矿井水害事故的发生对煤矿的安全开采带来很大的影响，快速准确地识别突水事件中所涉及的水源，是预防和解决矿井水害的关键手段之一。基于客观组合权和集对分析理论，针对矿井突水水源识别问题，以东滩矿4个含水层中的40个水样为识别样本，选取Na++K+、Ca2+、Mg2+、Cl-、SO42-和HCO3-共6项水化学指标作为识别因子，构建了矿井突水的水源识别模型，并利用该模型对10个检验样本进行水源识别。结果表明：客观组合权既避免了主观赋权法的主观随意性的缺点，又弥补了单一客观赋权法带来的片面性较强的不足，能合理地对识别指标进行权重赋值。Ca2+和Mg2+的权重明显高于其他离子，分别为0.28和0.25，二者的权重之和占比达53%，说明这2个指标在突水水源识别中起主要作用。利用客观组合权-改进集对分析模型对10个检验样本进行识别，准确率达到90%，说明该模型在矿井水源识别中具有一定的参考价值。模型出现误判的原因可能是误判水样取样钻孔位于断层附近，断层切割侏罗系红层水含水层与二叠系砂岩水含水层，是具有一定导水能力的天然通道，导致水样与砂岩水含水层存在一定的水力联系并发生了混合。建议在后续工作中应收集和监测各种含水层的水样数据，及时应用新的水样数据对模型进行修正，为提高模型的识别准确率奠定基础。

The occurrence of mine water hazards has a great impact on the safe mining of coal mines. Quickly and accurately identifying the water sources involved in water inrush events is one of the key means to prevent and solve mine water hazards. Based on the theory of objective combined weight and set pair analysis，aiming at the identification of mine water inrush water sources，40 water samples from four aquifers in Dongtan mining area were taken as identification samples，and six hydrochemical indexes including Na++K+，Ca2+，Mg2+，Cl-，SO2-4 and HCO-3 were selected as identification factors to build a water source identification model for mine water inrush. In addition，the model was used to identify the water source of another ten samples. The results show that the objective combination weight not only avoids the shortcoming of the subjective arbitrariness of the subjective weighting method，but also covers the shortage of strong one-sidedness from the single objective weighting method，and can reasonably assign the weight to the identification index. The weights of Ca2+ and Mg2+ are significantly higher than those of other ions，which are 0.28 and 0.25，respectively. The sum of the weights of Ca2+ and Mg2+ accounts for 53% of the overall weight，which indicates that these two indexes played major roles in the identification of water inrush sources. The 10 test samples were identified by the objective combination weight-improved set analysis model，and the accuracy rate reached 90%，indicating that the model has a certain reference value in mine water source identification. The reason for the misjudgment of the model may be that the water sampling hole is located near the fault，and the fault cuts the Jurassic red water aquifer and the Permian sandstone water aquifer，which are natural channels with certain water conductivity，resulting in water loss. There is a certain hydraulic connection and mixing with the sandstone water aquifer. It is suggested that the water samples data of various aquifers should be collected and monitored in the follow-up work，and the new water samples data should be employed to modify the model in time，so as to lay a foundation for improving the identification accuracy of the model.