针对尾矿库安全评价中存在的不确定性问题，以黄金洞尾矿库为例，引入未确知测度理论，分析评价对象与评价指标之间的关系，建立一个包含5个影响因素和18个影响因子的尾矿库综合安全指标评价体系，结合基于层次分析法—熵权法综合确定指标权重，构建基于改进熵权法—未确知测度模型。根据该模型的指标评价体系和分级模式确定指标测度函数，代入各指标实测值，得出多指标综合测度向量，并借助置信度识别准则准确判定尾矿库安全等级和各指标未确知测度重要度。结果表明：黄金洞尾矿库安全等级为Ⅱ级，尾矿库处于较安全状态。经比较，指标测度重要度从大到小为地基沉陷（B₃）>洪水漫顶（B₁）>安全管理（B₄）>坝体溃坝（B₂）>自然因素（B₅），可知地基沉陷对尾矿库安全影响程度最大，应加强对地基沉陷的管理。基于改进熵权法—未确知测度模型得到的评价结果与实际结果相吻合，为尾矿库安全性评价提供了一种可行的方法。

In recent years，the state has strict requirements on the safety production of mining enterprises.The safety problem has become one of the most important problems of each mining enterprise.Tailings pond is an indispensable part in the production process of mining enterprises，so the safety of tailing pond is closely connected with the local development.The dam break of tailing pond will cause great loss and serious damage to local environment and economic development，and even endanger people’s lives.By evaluating the safety of tailings pond reasonably，the safety accident of tailings pond can be avoided effectively，so as to ensure the normal safety production of mining enterprises.The evaluation of tailings ponds is characterized by uncertainty，complexity and variability，and there are many indexes factors affecting the safety of tailings ponds.The information conveyed by most of the influencing factors is of significant uncertainty and randomness，which makes the safety evaluation of tailings ponds a complex and changeable problem filled with many uncertain factors.Aiming at the problem of uncertainty in safety evaluation of tailings pond，the theory of unascertained measure was introduced to analyze the relationship between the evaluation object and the evaluation indexes.Taking the Huangjindong tailings pond as an example，by selecting 18 influencing factors，a comprehensive safety index evaluation system of tailings pond containing 5 types of influencing factors and 18 influencing factors was established.The comprehensive weight of indexes was determined by combining the analytic hierarchy process（AHP）-entropy weight method，and an unascertained measure model based on the improved entropy weight method was constructed.By quantifying qualitative factors，this model eliminates the differences among factors，improves the problem that the subjective and objective weights have small differences and cause the evaluation results to change greatly，and weakens the influence of weight values on the evaluation results.At the same time，the index measure function is determined according to the index evaluation system and classification mode of the model，and the measured value of each index is substituted into it to obtain the comprehensive measure vector of multiple indexes，and the safety level of tailings pond and the importance of unascertained measure of each index are accurately determined by the confidence recognition criterion.Results show that the security level of Huangjindong gold tailings is Ⅱ level，means the tailings is in a safe state.By comparing，the importance level of indicator measure from high to low is foundation subsidence（B₃），flood top（B₁），safety management（B₄），dam break（B₂），natural factor（B₅）.That means foundation subsidence has the greatest impact on the safety of tailing pond，it is necessary to strengthen management of ground subsidence.The evaluation results based on the improved entropy weight method-unascertained measure model are consistent with the actual results，which provides a feasible method for the safety evaluation of tailings pond.