood function over unkown frequencies, APRS is more efficient than GA. During Phase I, we will apply both GA and APRS to the optimization of gains for the F-18 inner loop command and stability augmentation system. The performance of GA & APRS in terms of solution reliability and function evaluations will be tested on a simulation. Phase II will involve complete optimization and testing of the flight control system using FA and APRS methods.