论文部分内容阅读
为更好地预测预防煤矿水害,遏制煤矿水害,针对煤层底板突水问题的非线性、小样本特点,通过研究支持向量回归机(SVR)原理,建立基于SVR的矿井底板突水量预测模型。从突水影响因素中选取属性特征,包括水压、含水层厚度、隔水层厚度、底板采动裂隙带深度和断层落差。用网格搜索法和5-折交叉验证法确定径向基核函数(RBF)及模型参数。运用模型对测试样本进行突水量预测。最后将构建的矿井突水SVR模型运用到国内某典型矿山的煤层底板突水预测中,对工作面煤层底板进行最大突水量预测。
In order to predict the prevention of water damage in coal mines and to prevent water damage in coal mines, aiming at the characteristics of nonlinear and small samples of water inrush from coal seam floor, a SVR-based forecasting model of water inrush from mined floors was established by studying the principle of Support Vector Regression (SVR). Attributes were selected from water inrush factors, including water pressure, thickness of aquifers, thickness of aquitards, depth of mining floor cracks and fault drop. Radial basis function (RBF) and model parameters were determined by grid search method and 5-fold cross-validation. Applying the model to predict the inrush of test samples. Finally, the constructed mine water inrush SVR model is applied to the coal floor water inrush prediction of a typical mine in China, and the maximum water inrush of the coal seam floor in the working face is predicted.