一种基于支持向量机和模型树的回归模型及其在采煤工作面瓦斯涌出量预测中的应用

A SVM and Model Tree Based Regression Model and Its Application in Predicting the Amount of Gas Emitted from Coalface

矿井瓦斯是煤矿生产过程中的主要不安全因素,能否准确预测采煤工作面上的瓦斯涌出量将直接影响矿井开采的经济技术指标.从数据挖掘与机器学习的角度看,瓦斯涌出量的预测问题是回归分析的经典应用.支持向量机和模型树在回归分析方法中显示出了优越的性能,本文应用支持向量机和模型树方法建立采煤工作面瓦斯涌出量的预测模型.实验结果显示,预测精度满足要求,是两种可行的、合理的预测方法.受此启发,本文提出了一种基于支持向量机和模型树的组合回归模型,并将其用于瓦斯涌出量的预测,实验证明组合模型的预测性能比单个的回归模型都要好.

The mine gas is one of the most unsafe factors in coal mine production.Predicting gas emission directly affects the economic and technological indicators of the mining process.From the data mining and machine learning point of view,it is a typical application of regression analysis.Support vector machine（SVM）and model tree have already demonstrated a superior performance among different regression models.This paper applies them to build prediction models for the amount of gas emitted from coalface.The experimental results show that their precisions suffice the practical use,and both are feasible and reasonable prediction methods.Inspired by their success,we single out a combined regression model based on support vector machine and model tree,and use it to predict the gas emission.The experiments show that the combined model significantly outperforms single regression models.