Efficient progressive polynomial approximate optimization

Abstract

This study proposes efficient progressive polynomial approximate optimization (ePPAO) which is a new function-based sequential approximate optimization algorithm. ePPAO finds an improved design than an initial design in a short amount of time and at low-cost. The proposed method adopts a screening method in every iteration to improve efficiency for large-scale problems. A new move limit strategy as a region management technique was developed based on the concept of the trust region method. To evaluate the proposed method, we solved well-known unconstrained and constrained problems and compared the results with two different algorithms from commercial software similar to the ePPAO concepts.. In addition, the proposed method is applied to an industrial design problem.