Research Of Mine Materials Planning Management System Based On PSO-SVM

As one of the main sources of China’s energy consumption, coal is of great importance inour country’s economy development, the extensive use of the Internet technology and mobileInternet technology, and the quickened proceeding of economy globalization, the marketcompetition becomes more and more ardent. Over all informatization level of China’s coalmine enterprise material plan management is relatively low, which causes the subsequentproblems of excess inventory, large amount of capital occupation and the increase ofproduction costs. In order to achieve the management philosophy of “Zero inventory”, thesupply chain should be extended to the coalface, fine management of the coal mine enterprisematerial plan needs to be made, information technical science should be employed to makevalid dynamic predication of the materials consumed in the production, intelligentmanagement system for the material plan report also needs to be built. All the above methodsalso act as the critical measures of minimizing inventory costs and improving the finemanagement level of the material supply chain.This thesis first analyzes the problems in the present material plan management of thecoal mine enterprise, and then demonstrates the relevant content of the material planmanagement. Based on the classified materials and the influential factors extracted from thesematerials needed by the underground mine operations as well as the analysis of the keyindicators, the key indicator system of the material demand forecast is worked out. As to theproblems of subjectivity and insufficient optimization of the parameters in the process ofparameter selection by the support vector machine model (SVM model), particle swarmoptimization method is adopted in optimal selection of the optimal parameters of the SVMmodel. According to the parameter results of the optimal selection, the SVM forecasts thematerial demand, for example，forecast shearer with the actual materials, and the forecastresults indicate that particle swarm optimization method optimizes the parameters of thesupport vector machine forecast model and improves its predicting accuracy. Based on the background of the author’s actual participation of the enterprise projecttopic “Supply chain e-commerce system of Shandong energy Zi mining group” and theanalysis of the business requirements of the Zi mining group’s plan management, this thesisworks out the whole framework of the mine’s enterprise material plan management systemand discusses the technology realization scheme. Data modeling is done based on the relevantmaterial demand forecast, particle swarm optimized SVM is chosen as the prediction modelfor the material demand forecast, based on J2EE development platform, Oracle10g data baseis adopted as the underlying data support platform, meanwhile, Activiti5process enginedesign is applied, therefore, a set of material plan management platform based onPSO—SVM according with the actual business of Zi mining group is developed. Finally, theresearch work of the dissertation has been summarized and the prospect of the further study ofthe system has been opened up.