台风灾害是可以预测的自然灾害，但准确地把握其演变动态，制定一个合理高效的应急物资分配计划，仍是我国当前救灾工作的难点。本项目考虑从台风演变的视角，研究基于群组需求更新的应急物资分配多时段决策问题。首先，分析台风的演变规律，探讨群组需求更新的运作机制。在此基础上，使用贝叶斯序贯决策方法建立基于群组需求更新的应急物资分配多时段决策模型。针对灾害前后阶段的不同特性，分别制定多时段决策规则。然后，将模型进行整合，使群组需求更新贯穿于灾害前后阶段的所有时段中，从而使决策者能够全程把握灾害演变动态，实现应急资源的全局最优配置。最后，通过实证研究，验证群组需求更新运作机制和贝叶斯序贯决策方法的有效性。研究结果可为政府针对台风或洪涝灾害的应急响应机制提供智力支持和决策依据，丰富我国自然灾害应急资源配置体系的研究内容和范畴。

The typhoon is a predictable natural disaster. However, it is still difficult in the national relief efforts to accurately forecast its evolution, and make a reasonable and efficient relief allocation plan. From the perspective of typhoon’s evolution, this project studies the multi-time decision making problem in relief allocation via group-based demand updates. First, the evolution law of typhoon is analyzed, and the mechanisms of group-based demand updates are explored. After that, by using Bayesian sequential analysis approach, the multi-time decision making models for relief allocation via group-based demand updates are built. In these models, according to the different characteristics of the stage before or after the disaster, a multi-time decision making rules is set respectively. Then, an integrated model is built to update the group-based demands throughout the stages before and after the disaster, so that policy makers can continuously monitor the dynamic disaster evolution, and eventually achieve the global optimal relief allocation plan. Finally, empirical results indicate the validity of the mechanisms of group-based demand updates and the Bayesian sequential analysis approach. The research results of this project can provide the intellectual support in emergency response to the typhoons or floods, and enrich the research contents of the relief allocation system for natural disasters in our country.