The Design And Implement Of Spare Requirements Prediction On Packs Of Power Communication Equipment

Spare parts management is one of the important parts of equipment management in enterprises, which is closely related to the production performance and economic benefits. On the process of spare parts management, it has not only to meet the (?)nance and repairing requirements, but also to avoid excessive funds used in purchasing and storing at the same time. Therefore, an accurate prediction of spare parts demand is very necessary to the inventory controlling and optimizing. Thus, researchers launched various studies and raised plenty of methods in order to improve the prediction quality of spare parts demand. Some of those methods have already achieved good results in certain areas, like aviation and military. However, from the current situation, researches targeted to power communication equipment have not been involved. Accordingly, it still needs to be explored that whether the existing approaches can help improve the situation in management of spare parts for electric power companies, which is still rely on the less accurate empirical formula.In this thesis, it aims to carry out statistical analysis and researches on the characteristics of existing running data about packs of power communication equipment and its spare parts, which is based on a provincial power company, in order to achieve the following three target outcomes. Firstly, obtaining the properties describing the classification characteristic of packs spare part from data analysis, to establish a classification model based on K-medoids clustering algorithm. It targets to divide the different types of packs spare into three categories as their importance to the power company. The classification result will provide reference to related department, in order to facilitate the management department to manage all types of situations based on different strategies.Secondly, building pack spare parts demand forecasting model with the RBF kernel function to v-SVR by performing a lot of theoretical and experimental analysis on data describing the characteristic of pack spare parts demand forecasting problems using class H pack spares as sample. Also forecasting and evaluating the prediction results from the natural monthly basis pack spares demand data.Thirdly, analyze the process of pack spares classification and spares demand prediction from statistic of the historical database system and processes of the company’s daily operation and management. Performing a detailed design of overall structure and specific functional modules and building the main functions of the pack spare classification and prediction applications. It makes a useful exploration in the application of the classifying and predicting model raised in this thesis in a real production environment.