Application of pattern search in worst-case flutter solution

This chapter examines the application of pattern search in worst-case flutter solution. The aeroelastic model with purely real valued uncertainties is reviewed, and the promising pattern search algorithm is introduced and applied to solve this type of optimization problem by taking the worst- or best-case flutter speed or aeroelastic damping or frequency as the objective function and taking the structured uncertainty operators as design variables. It is a valuable feature of the present approach because such correlation may enhance the confidence on the results especially when there lacks available validation methods. It should be noted that, as the number of design variables increases about five times compared with the last case, the CPU time cost is still within 10s for computing the lower and upper bounds of damping/frequency for each branch at a given airspeed, while the iterations required are typically within 350 and the objective function evaluation count is about 3000.