H公司基于动态架构人工神经网络的订单预测研究

发布时间：2018-04-14 19:59

本文选题：订单预测 + 神经网络　；参考：《华中科技大学》2016年硕士论文

【摘要】：在企业的生命周期中,订单预测是的血液,它可以帮助我们及时预测到市场的需求,快速的分析市场信息,为企业的市场战略营销和管理策略提供基础数据和分析。神经网络已广泛应用于自然科学和社会科学各个领域,并且都取得大量的成果。电子商务背景下对订单的处理速度和需求预测精度和速度都提出了更高的要求。本文研究H外贸公司的订单预测问题。文章首先分析了国内外神经网络和订单预测的研究现状,讨论了最小二乘法等理论知识,为后文中构建的一种动态架构人工神经网络模型奠定理论基础。接着基于神经网络构建了较优化的动态人工神经网络模型,利用最小二乘法改进了训练过程中的函数优化了非线性过程中的算法。构建出来的模型相比较于BP神经网络而言,结构更加简洁、理论上数据利用率更高。最后通过调研H公司的组织结构和订单预测现状,并进行数据的采集和分析,采用文章构建的动态神经网络模型进行订单预测的应用。应用结果表明,本文构建的动态神经网络模型是有效的、速度更快、结果更精确;发现了H公司在创新产品与业务、管理水平、订单预测和管理方面都还有较大的上升空间,因此也提出了公司管理需要注重市场研究和市场学习、保持产品的专注、了解全球文化的管理启示。本研究成果可以为外贸公司的订单预测提供有效的预测方法。
[Abstract]:In the life cycle of the enterprise, the order forecast is the blood, it can help us to forecast the market demand in time, analyze the market information quickly, and provide the basic data and analysis for the marketing and management strategy of the enterprise.Neural network has been widely used in various fields of natural science and social science, and has achieved a lot of results.Under the background of electronic commerce, the processing speed of orders and the precision and speed of demand prediction are required higher.This paper studies the order forecasting problem of H foreign trade company.This paper first analyzes the present research situation of neural network and order prediction at home and abroad, and discusses the theoretical knowledge of least square method, which lays a theoretical foundation for a dynamic architecture artificial neural network model constructed later.Then an optimized dynamic artificial neural network model based on neural network is constructed, and the least square method is used to improve the function in the training process to optimize the algorithm in the nonlinear process.Compared with BP neural network, the constructed model is simpler in structure and more efficient in theory.Finally, by investigating the organization structure of H Company and the status quo of order forecasting, and collecting and analyzing the data, the application of the dynamic neural network model constructed in this paper is used to predict the order.The application results show that the dynamic neural network model constructed in this paper is effective, faster and more accurate.Therefore, it is also pointed out that the management of the company should pay attention to market research and market learning, keep the focus of products, and understand the management enlightenment of global culture.The results of this study can provide an effective forecasting method for order forecasting of foreign trade companies.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP183;F274

【参考文献】