Dynamic Multi-Criteria Material Classification And Procurement Management

Material classification and procurement management are the core problems in the practice of supply chain management.There are too many kinds of materials in manufacturing enterprises.In the face of increasingly fierce market competition and growing customer demand,material classification is an important means to improve the efficiency of material management.Enterprises can effectively reduce costs and material supply risks by classifying materials intelligently and formulating corresponding procurement management strategies.There are three main problems in current material classification research:(1)unbalanced multi-criteria material classification,(2)classification results lagging behind the market demand,and(3)differentiated procurement management.To solve these problems,this thesis studied the dynamic multi-criteria material classification based on SMOTE-SVM and self-adaptive combined forecasting model and suggested procurement strategies.The main research work is as follows:(1)Aiming at the problem of unbalanced multi-criteria material classification,the design theoretical basis of the index system was analyzed,and the profit impact and supply risk dimensions of the Kraljic model were subdivided into multiple more specific indices.Taking company D as an example,the multi-criteria material classification index system was established.The SMOTE method was improved to over-sample the data set of multi-categories unbalanced materials,and then the materials were intelligently classified into four categories:strategic materials,leverage materials,non-critical materials and bottleneck materials through SVM algorithm.The improvement of the classification accuracy of minority categories of materials makes the multi-criteria material classification based on SMOTE-SVM more in line with the actual requirements of enterprises.(2)To solve the problem that the classification results of traditional models lag behind the market demand,the self-adaptive combined forecasting model was proposed.It integrated the advantages of multiple forecasting models,adopted self-adaptive iteration to adjust weights and regularization to avoid over-fitting problems,and achieved demand forecasting with higher accuracy.Based on the metabolic GM(1,1)model,BP neural network and LSTM model,the forecasting effect of the self-adaptive combined forecasting model was verified through the comparison of models.The forecasted product demand information was transformed into material demand information,and the dynamic multi-criteria material classification model was implemented.(3)The differentiated procurement management based on material classification was deeply studied.According to the four categories of materials after intelligentized classification,the characteristics of all categories of materials were analyzed qualitatively,and the appropriate purchasing strategies,inventory strategies and supplier management strategies were formulated.A tender procurement model of multi-source procurement with optimal quotation under capacity limitation was established.Through Monte Carlo simulation,the influence of four main parameters on the material procurement cost of the tenderee was quantitatively analyzed,and the corresponding tender procurement strategy was proposed.Precise procurement management had achieved the purpose of ensuring the stability of material supply and reducing costs and supply risks.