基于支持向量回归机的煤层瓦斯含量预测研究

为了对煤层瓦斯含量进行准确预测,应用支持向量回归机(SVR)理论建立煤层瓦斯含量预测模型,结合现场实测数据利用支持向量机(SVM)工具箱进行模型的求解及预测,并从均方根误差、希尔不等系数和平均绝对百分误差3个不同误差指标与人工神经网络预测模型进行比较分析。研究结果表明:SVR模型其预测精度及可行性高于神经网络模型,而且运算快,实时性较好,用于煤层瓦斯含量的预测较理想,具有良好的应用前景,可以为煤矿瓦斯防治提供理论依据。

Study on Prediction of Coal Seam Gas Content Based on Support Vector Regression

In order to accurately predict coal seam gas content,the theory of SVR is applied to establishing the prediction model of coal seam gas content,and SVM toolbox is used to solve the model and prediction with the measured data.From the three different error indicators of root-mean-square error,hill inequality coefficient and mean absolute percent error,a comparison and analysis is made with artificial neural network prediction model.The results show that,the accuracy and feasibility of SVR model prediction is much higher than that of the neural network model,and its computing speed is more satisfactory than the latter in terms of real-time.The model can better forecast the coal seam gas content and has a good prospect of application.It provides a theoretical basis for the prevention and control of coal gas.