中文摘要：

针对传统的相对熵集结模型在赋权方面存在的缺陷，本文基于熵可靠性对其进行优化。首先利用广义熵的可靠性分析，剔除原专家群组中可靠性较低的专家，建立新的专家群组；其次基于熵的可靠性，确定新群组中专家的权重；最后将权重应用于相对熵集结模型中，来集结专家群组的一致性。算例结果表明，优化后的相对熵集结模型是有效的，而且可靠性得到很大的提高。

英文摘要：

In view of the irrationality of the traditional REM assembly in empowering the weights, which can result in wrong evaluation results in a group decision-making, we present the REM based on the relative reliability of entropy. Firstly, we take advantage of the reliability of general entropy analysis about reliability of the results and then delete the original group of experts with the lower reliability, and create a new group. Secondly based on the reliability of entropy, we set up the weight of the new group of experts. Finally, we apply weights to relative entropy assembly model to assemble group consistency. At last the simulation results show that the optimized relative entropy assembly model is effective and that its reliability is greatly improved.