Abstract: This paper focuses on the determination of attribute weights in multiple attribute decision making problems.A novel method is proposed based on the Credal network.It is assumed that the weights are from various sources，including the previous information from experience and the values calculated according to the different decision makers’ preference using different approaches，etc.Because of the ability to capture and model expert knowledge under very general conditions，the Credal network is constructed and computed to obtain the composite weight intervals.Finally，two numerical examples are used to illustrate the effectiveness of developed method.

Key words: Credal network, multiple attribute decision making, weights, aggregation

摘要： 针对多属性决策中属性权重的确定问题，利用Credal网络提出了一种新的融合方法。该方法假设属性权重有多方面的来源，包括以前决策时的经验信息，不同决策者的偏好产生的权重信息，不同方法计算的权重信息，等等。建立了一个Credal网络模型进行推理计算，得到了各属性的综合权重区间，使之既包含了以前的经验信息，又包含了多个决策者给出的偏好信息。用2个数值算例说明了方法的有效性。

关键词: Credal网络, 多属性决策, 权重, 融合