基于K-CV&SVM的工作面煤层瓦斯含量预测

Prediction for Seam Gas Content of Coal Face Based on Cross-Validation & SVM

为进一步提高工作面煤层瓦斯含量预测的准确性,将交叉验证方法(K-CV)和支持向量机(SVM)相结合,建立预测模型。该模型在SVM的基础上采用交叉验证的思想,寻找最佳参数c&g,最大限度地消除由于个别样本的较大误差对预测模型的影响,提高预测模型的准确性。选取告成矿工作面煤层钻孔的实测数据进行实例分析,结果表明:该模型较单一SVM预测精度高,能有效预测工作面煤层瓦斯含量。

In order to further improve prediction coal seam gas content of coal face, the prediction model is established by K- fold cross-validation （K-CV） and support vector machine （SVM）. This model uses the method of cross-validation on the basis of SVM to find the best parameter c ＆ g. It can eliminate the influence on the prediction model of large errors of indi- vidual samples and improve the accuracy of the prediction model. By analyzing the drilling data from mining working face of Gaocheng coal mine,the results showed that the model has higher prediction accuracy than the model based on SVM. The model can effectively predict the working face of coal seam gas content.