煤矿事故发生之前往往伴随着大量的险兆事件。为了预防事故发生、提高安全管理保障体系可靠性，不仅要做好一般的事故预防工作，更应该关注高概率险兆事件，并对其进行科学识别上报和分级决策。然而，煤矿高概率险兆事件风险程度具有模糊属性，安全管理者对高概率险兆事件的识别具模糊属性，煤矿条件也具有不确定性，导致高概率险兆事件的衍生后果也是模糊的，因此，如何科学地对煤矿高概率险兆事件进行分级决策，进而明确对应有效的管理措施，是预防煤矿事故亟待解决的问题，也是本文的研究重点。

首先，运用网络爬虫技术，于应急管理部官方网站、中国安全生产网等网站进行煤矿安全轻伤事件信息爬取，归纳整理其特点，并基于课题组对煤矿险兆事件的大量研究，界定出煤矿高概率险兆事件，并确定出识别方法和分级方法。

其次，通过文献调研、人员访谈、案例分析和扎根理论等方法对煤矿高概率险兆事件影响因素进行研究，通过开放性译码、主轴译码、选择性译码、饱和性检验等环节得出5个核心范畴和26个主范畴，通过德尔菲法专家修正后，构建出包含5个一级指标和24个二级指标的煤矿高概率险兆事件评价指标体系。

接下来，基于煤矿高概率险兆事件评价指标体系，运用AHP-熵权法计算指标权重，并结合危险性指数对高概率险兆事件进行等级划分，构建高概率险兆事件AHP-熵权法分级模型。除此之外，在模糊数学理论基础上，引入HF-ELCTREⅡ方法，增加了一致集、不一致集、无差异集的概念，构建了煤矿高概率险兆事件犹豫模糊多属性分级决策模型。

最后，将两种分级模型应用于A煤矿，确定出A煤矿高概率险兆事件分级标准，根据计算评估结果，对两种方法进行对比分析，确定最优分级模型，并基于该模型的分级结果分别从高概率险兆事件不同级别提出具体的管理提升措施。研究结果表明，本文构建的煤矿高概率险兆事件犹豫模糊多属性分级决策模型具有较好的实用性和创新性，能为煤矿险兆事件的管理和事故预防提供决策依据和理论借鉴

There are a lot of near-misses hidden behind the coal mine accidents.In order to prevent accidents and improve the efficiency of safety management, we should put the focus of management on high probability near-misses, classify the influencing factors. However, the risk degree of high probability near-misses in coal mine has fuzzy attribute, safety manager's identification of high probability near-misses has fuzzy attribute, and coal mine conditions also have uncertainty, which leads to the consequences of high probability near-misses is also fuzzy. Therefore, how to make scientific decisions on high probability near-misses in coal mines and then make clear the corresponding effective management measures is an urgent problem to be solved to prevent coal mine accidents, and also the research focus of this paper.

First of all, by using web crawler technology, the information of coal mine safety minor injury accidents were crawled on the official website of the Ministry of Emergency Management, China Work Safety website and other websites. According to the characteristics of these accidents, the high probability near-misses in coal mine are defined, and the identification method and classification method are determined.

Secondly, through literature research, personnel interview, case analysis and grounded theory and other methods to study the high probability of coal mine near-misses factors. Then, five core categories and 26 main categories are obtained through open decoding, spindle decoding, selective decoding and saturation testing. Through the Delphi method, an index system of influencing factors of high probability near-misses in coal mine is constructed, which contains 5 first-level indexes and 24 second-level indexes.

Next, based on the index system of influencing factors of high probability near-misses in coal mine, AHP-entropy weight method is adopted to determine the weight of the index, which classifies the influencing factors of high probability dangerous events. In addition, based on the fuzzy mathematics theory, the introduction of HF-ELCTRE Ⅱ method, build a high probability of coal mine near-misses hesitant fuzzy multi-attribute decision model.

Finally, the two classification methods are applied to A coal mine, and the classification standard of A coal mine's high probability near-misses is determined respectively. According to the results of calculation and evaluation, specific management improvement measures are proposed from different levels of high probability near-misses. The research results show that the model of hesitant multi-attribute decision making in this paper has good practicability and innovation. It can provide the decision-making basis and theoretical guidance for the management and accident prevention of near-misses in coal mines.