多属性融合分布式光纤导水裂隙带高度预测方法

Prediction of water⁃flowing height in fractured zone based on distributed optical fiber and multi⁃attribute fusion

针对上覆岩层导水裂隙带发育高度预测这个研究难点,以光纤监测覆岩变形模拟试验为基础,融合测点频移值、高度、覆岩岩性等因素,提出一种基于深度学习的导水裂隙带发育高度预测方法。先将测点频移值分解为随机、周期和趋势等分量,建立预测模型,叠加后再利用预测频移值变化特征推测导水裂隙带的发育高度。实验结果表明本方法预测的发育高度平均误差为11 mm,其精度控制在允许误差范围内。

The prediction height of water-flowing in fractured zone of overburden strata is the difficulty of research.A deep learning model is proposed to estimate the height of the water-conducting fracture zone based on simulation experiments of fiber-optic monitoring overlying rock deformation.The key factors,such as the frequency shift value of the optical fiber sensor,the height of the sensor,the advance length of the working face,the lithology and structure of the overlying rock were select in this model.Firstly,the frequency shift value is decomposed into three components,random frequency shift value,periodic frequency shift value and trend frequency shift value.Then,deep learning model are built to predict each component,and prediction component are superimposed to form the frequency shift value of sensor.Finally,the development height of the water-conducting fracture zone is inferred in accordance with the relationship between curve of frequency shift value change and the development height of fracture zone.The experiments show that the average estimation error of such height is about 11 mm.The predicted heights are accurate and reliable,which can also meet requirements of an engineering test practice.