一类随机规划的蒙特卡罗回溯优化求解方法  

Monte Carlo Based Retrospective Optimization Method to Solve One Class of Stochastic Programming

作　　者：马新顺[1] 石彤菊[1] 

机构地区：[1]华北电力大学数理学院,河北保定071003

出　　处：《河北大学学报（自然科学版）》2008年第6期568-571,共4页Journal of Hebei University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(10471033);华北电力大学博士学位教师科研基金资助项目(200612005)

摘　　要：针对一类随机规划问题构造了基于蒙特卡罗的回溯优化求解法,该方法本质属于一种动态搜索算法,通过迭代求解一系列样本确定性优化问题并经样本容量逐渐增加过程而逼近随机问题的最优解,而迭代终止条件由需求的计算精度确定,并具体给出了近似解的计算方法及迭代终止条件.最后,通过算列验证了该方法的有效性.This paper present an essential dynamic search method named retrospective optimization algorithm based on a sequence of sample path approximation to the original problem with increasing sample size and decreasing the tolerance error. A stopping rule of the algorithm and a calculation of the approximating solution are studied and proposed. Numerical example with an expectation model is employed to demonstrate the efficiency for the presented algorithm.

关 键 词：随机规划 蒙特卡罗模拟 回溯优化法 样本近似方法 

分 类 号：O427.4[理学—声学]