考虑复杂需求特性的多级煤炭供应链网络优化

由于不同燃煤设备与装置对煤质的要求存在差异，需要将不同类别、品质的煤进行配煤加工，以满足客户差异化需求并减少环境污染。考虑到这一复杂产品需求特性，结合配煤加工与流量平衡等特有约束，建立了新的四级煤炭供应链网络混合整数规划模型，以确定网络中的矿井、物流转运中心与配煤加工中心的数量、位置及规模，并分配各条网络路径上的合理煤炭流量。鉴于问题的NP-hard特性，设计了一种遗传算法，对染色体采用了新的编码结构，并结合贪婪启发式算法生成初始种群，提高了求解效率。大规模实验算例表明，该算法的求解质量明显优于混合遗传算法与模拟退火算法；同时，随着算例规模的增大，与LINGO软件相比，算法在计算时间方面的优势越来越显著。

Optimal design for multi-echelon coal supply chain network with complex demands

Because there are some major differences in coal product’s quality among different coal equipment’s require-ments, it is necessary to carry out coal blending process with different kinds of coal, so as to meet customer individual requirements and reduce the environment pollution. Considering these complex characteristics, coal blending process and flow balance constraints, a new mixed-integer programming model for four-echelon coal supply chain network is estab-lished, in order to decide the number of various facilities, their locations and scales, and the allocation of the correspond-ing logistics flows. Because it is NP-hard in nature, a genetic algorithm with a new encoding structure is designed, and the initial population are created by greedy heuristics to accelerate the convergence of GA. Random instances show that the solution quality of GA is superior to hybrid GA and simulated annealing. Meanwhile, the GA outperforms LINGO soft-ware with respect to computational time when the instances become larger.