Forecasting And Optimizing Safety Goods Stockpile Of Storage Management

The storage is huge investment for manufacturing enterprises. And how to control storage cost is important in corporation management. OP (Order Point) is a traditional measure which can not make further control. But JIT can make up this shortage.But JIT method is usually an ideal state. It is impossible to realize zero storage in practical application. No matter which enterprise cannot all achieve zero storage standards. The best way is minimize the storage. In order to avoid the loss which comes from the urgent provide, many enterprises usually make a certain safety stock which according to the phase and environment.Safety Goods Stockpile is a buffer to compensate the storage when actual needs exceed anticipant demands. Some dressing corporations set up Safety Goods Stockpile based on the experience, which leads to an undesirable effect. The key of storage control is how to choose the best method to forecast the Safety Goods Stockpile of the model.The paper makes some work in forecasting and optimizing Safety Goods Stockpile:(1) Under uncertainty environment, the relation between USL and OSL is analyzed. The simplified relation is studied between Safety Goods Stockpile and parameters. Through math experiment, it is easy to get single service level is higher than period service level. And, it is safer for managers to take period service level to control service quality. The relation of simplified model is found with the method of linear multivariate regression and BP neural network methods.(2) Take clothing enterprise as example building BP nerve network model, using many index affecting Safety Goods Stockpile sample book to carry out training on nerve network Establish the model of BP nerve network. To reach the purpose forecasting to Safety Goods Stockpile. Effect approaches the experience certificate.(3) There exists convergence speed is slow problem in neural networks. In order to solve the problem of convergence speed, Genetic Algorithm is used to optimize BP nerve network. Simulation results show the validity of the algorithm.