2014 年 8 巻 6 号 p. JAMDSM0076
In demand-synchronized production, when constructing flexible production systems to produce a variety of products while maintaining high efficiency, a preparation lead time is required because of such physical constraints as making setup changes and supplying facilities. Preparation lead time must also include time to arrange suitable software conditions. If forecasting required production volumes several days in advance with a certain degree of precision is possible, making systematic preparations for constructing the production system is also possible. The objectives of this research are to precisely forecast demand fluctuation within several days of delivery and thus have the ability to forecast production volume. We propose a method for forecasting production volumes, based on an analysis of past order data and the determination of their similarity by Fourier analysis. We confirm the efficiency of our proposed method by conducting experiments and forecasting production volumes using actual orders.
