煤矿是我国的主要能源，随着煤矿开采的不断深入，矿井瓦斯浓度逐渐增大，瓦斯事故日趋严重，这时刻威胁着国家和人民的生命财产安全，对煤矿生产中的瓦斯有效监测是十分迫切的。虽然煤矿企业对瓦斯的安全监测也建立了相应的监测系统，但它主要是针对局部的监测与管理，其缺乏对监测数据的有效处理以及深度分析，而本课题就是利用数据挖掘分析方法试图对相关的监测数据进行深度分析，以达到更好的预测功能。 本文的研究目的在于将数据挖掘技术应用到煤矿瓦斯监测系统的数据分析中，也就是，针对监测系统所采集的相关数据，引用聚类、时间序列分析等新型技术对其数据进行深度分析和处理，从而为煤矿生产中的瓦斯监测及预测提供一种探索和指导。 在介绍课题背景之后，对课题中涉及的相关技术进行了描述，然后通过分析煤矿瓦斯监测数据指标的特点，建立了数据挖掘处理模型，在此处理模型中，主要涉及以下几方面的应用：应用模糊K-均值法对煤矿瓦斯安全进行评级；应用时间序列分析方法中的ARIMA模型对瓦斯的日均浓度进行预测，把握瓦斯浓度的分布及变化趋势；通过运用数据挖掘中的关联规则挖掘方法对安全监测数据进行多维关联规则挖掘，发掘数据间有意义的强关联规则，并提出了一种基于规则前件的最小支持度阈值的估计方法，用于关联规则挖掘中最小支持度阈值的确定。实证研究表明，模型性能良好，简便易行，所采用的方法合理有效。

Coal-mine is a major source of energy in China, with the continuous deepening of coal mining, coal mine gas concentration increases gradually and gas accidents become increasingly serious,which is threatening the country and the people's lives and property, the effective monitoring of the gas in the coal mine production is very urgent. Although coal mining enterprises also established a monitoring system of gas safety, it is mainly suit to local monitoring and management, lacking of effective treatment and deep analysis of the monitoring data. In order to achieve better prediction function, this study use data mining analysis methods to make in-depth analysis of the relevant monitoring data. The purpose of this paper is that the data mining techniques are applied to the analysis of the data of the coal mine gas monitoring system, that is, we use clustering, time series analysis, and other new technologies to make in-depth analysis of the data which is collected by the monitoring system, to provide an exploration and guidance for coal mine gas monitoring and forecast. After introducing the topic background, a description of the technology involved in the subject, and then by analyzing the characteristics of the coal mine gas monitoring data indicators, set up a data mining processing model, and on this basis to do the following aspects: Application of fuzzy K-means method for coal mine gas safety rating; application of ARIMA model of time series analysis method to predict the daily average concentration of the gas, which is to grasp the gas concentration distribution and trends; In order to excavation strong association rules ,we mine multi-dimensional association rules on safety monitoring data through using the method of association rules,and proposes a method for estimating the minimum support threshold based on rule antecedent, which is used to determine the minimum support threshold in mining association rules.Empirical studies have shown that the model performance is good, simple, the method used is reasonable and effective.