| 基于神经网络的坑道近水平定向孔轨迹预测研究 |
| 投稿时间：2023-08-27 修订日期：2023-11-22 点此下载全文 |
| 引用本文： |
| 摘要点击次数: 232 |
| 全文下载次数: 0 |
|
| 基金项目:陕西省自然科学基础研究计划“融合钻柱模型参数不确定性的近水平钻进智能控制系统” |
|
| 中文摘要:针对当前近水平定向钻孔轨迹测量位置滞后于钻头位置，无法实时获取滞后区钻孔轨迹实际参数，施工过程中控制轨迹时需要人工预测该部分轨迹并为下一步轨迹调整提供依据。为了降低人为因素影响，提高轨迹预测的准确性。基于BP神经网络建立了用于煤矿井下近水平定向孔轨迹控制的孔底空间参数预测模型。选取随钻测量仪器位置及其之前12m范围的倾角、方位角等13个钻孔空间和轨迹控制参数，经过变换后作为输入参数，构建了一个具有11个输入参数和2个输出参数的4层BP神经网络预测模型，该模型以不同矿区的6个钻孔502组数据为训练样本，得到了网络预测模型参数，并将12组测试数据的预测结果与24名从业技术人员的经验预测结果进行了对比分析。研究结果表明：采用logsig激活函数和（9×6）节点的双隐含层BP神经网络模型，对孔底空间参数（倾角、方位角）的预测绝对误差平均值分别达到0.51°和0.68°，且预测误差服从正态分布，预测结果绝对误差平均值较从业5年以上的技术人员低了35%，现场应用效果较好，满足煤矿井下定向钻进轨迹控制的需要，并为智能定向钻轨迹智能控制提供了理论与实践基础。 |
| 中文关键词:神经网络 坑道钻探 近水平定向孔 轨迹预测 预测模型 |
| |
| Study of the predicting Model for Directional Drilling Path controlling based on Neural Network in Coal Mine |
|
|
| Abstract:As the current measurement position of near-horizontal directional drilling trajectory lags behind the bit position, the actual parameters of drilling trajectory in the lag area cannot be obtained in real time, and the trajectory control in the construction process needs to manually predict this part of the trajectory and provide a basis for the next trajectory adjustment. In order to reduce the influence of human factors and improve the accuracy of trajectory prediction.A forecasting model is established based on BP neural network that used for controlling underground directional drilling path in coal mine: The model is a four-layer BP neural network, and it chooses 11 input parameters which are changed from path parameters (dip angles and azimuths from 12m before MWD) and control parameters: Then 2 parameters including the dip angle and azimuth at bit are output: 502 groups training data of 6 boreholes from different mining areas are used for training the network model: Then the forecasting result of 12 groups test data are compared with prediction results of artificial experience from 24 technicians: The study result shows that: the mean absolute error of the dip angle and azimuth at bit are only 0.51° and 0.68° predicted by the forecasting model based on BP neural network, that uses logsig activation function and has a double-hidden-layer which has point structure of 9×6, and the prediction error obeys normal distribution: Compared with technicians who work more than 5 years, the accuracy prediction results from BP neural network model is 33.9% lower and meets the needs of drilling path control. |
| keywords:Neural network Tunnel drilling Near-horizontal borehole Drilling Path control Forecasting model |
| 查看/发表评论 下载PDF阅读器 |
