Abstract: Aiming to overcome the low accuracy of demand forecasting in supply chain, this paper uses the genetic algorithm to estimate the developing coefficient and the control variable of the GM（1, 1） model and predicts the demand of every level in supply chain with this forecasting model, then uses a negotiation algorithm based on game theory to optimize the demand forecast when demand forecast disruption occurs among every level of supply chain. Results are very encouraging as this optimization clearly improves the accuracy of demand forecasting in supply chain.

Key words: supply chain, demand forecasting, game theory, grey model, Genetic Algorithms（GA）

摘要： 为解决供应链中需求预测精度不高的问题，在传统GM（1，1）预测模型的基础上，采用了一种基于遗传算法调整发展系数和内生灰作用量的灰色预测模型，运用此模型对供应链中各级需求量进行预测，使用博弈理论构建了供应链中各级需求预测出现误差时的协商策略用以对预测结果的优化。实验结果表明，协商策略获取了符合双方利益的需求量，预测结果有了较高的精度。

关键词: 供应链, 需求预测, 博弈理论, 灰色模型, 遗传算法