L公司服务器配件需求预测研究

发布时间：2018-04-04 08:38

本文选题：服务器　切入点：配件　出处：《北京交通大学》2017年硕士论文

【摘要】：近年来,国家大力发展信息化建设,我国信息产业实现跨越式发展,以金融、电信、交通、教育等为代表的重点行业对服务器的需求迅速增长。L公司为国内知名的服务器制造商,以L公司为中心的服务器供应链中,向上游供应商采购服务器配件是影响公司效益的重要环节之一。本文将对L公司服务器配件需求预测予以研究,目的在于通过选择更加符合L公司实际情况的配件分类和预测模型,使得服务器配件分类更加细致,需求预测结果更加准确。L公司采购部门可以依照优化后的预测结果,对服务器配件进行采购。本文在相关理论研究的基础上,分析了L公司服务器配件需求预测现状,主要包括配件分类现状和配件预测现状。在分类问题上,L公司服务器配件存在分类交叉、客户满意度不高的问题。由于分类不合适,以及预测方法亟待改进,导致配件需求预测结果与实际值偏差较大。针对L公司服务器配件需求预测存在的问题,本文主要从两个方面开展工作:一是对配件分类进行优化;二是在优化配件分类的基础上,对配件需求预测进行优化。在现状分析的基础上,本文建立了需求预测优化的两阶段模型。第一阶段对L公司服务器配件建立分类模型,在传统帕累托分类的基础上新增了订单周期、采购时间和下单次数等影响因素,求解算法采用了 ID3算法。第二阶段建立了ARIMA-BP神经网络模型,该模型在传统的时间序列预测模型上,增加了对非线性特征数据的预测,使预测结果更加接近实际值。模型以优化配件分类和采购需求的预测结果为目标,对L公司服务器配件进行需求预测。本文结合L公司实际调研得到的数据对所建立的模型进行求解。首先利用MATLAB软件对第一阶段服务器配件的分类模型进行求解,得到更加细致的服务器配件分类结果和具有预测意义的配件种类;其次利用SPSS软件求解ARIMA模型;利用MATLAB软件求解BP神经网络。最后,将线性预测数据和非线性预测数据汇总得到优化后的预测结果。根据模型结果,对L公司服务器配件采购需求预测优化效果进行分析。一方面检验模型的实用性,另一方面验证优化结果对服务器配件采购需求预测的改善作用。研究结果表明,优化后的需求预测结果比优化前的预测结果更加接近实际值。
[Abstract]:In recent years, the state has made great efforts to develop information construction, and the information industry in our country has developed by leaps and bounds to finance, telecommunications, transportation,Education and other key industries represented by the rapid growth of demand for servers. L is a well-known domestic server manufacturers, L company as the center of the server supply chain,Purchasing server accessories from upstream suppliers is one of the important links that affect the efficiency of the company.The purpose of this paper is to make the server accessories classification more detailed by selecting the fitting classification and forecasting model which is more in line with L company's actual situation.The demand forecast result is more accurate. L purchasing department can purchase the server accessories according to the optimized forecast result.On the basis of relevant theoretical research, this paper analyzes the current situation of L company server accessories demand forecasting, mainly including the status of accessories classification and accessories forecasting.In the classification of the company's server accessories there is a cross-classification, customer satisfaction is not high.Because the classification is not suitable, and the forecasting method needs to be improved urgently, the forecast result of spare parts demand deviates greatly from the actual value.Aiming at the problems of L company's server accessories demand forecasting, this paper mainly carries out the work from two aspects: one is to optimize the accessories classification; the other is to optimize the spare parts demand prediction on the basis of optimizing the accessories classification.Based on the analysis of the present situation, a two-stage model of demand forecasting optimization is established in this paper.In the first stage, the classification model of L company's server accessories is established. Based on the traditional Pareto classification, the influence factors such as order cycle, purchase time and order number are added. The ID3 algorithm is used to solve the problem.In the second stage, the ARIMA-BP neural network model is established. In the traditional time series prediction model, the prediction of nonlinear characteristic data is increased, and the prediction results are closer to the actual value.The model aims at optimizing the classification of accessories and forecasting the purchasing demand, and forecasts the demand of L company's server accessories.In this paper, the established model is solved by combining the data obtained from the actual investigation of L Company.First, the classification model of the first stage server accessories is solved by using MATLAB software, and the more detailed classification results of server accessories and the kinds of accessories with predictive significance are obtained. Secondly, the ARIMA model is solved by using SPSS software.BP neural network is solved by MATLAB software.Finally, the linear prediction data and the nonlinear prediction data are summarized to obtain the optimized prediction results.According to the result of the model, this paper analyzes the effect of forecasting and optimizing the purchasing demand of server accessories in L Company.On the one hand, the practicability of the model is tested; on the other hand, the optimization results are verified to improve the demand prediction of server accessories.The results show that the optimized demand forecasting results are closer to the actual values than those before the optimization.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F274;F49

【参考文献】