A Research on Demand Forecasting and Supply Contract Support System Development for Innovative Product

本研究主要研究两个问题：一是需求预测问题，二是利用需求预测的最优供应合同问题。在需求预测中，我们构建了三层三次神经网络，其中中间层由两组组成，每组包含六个客户(代理人)类型，选择产品规格及其商家评价、价格折扣信息、同类产品的历史销售数据等信息作为网络的输入，并将网络反馈到中间层，以动态更新Sigmoid函数的参数。利用需求预测信息，进一步构建了以零售商利润、制造商利润和供应链综合利润为目标的初始订货量和期权数量优化模型。此外，我们还发现了期权合约模型的赢价区间。

This research deals with two problems : One is demand forecasting and the other one is optimal supply contract utilizing the demand forecasting. In the demand forecasting, we constructed three layers cubic neural network in which middle layer consists of two groups contains six customer(agent) types in each, when those information, such as product specification and its evaluation by merchandizers, price discount information, historical sales data of similar product were selected to be input of the network, and output of the network feedback to middle layer so that to dynamically update parameters of sigmoid function. Utilizing demand forecast information, we further constructed option models that optimize initial ordering quantity and option quantities for retailer's profit, for manufacturer's profit, and for integrated supply chain profits. Furthermore, we found that wining price zone of the option contract model.