Akademik Veri Yönetim Sistemi
19th International Logistics and Supply Chain Congress, Gaziantep, Türkiye, 21 - 22 Ekim 2021, ss.359-364
Despite the long history of the research on inventory optimization, the field is still attracting a significant amount of academic interest. This is partially because that lot-sizing decisions relate with an enormous number of worldwide logistics activities, and partially because the uncertain nature and complex components of many problems in the field, which remain to be solved. Moreover, exact solution approaches developed for many problems require long computation times, high-quality computing equipment and/or commercial software. As a result, the needs of decision makers in practice for decision support tools that provide close-to-optimal results using a feasible amount of computation time/effort motivate many researchers to develop heuristic algorithms with lower computation requirements. With a similar motivation, in this research, we provide an adaptation of Restricted Dynamic Programming, an existing heuristic approach known to reduce the computation and memory requirements of the classical Dynamic Programming method, for solving single product lot-sizing problems with stochastic demand. To the best of our knowledge, this is the first attempt to adapt Restricted Dynamic Programming approach to stochastic lot-sizing problems. The added value of the proposed algorithm is shown through a numerical comparison with the results obtained by the classical Dynamic Programming method.