论文部分内容阅读
针对丁集矿井壁冻结压力进行不同监测水平的现场实测,发现冻结压力随时间和环境而变化,受层位深度、岩土含水率、冻结壁平均厚度和平均温度等因素的影响,具有明显的不确定性,以变异系数表征其不确定程度。在此基础上优化传统的RBF神经网络,把变异系数引入模糊中心值和权值学习策略中,建立深厚冲积层井壁冻结压力预测模型。该模型以层位深度、含水率、冻结壁平均厚度和平均温度为输入信息量,区分黏土层与钙质黏土层,采用两淮地区7只井筒33个监测水平的样本数据进行训练学习,最后通过口孜东矿井壁冻结压力预测分析进行模型验证。结果表明:现场实测值与预测值拟合度好,模型算法高效,精度合理,为两淮地区深厚冲积层立井冻结压力的分析与预测提供可靠依据。
According to the field monitoring of different monitoring levels of the freezing pressure in the shaft wall of Dingji Mine, it is found that the freezing pressure changes with the time and the environment. Affected by the factors such as the depth of the layer, the moisture content of the soil, the average thickness of the frozen wall and the average temperature, Uncertainty, the coefficient of variation to characterize the degree of uncertainty. On the basis of this, the traditional RBF neural network is optimized, and the coefficient of variation is introduced into the fuzzy center value and weight learning strategy to establish a wellbore pressure prediction model for deep alluvium. In this model, the layer depth, moisture content, the average thickness of freezing wall and the average temperature are used as the input information to distinguish the clay layer from the calcareous clay layer. The data of 33 monitoring levels from 7 wells in the two Huaihe areas are used for training and learning. Finally, The model verification is carried out by predicting and analyzing the frozen pressure of the borehole in the mouth of Zitong. The results show that the measured values of field and predicted values fit well, the model algorithm is efficient and the accuracy is reasonable, which provides a reliable basis for the analysis and prediction of the freezing pressure in deep alluvial-layered wells in the Huaihe area.