Development Of Demand-Forecasting Module In Material Requirements Planning System

With the development of technology, competition between manufacturing organizations becomes more and more intense, and an excellent MRP (Material Requirements Planning) system becomes to be the key to get the success for an enterprise. The enterprise can use the MRP system to manage inventory, save expenditure, offer a good solution when emergency happens. As an important part of the MRP system, Demand- Forecasting module determines an MRP system's merit. With the forecast results of Demand-Forecasting module, the enterprise can manage inventory better. Demand-Forecasting module enables materials to be sufficiently prepared, in terms of inventory, for their customers, and it also affords the firm opportunities to control costs.New SASS (Stock Adjustment Simulation System) is a MRP system developed for Fujicco which is a manufacturing organization. The writer engaged in the development of this system. The MRP system Fujicco used to adopt is not suitable for itself. With analyzing the requirements of this enterprise, New SASS was developed. The writer is the designer and programmer of the Demand-Forecasting module in New SASS. This paper is base on the writer's experiences gained during the development.This paper includes these aspects:1. Time series models used in the module were introduced: Single Moving Average Model, Exponential Smoothing Model, Autoregressive Model, Moving Average Model, Autoregressive Moving Average Model. During the development, these models were designed to be a class library, so they can be reused in other projects later. And in the class library a class was designed to use all these models to forecast, with the analysis of the MAPE, it can choose the best model for demand forecasting.2. In the paper, US Census Bureau's X-11 Seasonal Adjustment procedure was introduced. In the Demand-Forecasting module, X-11 Seasonal Adjustment procedure is used to analyze data, to get rid of the season factors and trend factors of the data, so we can get more accurate forecast results. X-11 Seasonal Adjustment procedure was designed to be a class library too. As we know, X-11 Seasonal Adjustment procedure is generally used to analyze data by month or by quarter, but in New SASS, it is used to analyze data by week.3. In the paper, the modules in New SASS and the database used in New SASS were simply introduced.4. The design and implementation of Demand-Forecasting module were discussed in detail. The module includes two modes: batch mode, simulation mode. The batch mode includes two parts: renewal goods connecting and Demand-Forecasting. New goods are lack of historical data, so renewal goods connecting was designed to supply data for Demand-Forecasting. With the connecting of old goods data, new goods can be forecasted. After renewal goods connected, Demand-Forecasting began. The implementation of Demand-Forecasting was discussed in detail too. As the most important part of Demand-Forecasting, data analyzing and data correcting was discussed in the paper. With X-11 Seasonal Adjustment procedure, we can get rid of the trend factors and season factors of data. With data correcting, we can correct the abnormal data. By this way, we can get more accurate forecast results. Besides, other details as dividing data by week into data by day were introduced in the paper. Simulation mode was discussed in the paper too. Simulation mode uses the same algorithm used in batch mode. The difference from batch mode is that in simulation mode, customers can simulate Demand-Forecasting by set parameters themselves, and the results will be shown in the screen. With this interactive mode, customers can manage inventory better.5. TZBM (Three-zone Buffer Management)'s applications to Demand- Forecasting was discussed in the paper. TZBM is used to avoid unnecessary computer work and get more accurate forecasting results. TZBM is an approach presented for solving real world problems with many constraints, which are usually self-contradictory and difficult to be satisfied simultaneously. TZBM divides the complicated constraints into the hard and the soft, and then relaxes some soft constraints in a certain extent according to the current situations for avoiding the whole system trapping in the situations of deadlocks and violating constraints. All the individuals in the solution are divided into three zones according to the situations of satisfying constraints, and then TZBM pays a particular attention to improving the Red-zone, since all the constrained causes exist in this zone. In this system, TZBM was applied to solve data correcting in Demand-Forecasting procedure for avoiding unnecessary compute work.6. At last, an example was shown in both modes. And the analysis of the example was printed to show that the Demand Forecasting can get accurate forecast results and the procedure of Demand-Forecasting is efficient.New SASS is an excellent MRP system which includes an excellent Demand Forecasting module. Through various means as renewal goods connecting, forecasting with suitable time series models, adjusting data with X-11 Seasonal Adjustment Procedure, it can get more accurate forecast results. Besides, it can avoid unnecessary calculating by using TZBM. The module has been published with New SASS, and it gained praise from customers because of its excellence. As the designer and programmer of this module, I hope the readers can learn something useful in this paper for developing Demand-Forecasting module.