论文部分内容阅读
在大量实测数据的基础上探究了经验公式法的局限性并基于理论层面分析了BP神经网格法局限性。探究了支持向量机计算导水裂隙带发育高度的优势,根据43个收集到的顶板裂隙带发育高度实测数据,依照参数获取简便,与导水裂隙带发育高度相关性相对较高的原则,以采高、顶板覆岩强度以及开采方式作为特征因子,建立导水裂隙带发育高度回归模型。利用模型预测了谢桥1121工作面及白庄7507工作面导水裂隙带发育高度。结果表明:运用SVM回归模型预测导水裂隙带发育高度具有较好的可行性。
On the basis of a large number of measured data, the limitations of the empirical formula method are explored and the limitations of the BP neural network method are analyzed based on the theory. Based on the 43 collected height data of roof fracture zone development, according to the principle of simple parameter acquisition and relatively high correlation with the development of water-conducted fracture zone, Mining height, roof overburden strength and mining methods as the characteristic factors, a regression model was established for the developmental height of water-bearing fractured zone. The model was used to predict the developmental height of hydraulic fractured zone in Xieqiao 1121 working face and Baizhuang 7507 working face. The results show that it is feasible to predict the developmental height of aqueduct with SVM regression model.