为研究大佛寺井田煤层气富集高产规律及开发潜力区的分布，采用液氮吸附、核磁共振、等温吸附等实验，研究了该区域的储层特征、煤层气成因、富集因素；采用主成分方法结合支持向量机进行了含气量预测，采用灰色关联方法分析了区域煤层气产能主控因素，并总结了该区域的煤层气高产模式，最后通过熵权法和随机森林算法对该区域煤层气开发潜力区做出了预测。论文取得了以下主要研究成果：

（1）大佛寺4号煤有机显微组分含量较高，以惰质组为主，其次为镜质组。孔隙以植物组织孔为主，裂隙以张裂隙为主，连通性较强。吸附特征表现为平衡水基的兰氏体积为9.30~10.76 m³/t，兰氏压力为3.56~4.72 MPa；干燥无灰基的兰氏体积11.68~17.2 m³/t，兰氏压力为3.56~4.72 MPa；渗透率在区域上的特征为东部渗透性好，西部较差。储层压力在区域上的特征是，井田中部沿东西向储层压力变大，向南北两端逐渐减小。

（2）大佛寺井田含气量与镜质组、煤厚、埋深和顶板泥岩厚度以及10米煤泥岩厚度呈正相关，与挥发分、灰分、惰质组呈负相关；井田内煤层气成因属于次生生物成因，煤层气成藏模式属于典型的盆缘缓坡晚期生物气成藏模式；根据主成分法筛选出煤厚、延安组泥岩厚度两个与含气量相关的主要因素，利用支持向量机预测的含气量，预测结果显示R²值达到0.795，均方误差为0.02，确认不灵敏参数ε、核函数参数γ、惩罚参数C分别为 0.01、1、32。

（3）大佛寺煤层气直井和水平井产能主控因素有所区别，直井产能主控因素强到弱依次为：煤厚、埋深、渗透率、灰分、顶板厚度、含气量、储层压力，而水平井则为：含气量、煤厚、储层压力、灰分、渗透率、顶板厚度、埋深；大佛寺井田两种煤层气高产模式分别为鼻状构造轴部高产模式和单斜向上倾高部位高产模式。

（4）利用熵权法将井田煤层气开发潜力区分为5类：①不利区；②中等区；③较有利区；④有利区；⑤极有利区；根据Pearson相关系数和显著性检验分析，选取8个参数（含气量、灰分、煤层净厚度、构造位置、顶板厚度、渗透率、储层压力、埋深）用于模型建立。采用随机森林算法同样将煤层气开发前景分为5类，两种模型预测结果基本吻合，但随机森林算法对西部有利区的划分更加细致。从两种预测分析结果的整体空间分布来看，井田内的中、东部评价结果基本与目前开采效果一致。井田的东南部不适于后续部署井位，而井田的西部存在高有利区分布，后续开发部署井位可着重考虑。

In order to explore the CBM enrichment and high-yield law and the distribution of development potential areas in the Dafosi Minefield, experiments such as liquid nitrogen adsorption, nuclear magnetic resonance, and isothermal adsorption were used to study the reservoir characteristics, CBM genesis and enrichment factors in this area.Principal component method combined with support vector machine is used to predict the gas content. The gray correlation method is used to analyze the main controlling factors of regional coalbed methane production capacity. Summarized the high production mode of coalbed methane in this area, and finally predicted the potential area of coalbed methane development in this area through entropy weight method and random forest algorithm.This paper has achieved the following main research results:

(1) Dafosi No. 4 coal has a relatively high content of organic microscopic components, with the inert group as the main component, followed by the vitrinite group. The pores are mainly plant tissue pores, and the fissures are mainly tension fissures with strong connectivity. The adsorption characteristics are as follows: the Langmuir's volume of the equilibrium water base is 9.30~10.76 m³/t, and the Langmuir's pressure is 3.56~4.72 MPa. The Langmuir's volume of the dry ash-free base is 11.68~17.2 m³/t, and the Langmuir's pressure is 3.56~4.72 MPa. The regional permeability is characterized by good permeability in the east and poor in the west. The characteristic of the reservoir pressure in the wellfield is that the reservoir pressure in the middle of the wellfield increases along the east-west direction and gradually decreases toward the north and south ends.

(2) The gas content in the Dafosi Minefield is directly proportional to the vitrinite, coal thickness, burial depth, roof mudstone thickness and 10m coal mudstone thickness, and inversely proportional to the volatile component, ash content and inertinite. The deep-water circulation and poor runoff in the aquifers of the Yan’ an Formation are conducive to the enrichment of coalbed methane. The genesis of coalbed methane in the minefield belongs to secondary biological genesis, and the coalbed methane accumulation model belongs to the typical late-stage biogas accumulation model on the gentle slope of the basin margin. The two main factors related to gas content, coal thickness and mudstone thickness of Yan’ an Formation, were screened out according to the principal component method. And then using the gas content predicted by the support vector machine, the prediction result shows that the R² value reaches 0.795, and the mean square error is 0.02. It is confirmed that the penalty parameter C, the insensitive parameter ε, and the kernel function parameter γ are respectively 32,0.01, 1.

(3) The main control factors for productivity of Dafosi CBM vertical wells and horizontal wells are different. The main control factors for productivity of vertical wells are as follows: coal thickness, burial depth, permeability, ash content, roof thickness, gas content, reservoir pressure ; The main controlling factors of horizontal well productivity from strong to weak are: gas content, coal thickness, reservoir pressure, ash content, permeability, roof thickness, and buried depth. The two high production modes of coalbed methane in Dafosi mine field are the high production mode at the axial part of the nose structure and the high production mode at the monoclinic up-dip high part.

(4) Using the entropy weight method and the classification method of natural discontinuities, the prospects of coalbed methane development in the wellfield are divided into five categories: unfavorable area, medium area, more favorable area, favorable area, extremely favorable area. According to Pearson's correlation coefficient and significance test analysis, 8 parameters (gas content, ash content, net thickness of coal seam, structure location, roof thickness, permeability, reservoir pressure, burial depth) are selected for model building. The random forest algorithm is also used to divide the CBM development prospects into 5 categories. The prediction results of the two models are basically consistent, but the random forest algorithm is more detailed in the division of the favorable areas in the west. Judging from the overall spatial distribution of the two prediction analysis results, the evaluation results of the central and eastern parts of the minefield are basically consistent with the current mining results. The southeastern part of the minefield is not suitable for subsequent deployment of well locations, and the western part of the minefield has high favorable areas, and subsequent development and deployment of well locations can be emphatically considered.