Study Of Product Demand Forecast Base On ANN

Product forecast is the key factor of supplier chain management in manufacturing factory. Base on this, factory can define the reasonable material purchasing plan, production plan, manpower plan and inventory plan. Forecast study is important for factory's management. In this article, this subject is studied base on ANN as below issues:(1) Introduced ANN, especially the principle and algorithm of BP network. Discussed constitution, formulation and characteristic of it, as well as the sample selection and pretreatment. It was proved 3 layers BP network can realize all complicated input and output relationship.(2) Introduced product requirement forecast system in Atlascopco company. Introduced the important role of planning BOM in the forecast system. It was proposed to formulate planning BOM with BP network, then studied and prepared basic mode design according this. Meanwhile, we also make compared study with current method: use average historical data.(3) Introduced program realization of BP network. It was in detail discussed the forecast system mode with BP network to be realized with MATLAB program. Make forecast for the planning BOM with trained network, make sure the effectiveness and liability of this mode. The result showed the new system with BP network is better obviously than the old system which only use historical data average result in the factory. Normally for manufacturing factory, it is no need to study forecast because they can get it from sales department, but for Atlascopco Wuxi, factory planning have to make forecast because we only can get the forecast of product range from sale, this can not be taken as input of ERP system, so we have to involve in the forecast study. Planning BOM is the key factor to translate the sales forecast to product requirement forecast. ANN is the way to set up mode by data itself, it can learned the internal relationship of the data to get forecast result. Good feasibility, anti-annoy ability and self-learning ability are the advantage of ANN. Base on all of this, ANN provide a good tool for the forecast study, we can use it to make our forecast more accurate and more scientific, then can help factory to get more reasonable forecast information.