TWO-STEP METHOD FOR ACTIVE INEQUALITY CONSTRAINTS FOR DIRECT SEARCH OPTIMIZATION

摘要

We demonstrate in this paper that for steady-state optimization involving a large number of inequality constraints a two-step procedure may be a useful approach. In the first step, the active inequalities are identified, and in the subsequent step the active inequality constraints are treated as equality constraints. This approach overcomes the problem of seeking variables on a surface, and also decreases the dimensionality of the optimization problem in the second step. Five test problems show that this approach leads to fast convergence to a very accurate optimal solution.