摘要
面向订单生产的钢铁企业实际生产过程中,在订单量、设备生产能力不匹配的约束条件下,加热、退火等工序会出现混合加工情况,如何合理计算混合加工时产品的资源消耗量是提高产品成本准确性的关键。本文提出了基于粗糙集和BP神经网络相结合的混合加工资源消耗计算方法,通过粗糙集提取了资源消耗的必要影响因素,确保资源分配对象划分准确,利用BP神经网络建立影响因素与资源消耗的分配关系,通过上述关系计算混合加工时的资源消耗量。最后通过实例,应用5-fold交叉验证方法,提高了资源分配模型的泛化能力,并与回归分析法对比,验证了该方法具有较高的精度和收敛速度,为混合加工作业在成本动因很难制定情况下,准确计算资源消耗提供了方法支持。
In actual make-to-order production of iron and steel enterprises, when order quantity can' t match capacity of equipment, mixed-processing will take place on activities of heating, annealing, etc. How to reasonably calculate products' resource consumption on mixed-processing activities is the key to improving the accuracy of product' s cost. The method of calculating resource consumption at mixed-processing activities which is based on the rough set and BP neural network is proposed. In order to divide resource allocation objects accurately, the necessary influencing factors of resource consumption are extracted by means of the rough set. The allocation relation between resource and influencing factors based on BP neural network is established, and quantity of re- source consumption on mixed-processing activities is predicted through the above allocation relation. Finally, through actual examples, using 5-fold cross validation method, the resource allocation model' s generalization ability is increased. Compared with other methods of regression analysis, this method has better accuracy and convergence rate. It provides method for reasonably calculating resource consumption on mixed-processing activities, of which cost driver is difficult to determine.
出处
《运筹与管理》
CSSCI
CSCD
北大核心
2013年第1期179-186,共8页
Operations Research and Management Science
基金
国家自然科学基金资助项目(70872014
71172137)
关键词
资源分配方法
混合加工作业成本
BP神经网络
粗糙集
钢铁
cost control method
resource allocation
mixed-processing activity cost
BP neural network
rough set
Iron and steel