铁矿预配料的原料种类繁多、化学成分差异较大,且下料槽个数有限、生产约束多,原料下料次序难以确定.针对该配料调度难题,本文提出了一种基于聚类算法和组合优化的铁矿混匀过程预配料智能调度方法.分别根据原料成分中SiO2,TFe含量的差异,采用Canopy-Kmeans聚类方法进行两次聚类,然后综合考虑各项约束条件,利用融合专家规则的组合优化和小范围穷举思想对聚类结果进行组合与排序,得到原料共槽方案与共槽下料次序,以保证在有限下料槽的情况下配完所有原料,且配得的混匀料化学元素含量始终尽可能稳定.经我国某钢铁厂实际生产数据验证,所提方法与现有人工计算方法相比,大幅缩减了运算时间,且矿物化学元素指标的波动小,具有实用价值.%This paper presents an intelligent scheduling pre-blending approach based on clustering algorithm and com-binatorial optimization in the burdening process of iron ore. The proposed approach is applied to solve the tough problems which result from the finite chutes, many production constrains, the undeterminable sequence of raw materials and the various raw materials composed of quiet different chemical elements. Firstly, the raw materials are clustered preliminar-ily by Canopy-Kmeans clustering method according to the differences in the content of the SiO2 and TFe respectively. Then, considering all the practical constraints, the chutes scheme of raw materials and the sequence of raw materials are obtained by combinatorial optimization combined with experts'rules and small-scale exhaustive algorithm so that all the raw materials can be scheduled within finite chutes and the chemical elements fluctuations of scheduled blending materials can be as smooth as possible. This pre-burdening approach could benefit in reducing the computing time significantly and reducing the fluctuations of chemical elements, which has been proven by being applied in a practical steel plant in China. Additionally, this approach has appreciable practical significance after compared with original artificial calculation method.
展开▼
