中文摘要：

为保证矿井生产环境安全,以井下数据采集系统为基础,提出了最短距离聚类融合算法,以克服传感器测量误差和系统误差,实现对井下环境信息的数据级融合,与已有文献相比提高了融合精度。在此基础上,设计了概率神经网络分类器对数据级融合结果进行决策级融合,完成了矿井环境安全等级划分,实现了矿井环境安全状态的智能监测。实践表明：此两级融合方法在矿井环境监测系统中是可行的、有效的。

英文摘要：

In order to ensure environment safety of underground mine,a minimum-distance clustering fusion algorithm is put forward to overcome measurement error and system error of sensor,and achieve data-level fusion of underground environment information,compared with other mentioned references,fusion precision is improved.On this basis,probabilistic neural network classifier is designed to carry on decision-level fusion on data-level fusion result and achieve grade safety partition of underground mine environment.Practice shows that the two-step fusion methods are feasible and effective in underground mine environment monitoring system.