Multiobjective Optimization Using Adjoint Gradient Enhanced Approximation Models for Genetic Algorithms

摘要: In this work, a multiobjective design optimization framework is developed by combining GAs and an approximation technique called Kriging method which can produce fairly accurate global approximations to the actual space provide function evaluations efficiently. It applied wing planform problem its results demonstrate efficiency applicability of proposed framework. Furthermore, improve propsed using adjoint gradients two different approaches are tested. The show that gradient efficiently replace computationally expensive sample data needed for constructing models, gradient-based techniques be utilized refine candidates obtained through model based genetic algorithms.