论文部分内容阅读
在立井冻结施工过程中,及时了解不同深度、不同岩性地层下冻结壁交圈时间及其形成特性是实现科学施工的前提和基础。针对现有立井冻结设计存在的客观问题,应用神经网络系统理论,合理确定输入参数及输出参数,在学习训练的基础上,建立了胡家河矿井筒冻结施工信息的神经网络预测模型。采用该模型分别对主井、副井及风井冻结壁的交圈时间、内外侧扩展范围、平均扩展速度、有效厚度、井帮温度、荒径范围、平均温度等特性参数进行了工程预报,并对实测数据及预测数据进行了对比分析。结果表明:现场实测结果与预测结果基本吻合,预测模型准确度高,适用性广,为科学设计立井施工方法及其支护方案提供了理论依据。
In the process of shaft freezing construction, it is necessary to keep abreast of the different depths and the time of formation of the frozen circle in different lithology stratums and the forming characteristics of them. This is the precondition and foundation for the scientific construction. In view of the objective problems existing in the existing shaft freezing design, neural network system theory is applied to reasonably determine the input parameters and output parameters. On the basis of learning and training, a neural network prediction model of mine shaft freezing construction information is established. This model is used to forecast the parameters of the circle, the extension of the inner and outer sides, the average expansion velocity, the effective thickness, the wellbore temperature, the wasteland range and the average temperature of the main shaft, And the measured data and forecast data were analyzed comparatively. The results show that the field test results are in good agreement with the prediction results, and the prediction models are of high accuracy and wide applicability. This provides a theoretical basis for the scientific design of shaft construction methods and support schemes.