BP神经网络在某深基坑工程监测中的应用研究

采用BP神经网络模型对某深基坑工程的变形进行了预测,通过与实测数据的对比分析表明:BP神经网络模型在深基坑工程监测中具有良好的容错性,在围护桩顶竖向及水平位移、围护桩深层水平位移、锚索内力预测方面具有较高的准确度,与实际内力及变形情况较为吻合;BP神经网络模型在地下水位预测中的准确度较低、数据结果离散性较大,不能充分反映水位变化的实际情况;BP神经网络模型在进行短期预测时准确率高,预测曲线与实测数据曲线吻合性较好;在进行锚索内力监测时,当传感器不能及时采集数据的情况下,BP神经网络能发挥预测准确数据的作用;BP神经网络模型隐含层数的选取对预测结果的准确性有一定的影响。

Application of BP Neural Network in a Deep Pit Monitoring

Predicted the deformation of a deep pit by using the BP neural network model, it shows that by compare the measured data with the prediction data analysis： the BP neural network model has strong fault-tolerance in deep pit monitoring,and has higher accuracy in the vertical and horizontal displacement of enclosure pile top, deep hol-izontal displacement of enclosure pile and anchor force prediction. The actual internal force is tally with the deformation. The BP neural network model in the prediction of groundwater level have low precision and high discrete, can not reflect the change of water level; BP neural network model can play a role in predicting the data accurately in the internal force monitoring when the sensor cannot collecting the data on time; The selection of the hidden layers of BP neural network model has certain affection for the accuracy of prediction results.