Abstract: The traits of supply uncertainty in reverse logistic system are analyzed firstly. And then based on grey prediction prerequisite, a new stricter jump point judgment is advanced. As a result, an improved jump-point grey model is developed and its arithmetic approaches are presented.On this basis, grey prediction system and logic frameworks of grey prediction agent are designed. Finally, an online prediction system by grey modeling jump series on reverse logistics based on agent is brought forward

Key words: Reverse logistics; Online prediction; Jump series; Grey prediction; Agent

摘要： 分析了逆向物流供应不确定性的特点。在分析灰预测条件的基础上，提出了一种更严格的跳变点判断标准，建立了跳变点序列的IJGM 模型，并给出了相应的算法和步骤。在该基础上，给出了逆向物流供应的灰预测体系和灰预测Agent 的逻辑结构。最后，为实现在线预测提出了基于Agent 的逆向物流跳变灰预测生产监控框架。

关键词: 逆向物流；在线预测；跳变序列；灰预测；Agent