基于特征选择的煤层瓦斯含量影响因素分析

A Factor Analysis of Gas Content in Coal Seam Based on Feature Selection

煤层瓦斯含量受到多种条件的影响。研究者用数学模型来表示各物理化学参量对煤层中瓦斯含量的影响程度,多数模型中最重要的参量是煤层的埋深,但一些模型也发现煤层的理化性质和顶底板岩性的重要性。文章采用随机森林和Lasso回归等数据挖掘技术进行了特征选择,对不同因素对煤层瓦斯含量影响程度进行了比较。结果显示影响瓦斯含量各因素的重要性有所差别,最重要的因素是基岩厚度和埋深。同时对于特征选择的方法,提出了建议。

The gas content in coal seam is controlled or impacted by several factors, and the model may be different in different coal mine area. A similar pattern can be found that the depth of coal seam and thickness of overlying rock are the key factors that impact coal gas content in most models. The other factors show impacts in some models, including physical and chemical characteristics, lithology of roof and floor. The impact factors in different coal mine districts were analysis using method of the Random Forest, and the Lasso Regression. The most important parameters that impact gas content were found to be thickness and depth of coal seam. At the same time, the methods usually used for feature selection were discussed.