Emergency decision making (EDM) can effectively reduce the loss caused by emergencies. The generation and selection of the alternatives are two key stages of EDM. To improve the efficiency of EDM, we propose a two stage EDM method. In the first stage, a knowledge-unit based case-based reasoning (KU-CBR) method is provided to generate alternatives. In this method, an emergency case representation model based on knowledge-unit is constructed to represent emergency, and a scenario similarity measurement is proposed to retrieve similar historical scenarios, and the alternatives are generated with reference to historical scenarios’ solutions. In the second stage, a decision-making process with linguistic intuitionistic fuzzy preference relations (LIFPRs) is developed to select the optimal alternative. Firstly, to determine missing values in the incomplete LIFPR, a programming model based on additive consistency is constructed. Subsequently, a model is built to obtain the LIFPRs with acceptable additive consistency. Afterward, in order to rank alternatives, a model for deriving the linguistic intuitionistic fuzzy (LIF) priority weights of the alternatives is constructed, and the priority weights can obtained conveniently and quickly. The two-stage EDM method combines the advantages of KU-CBR and LIFPRs, which is demonstrated by a case application of typhoon emergency, and the decision-making process with LIFPRs is compared with an existing method to demonstrate its practicability.

应急决策（EDM）可以有效减少突发事件造成的损失。备选方案的生成和选择是 EDM 的两个关键阶段。为了提高 EDM 的效率，我们提出了一种两阶段 EDM 方法。在第一阶段，提供了一种基于知识单元的基于案例的推理（KU-CBR）方法来生成备选方案。该方法构建了基于知识单元的突发事件表示模型来表示突发事件，并提出场景相似度度量来检索相似的历史场景，并参考历史场景的解决方案生成备选方案。在第二阶段，开发具有语言直觉模糊偏好关系（LIFPRs）的决策过程来选择最佳替代方案。首先，为了确定不完全 LIFPR 中的缺失值，构建了基于加性一致性的规划模型。随后，建立模型以获得具有可接受的添加剂一致性的 LIFPR。然后，为了对备选方案进行排序，构建了备选方案的语言直觉模糊(LIF)优先权权重模型，可以方便快捷地获得优先权权重。两阶段 EDM 方法结合了 KU-CBR 和 LIFPRs 的优点，并通过台风应急的案例应用进行了论证，并将 LIFPRs 的决策过程与现有方法进行了比较，证明了其实用性。为了对备选方案进行排序，构建了备选方案的语言直觉模糊（LIF）优先权权重推导模型，可以方便快捷地获得优先权权重。两阶段 EDM 方法结合了 KU-CBR 和 LIFPRs 的优点，并通过台风应急的案例应用进行了论证，并将 LIFPRs 的决策过程与现有方法进行了比较，证明了其实用性。为了对备选方案进行排序，构建了备选方案的语言直觉模糊（LIF）优先权权重推导模型，可以方便快捷地获得优先权权重。两阶段 EDM 方法结合了 KU-CBR 和 LIFPRs 的优点，并通过台风应急的案例应用进行了论证，并将 LIFPRs 的决策过程与现有方法进行了比较，证明了其实用性。