Font Size: a A A

The Research And Implementation Of Inventory Forecasting Model Based On ARIMA-RBF

Posted on:2020-11-23Degree:MasterType:Thesis
Country:ChinaCandidate:X Q ShiFull Text:PDF
GTID:2428330626450751Subject:Software engineering
Abstract/Summary:PDF Full Text Request
In the global computer industry,the competition among computer manufacturers is becoming increasingly fierce,companies are constantly seeking management methods that can adapt to the development of the market.In order to improve the company's core competitiveness and service quality,efficient after-sales maintenance service is particularly important,maintaining adequate inventory is the basis of ensuring quality after-sales service,appropriate inventory management can effectively improve the service level of enterprises.So inventory control arises at the historic moment.Inventory control is to control the inventory in the best state,neither excessive backlog nor shortage of inventory can occur.Related parts is a special part of computer parts.Related parts are also called additional parts,it can be used alone.In addition,When the main parts are used in the maintenance process,related parts needs to be bundled and used.Because of its particularity,if the quantity of related parts is insufficient,the main parts can not be used,which will cause idle products and engineering detention,it will bring greater impact.In view of this situation,it is of great significance and practical value to study the inventory forecasting model related parts.This thesis mainly completes the following work:(1)For the product characteristics of the related parts,the statistical rules of data are introduced,and the related factors affecting the demand of the connected materials are analyzed.In order to reduce the influence of outliers on the accuracy of demand forecasting,the quartile distance method is used to screen outliers and eliminate them.(2)The ARIMA-RBF combined forecasting model is proposed to predict the demand for related parts.Firstly,the ARIMA model is used to predict the demand,and then the defect of the ARIMA model is solved by the RBF neural network which is good at processing multidimensional vectors,the prediction error of ARIMA is corrected by adding the demand influencing factors.The core of the RBF neural network is the selection of the basis function center.This thesis finds the clustering center of the training samples through k-means,and uses the cluster center as the base function center of the RBF neural network.The center selected by this method is more representative.(3)This thesis compares the ARIMA-RBF combined forecasting model with the experimental results of multiple predictive methods such as single ARIMA forecasting method and exponential smoothing method,and proves that the combined model effectively improves the accuracy of demand forecasting of related,and it has good effect.
Keywords/Search Tags:Inventory forecast, Autoregressive Integrated Moving Average Model, Radial basis function neural network, Related part
PDF Full Text Request
Related items