用地球物理测井资料预测煤层气含量——基于斜率关联度—随机森林方法的工作案例

煤层气含量是煤层勘探开发研究的重点参数之一,由于煤层气含量受多因素影响,能有效预测其含量至关重要。本文将斜率关联度法与随机森林算法相结合,以地球物理测井资料为基础进行煤层气含量预测。首先利用改进的斜率关联度法,计算得到对煤层气含量敏感的测井曲线,再利用交叉验证法探究合适的随机森林决策树个数,并结合选出的超参数利用随机森林算法预测煤层气含量。以沁水煤田柿庄北区3号层为例,对该区块进行评价预测,并将预测结果与多元回归模型拟合结果进行对比,同时对本文方法模型的泛化性进行研究分析。结果表明,应用斜率关联度法对测井曲线与煤层气含量进行分析计算能准确有效地找到可用于煤层气含量预测的测井曲线;用随机森林算法训练得到的模型预测非夹矸段煤岩的煤层气含量准确,计算结果可信度高,在夹矸段预测能力较弱,总体对煤层气勘探开发有指导意义,具有实际应用价值。

The exploration of predicting CBM content by geophysical logging data:A case study based on slope correlation random forest method

Coalbed methane content is one of the key parameters in coal seam exploration and development research.Due to the influence of many factors on coalbed methane content,it is very important to predict coalbed methane content effectively.In this paper,slope correlation degree method and random forest algorithm are combined to predict coalbed methane content based on geophysical logging data.Firstly,the improved slope correlation degree method is used to obtain the favorable geophysical logging curves for CBM content prediction,and then the cross validation method is used to explore the appropriate number of random forest decision trees,and the random forest algorithm is used to predict the coalbed methane content for the logging curve sequence with positive correlation.With the No.3 seam in Shizhuang north area of Qinshui coalfield as an example,the block was evaluated and predicted with the results compared with the results of multiple regression model,and the anti-interference ability of the model was studied and analyzed.The results show that the application of slope correlation method to analyzing and calculating the geophysical logging curve and coalbed methane content can accurately and effectively find the logging curve that can be used to predict the content of coalbed methane,the model trained by random forest algorithm is accurate in predicting the content of coalbed methane in the non-gangue section,and the calculation result has high reliability,but the prediction ability is weak in the gangue section.The results obtained by the authors are of guiding significance to the exploration and development of coalbed methane and have practical application value.