论文部分内容阅读
针对DIRECT全局优化方法估值次数多、收敛速度慢的缺点,提出一种基于径向基函数元模型的改进DIRECT算法.通过分析DIRECT算法的采样点信息,从中识别包含局部最优或全局最优点的最优区域.收集最优区域中的采样点构造径向基函数元模型,并在该元模型上搜索全局最优点,从而提高了DIRECT算法的收敛速度.最后,将该方法应用于数值计算以及压力容器的优化设计,结果证明了该方法的实用性与工程有效性.
In view of the shortcomings of the DIRECT global optimization method such as a large number of evaluation times and slow convergence rate, a modified DIRECT algorithm based on the radial basis function meta-model is proposed. By analyzing the sampling point information of DIRECT algorithm and identifying the local optimal or global optimal point The best region of the DIRECT algorithm is constructed.The sampling points in the optimal region are collected to construct the radial basis function meta-model and search the global optimal point on the meta-model to improve the convergence speed of DIRECT algorithm.Finally, the method is applied to numerical calculation As well as the optimization design of pressure vessel, the result proves the practicability and engineering effectiveness of this method.