嵌套分区算法框架下基于序的优化方法研究

为有效解决随机资源分配问题,提出了一种嵌套分区算法框架下基于序的优化方法.该方法将序优化与最优计算量分配技术融入嵌套分区算法框架,利用"序比较"思想进行算法的局部寻优,极大地降低了算法的计算负担,而最优计算量分配技术则能够智能地对有限的计算量进行合理的分配,进一步提高序优化的收敛速度及结果的可靠性.嵌套分区方法保证了每一步均对全体可行域进行采样,从而保证了算法的全局收敛性.给出了算法实施的具体步骤并证明了收敛性.用该算法解决标准作业车间调度问题,并将仿真结果与其他算法进行比较,证明了本文算法的收敛速度与优化质量均优于其他算法.

Optimization method based on order under framework of nested partitions algorithm

To solve the stochastic resource allocation problem effectively,an optimization method based on order under the framework of Nested Partitions(NP) algorithm was proposed.This method combined ordinal optimization and Optimal Computing Budget Allocation(OCBA) technique with NP framework.Ordinal comparison method was used to greatly reduce computation budget.And OCBA technique could improve the convergence rate and reliability of result further by allocating the computing budget intelligently.However,the NP method guaranteed sampling from entire feasible region in each iteration and thereby guaranteed the global convergence of hybrid algorithm.Detailed operation steps were given out and convergent result was proved.Job shop scheduling benchmark problems were used to verify this algorithm and the result of simulation,which was compared to other well-known algorithms,indicated that this algorithm was better than several other algorithms both in convergence rate and quality of optimization.