论文部分内容阅读
利用集成专家意见的在线序列预测算法研究多产品多阶段报童问题.将任意的一个固定订购量策略看成一个专家意见,用弱集成算法综合考虑众多专家意见构建在线订购策略,并基于弱集成算法的竞争性理论给出在线订购策略的理论保证.首先给出了两产品多阶段报童问题的在线订购策略,证明了其实现的累积收益可与最优固定订购量策略实现的累积收益相当.然后,将两产品多阶段报童问题的在线订购策略及其理论结果推广到多产品多阶段报童问题中.最后在不同的需求类型下,通过数值算例表明构建的在线订购策略相对于最优固定订购量策略具有较强的竞争性能.
This paper studies the problem of multi-product multi-period newsboy by using online sequence prediction algorithm with integrated expert opinion.An arbitrary one fixed order quantity strategy is regarded as an expert opinion, and the weak integration algorithm is used to consider the opinions of many experts to build an online ordering strategy, and based on the weak integration algorithm , This paper gives the theoretical guarantee of online ordering strategy.Firstly, an online ordering strategy of two-product multi-period newsboy problem is given, which proves that the cumulative revenue achieved can be equal to the cumulative revenue achieved by the optimal fixed-volume strategy , The online ordering strategy and its theoretical results of the multi-stage newsboy problem of two products are extended to the multi-product multi-stage newsboy problem.Finally, under different types of demand, numerical examples show that the online ordering strategy is better than the best fixed order Quantitative strategy has a strong competitive performance.