Research On Cluster Analysis And Demand Predicting Method Of CN Enterprise Spare Part Inventory

With the rapid development of economy, resulting in per-capita disposable income increased and user demand diversification. To meet the consumer demand, production and manufacturing enterprise increase the product types and spare parts inventory. Huge inventory in the waste of resources caused by increased the operating risk of the enterprise. To assist enterprises in spare parts inventory management, maintain a reasonable inventory levels, taking the CN enterprise as the research object, the paper stock of spare parts demand forecasting methods were studied for the enterprise to develop a reasonable plan to provide spare parts inventory ideas.First, based on field investigation, analyzed about the present situation of CN enterprise, spare parts management process, the problems in the study of characteristics of the spare parts inventory, provide factual basis for follow-up study. On the basis of previous literature on classification of spare parts, spare parts prediction theory to analyze, compare features and applicability of various types of classification, prediction methods for selecting appropriate classification, prediction methods theoretical foundation.Then based on the data of CN enterprises of spare parts inventory records, to the inventory demand distribution for classification, based on the analysis of the clustering analysis method, choose the degree of similarity between Minkowski distance measure, using the average method to measure the distance between the class, class selection system clustering method to classify the spare parts inventory. Results show that the CN enterprise about 80% in the spare parts for slow moving spare parts.Finally according to the result of spare parts classification, selection of ARIMA model to simulate the stable spare parts demand forecasting, through the ADF test, seasonal difference, ACF/PACF truncated judgment, determine the SARIMA model for final prediction method. Type selection of combination forecast method for slow moving spare parts demand prediction, based on fuzzy demand sequence particle processing and optimization objective function and constraint conditions, the selected model of SVM parameters, FIG-SVM model was constructed. With a variety of forecasting methods are compared, and the precision showed that SARIMA, FIG-SVM has specific requirements in solving regular spare parts demand prediction problem has obvious advantages.