煤层气储层渗透率的预测方法  

Prediction method of reservoir permeability of coal bed methane

作　　者：边莉[1] 张欣欣[2] 李婷婷[2] 车向前[3] Bian Li Zhang Xinxin Li Tingting Che Xiangqian(School of Electronics ＆ Information Engineering, Heilongjiang University of Science ＆ Technology, Harbin 150022, China School of Electrical ＆ Control Engineering, Heilongjiang University of Science ＆ Technology, Harbin 150022, China School of Computer ＆ Information Engineering, Heilongjiang University of Science ＆ Technology, Harbin 150022, China)

机构地区：[1]黑龙江科技大学电子与信息工程学院,哈尔滨150022 [2]黑龙江科技大学电气与控制工程学院,哈尔滨150022 [3]黑龙江科技大学计算机与信息工程学院,哈尔滨150022

出　　处：《黑龙江科技大学学报》2016年第5期480-484,共5页Journal of Heilongjiang University of Science And Technology

基　　金：国家自然科学基金项目(51504085);哈尔滨市科技局项目(2015RQQXJ009)

摘　　要：煤层气储层渗透率参数的强非线性,使得常规参数难以预测,由此提出一种交叉熵优化支持向量机的煤层气渗透率预测新方法。利用UCI数据库中的数据,将该算法预测结果与粒子群算法、遗传算法的预测结果进行对比。结合某矿区实际地质特性,用该算法对该矿区煤层气储层渗透率参数进行预测,得出大部分被测试数据的相对误差率达到1%左右,除了个别数据超过20%,总体分布较稳定。该矿区渗透率参数的预测值和真实值分布图大体一致。该算法可行,且优于其他算法,可以为煤层气储层参数的预测提供算法保障。This paper presents a novel method tailored for coal bed gas permeability,based on cross entropy optimization support vector machine to address the notorious prediction of conventional parameters due to the stronger nonlinearity associated with the permeability parameters behind coalbed gas reservoir.The algorithm is validated by using the data in UCI database,and then comparing the prediction results with the those of particle swarm algorithm and genetic one,suggesting its feasibility and consequently distinct advantages over other ones. The prediction of the coalbed gas reservoir permeability in a mining area with the actual geological characteristics,using the proposed algorithm,indicates that the algorithm operates within the relative error of about 1% in the data measurement except more than 20% in some individual cases,with a relatively stable distribution. The general consistency between the predicted value of permeability parameter and the distribution map of true value of the mining area points to the conclusion that the algorithm can provide a safe prediction of the parameters behind coal bed gas reservoir.

关 键 词：煤层气 渗透率预测 交叉熵算法 支持向量机 

分 类 号：TD713.2[矿业工程—矿井通风与安全]