点击次数：

论文类型：会议论文

发表时间：2015-06-15

收录刊物：EI、CPCI-S、Scopus

卷号：2C

摘要：The sensitivity information of objective and constraint functions is required in gradient-based optimization algorithms. Usually the more accurate the gradients are estimated, the faster the convergence speed will be. Commonly used gradient estimation methods, e.g., the finite differences, either have low estimate accuracy or require much computing budget. This paper presents a metamodel-based gradient estimation (MGE) method. In our numerical investigations, the proposed MGE method achieves more accurate estimates of the sensitivity information than the forward difference method (FDM) with only one more point. This article explores the sensitivity of the proposed method to the sampling strategy, the metamodel type, the finite difference step on the metamodel and the dimensionality of the design space, and offers some recommendations. Finally, we combine the new MGE method with the method of moving asymptotes (MMA) and apply them to the shape design of a turbine disc and an axial compressor blade. The results from the two engineering examples confirm the utility of proposed method.