A Pattern Search Algorithm for Feedback-Control System Parameter Optimization

The problem of parameter optimization for the purpose of synthesis and design of feedback control systems is considered. A flexible performance index, which permits direct consideration of performance specifications such as rise time, settling time, peak overshoot, and bandwidth, in addition to the usual analytic forms, is developed and adapted to a modified version of Rosenbrock´s pattern -search algorithm. Modifications of Rosenbrock´s method include: 1) scaling of parameters to ensure coordinate orthogonalization; and 2) employing variable step sizes at the beginning of each new stage, the steps being a function of previous failures and successes. The efficiency of each stage is thus improved and the sensitivity to initial estimates of step size is reduced.