采煤工作面瓦斯涌出量的固有模态SVM建模预测

Prediction of gas emission from coalface by intrinsic mode SVM modeling

首先提出了一个依据EMD(empirical mode decomposition)方法提取固有模态分量进行SVM建模实现采煤工作面瓦斯涌出量预测的技术方法.利用瓦斯涌出量的历史记录数据,通过EMD分解得出其固有模态函数,即IMF分量,然后,对应于每个固有模态分别利用SVM函数拟合方法进行外推预测,再把不同固有模态的预测结果进行叠加重构合成,获得瓦斯涌出量的理论预测结果.从监测结果的实例分析发现,与常规SVM方法相比,EMD方法的引入能够大幅度提高理论模型的预测精度,并给出监测数据极为吻合的预测结果.实际应用表明,在采煤工作面瓦斯涌出量预测建模中,固有模态的提取和SVM方法的实施都充分利用了样本数据本身驱动的自适应性质,从而为保障优异的预测效果提供了良好的理论基础.

A technique to predict the gas emission from the coalface is presented which is realized by the intrinsic mode SVM modeling on the basis of the intrinsic mode components drawn out from the observed dada by means of the EMD （empirical mode decomposition） method. The intrinsic mode functions, that is, IMF components, are obtained by the EMD analysis of the historical recording dada of gas emission, and after the prediction of each intrinsic mode is carried out by the extrapolation of its regression function determined by the SVM function regression approach, then the prediction result of gas emission is derived through the reconstruction summing all prediction results corresponding to different intrinsic modes. Prom an application example related to the monitoring data it can be seen that the introduction of EMD method into the theoretical modeling to predict the gas emission from the coalface obviously improves the accuracy in comparison with the conventional SVM method, to have the prediction results agreement with the monitoring data. The theoretical analysis shows that in modeling the gas emission from the coalface, the extraction of intrinsic modes and the operation of SVM method make full use of the sampling data driven adaptive performances, and hence provide better theoretical fundamentals for guarding perfect prediction efficiency.