Research Of Spare Parts Inventory Management Optimization Methods In Manufacturing Industry

With the increasingly high level of automation in the manufacturing industry, equipment becomes more and more important in routine production. Device management, which includes spare parts management, operation and management and maintenance management, has become an important part of business management for manufacturing enterprise. In many companies, equipment management fees accounted for more than8%of production costs, and cost of spare parts accounted for70%-80%of the maintenance cost. However, some spares have scrapped and can no longer support the production because of long time leave unused, while some others failed to keep accurate inventory levels thus seriously affected the production continuity. In addition to spare parts demand uncertainty, inventory management of different kinds of spares has also become an increasingly prominent problem, so scientific spare parts management which guarantees lowest cost and the optimal spare parts supply and management while continuously improve reliability, maintainability and efficiency of equipments has becomes an concerned issue.In this study, rather than focusing on process fragments, complete process of spares management consists of spares classification, demand forecasting and inventory management is proposed to support business management on the basis of extensive research on spares management. The proposed ABC analysis is used to classify the spare parts which considering both qualitative and quantitative evaluation. After that, different demand forecasting algorithms are proposed to predict spares demand:for continuous time series, exponential smoothing is used, for interrupted time series, classical croston method is recommended,for fluctuating and unusual time series, weibull distribution is used to fit failure probability curve of spares, demand forecasting can be calculated by estimating its failure probability. Inventory policy optimization is then proposed to complete the spares management process.For some kinds of spares demand forecasting which lack of historical data, an integrated algorithm of principal component analysis and support vector machine is proposed, principal component analysis is used to reduce operational dimension because of various factors affecting spares demand.SVM is used to train the sample data and obtain optimal regression relationship between various factors and spares demand. At last, a group of data of spares from a submarine is used to test the model, and results show that PAC and SVM-based forecasting model produced more accurate forecasts then SVM-based model and can provide effective forecasting result.