序列近似优化方法

随着工程优化中仿真模型精度和计算时间的不断提高，常规的智能优化方法难以在可接受的计算代价中得到最优解。序列近似优化方法通过将近似模型技术引入优化过程，并采用不断更新采样点的方法来指导寻优，在基于计算耗时模型的优化中得到了越来越广泛的应用。通过论述序列近似优化方法的若干关键技术及其发展现状，可有效指导其在工程优化中的应用，并给出了序列近似优化方法可能的改进方法及发展趋势。
As more and more high precise time-consuming models are revealed into optimization procedure, general intelligence optimization algorithm cannot get a desirable result in feasible computing cost. The developing sequential approximate optimization approach is aimed to overcome this drawback by introducing the approximate model in the optimization procedure, which adds infill points sequentially to search the promising areas, has made it more and more practical technique for time-consuming engineering optimization. An overview of the sequential approximate optimization algorithm and its core techniques were given, which can expend the application of the algorithm. Finally, some relevant improved methods and new developmental trends concerning SAO(sequential approximate optimization) were presented.