基于成本动因BP神经网络的铁路物流货运成本预测

发布时间：2018-10-16 11:35

【摘要】：铁路货物运输是铁路运输的重要组成部分,合理的控制货运成本可以有效地降低铁路运输总成本,而加强货运成本管理的重要前提就是准确及时地进行货运成本预测。选择合理的预测方法,可以提前为铁路运输企业提供有效的成本信息,为企业应对复杂多变的市场环境提供保障。成本预测方法很多,但并不是所有预测方法都符合铁路运输成本的特点,适合我国铁路运输的实际情况。本文通过比较不同预测方法得到的铁路货运成本预测结果,确定一个符合铁路运输成本特点的预测方法。通过分析我国铁路运输成本的构成及影响因素可以得出,在铁路运输成本支出中间接成本占较大比重,而且运输成本数据具有较强的非线性,在对运输成本进行预测之前,既要选择恰当的间接费用分配指标合理地分配客货运成本,又要充分考虑铁路运输成本预测自身特性。基于以上两点考虑,选择作业成本法和BP神经网络相结合的方法对铁路货运成本进行预测。首先,从作业角度出发,利用成本动因理论,选取对货运成本影响较大的成本动因作为分配间接成本的指标,并结合定性分析和定量分析利用聚类分析法合并成本动因,再利用BP神经网络很强的学习能力、容错能力以及非线性映射能力,解决运输成本样本采集不精确和成本数据呈非线性关系的问题。以所选成本动因为输入变量,在MATLAB软件中反复进行信号的正向传播和误差的反向传播过程,不断学习训练,存储学习结果,获得预测结果。并将结果与灰色系统预测方法所得的预测结果进行比较。最后以某铁路运输企业货运成本预测为例,分别用成本动因BP神经网络预测方法和灰色系统预测方法进行预测,并分析。结果表明,通过作业成本法选择成本动因,然后利用BP神经网络模型进行成本预测的方法比灰色系统预测法更适合用于国铁路货运成本预测。
[Abstract]:Railway freight transportation is an important part of railway transportation, reasonable control of freight cost can effectively reduce the total cost of railway transportation, and the important premise of strengthening freight cost management is to accurately and timely forecast freight cost. Choosing reasonable forecasting method can provide effective cost information for railway transportation enterprises in advance and provide guarantee for enterprises to deal with complex and changeable market environment. There are many cost forecasting methods, but not all of them are in line with the characteristics of railway transportation cost and are suitable for the actual situation of railway transportation in China. By comparing the forecasting results of railway freight cost with different forecasting methods, this paper determines a forecasting method which accords with the characteristics of railway transportation cost. By analyzing the composition and influence factors of railway transportation cost in China, it can be concluded that indirect cost accounts for a large proportion of railway transportation cost expenditure, and the transportation cost data have strong nonlinearity. It is not only necessary to select appropriate indirect cost allocation index to reasonably allocate passenger and freight costs, but also to fully consider the characteristics of railway transportation cost prediction. Based on the above two considerations, the combined method of activity-based costing and BP neural network is chosen to predict the freight cost. First of all, from the point of view of activity, using the cost driver theory, select the cost driver which has a great influence on freight cost as the index to allocate indirect cost, and combine qualitative analysis and quantitative analysis to combine the cost driver with cluster analysis. Then the BP neural network is used to solve the problem that the transportation cost sampling is imprecise and the cost data is nonlinear by using the strong learning ability, fault-tolerant ability and nonlinear mapping ability of the BP neural network. Taking the selected cost driver as the input variable, the forward propagation of the signal and the backward propagation of the error are repeatedly carried out in the MATLAB software, and the learning results are stored and the prediction results are obtained. The results are compared with those obtained by the grey system prediction method. Finally, taking the freight cost prediction of a railway transportation enterprise as an example, the BP neural network forecasting method and the grey system forecasting method are used to predict the cost driver, and the analysis is made. The results show that the cost driver is selected by activity-based costing method, and then the cost forecasting method based on BP neural network model is more suitable than the grey system forecasting method.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F253.7;F532;F224

【相似文献】