论文部分内容阅读
神经网络预测为深基坑预测提供了一种有效的路径。运用哪种模型较优,输入层、输出层、隐含层参数如何选取,对预测的结果都有一定的影响,本文结合实际轨道交通工程案例,以深基坑沉降监测数据为例,对常见的几种神经网络预测模型进行了对比分析,对几种模型的残差、均方根误差(RMSE)和绝对平均误差(MAE),收敛次数这几个方面进行对比,结果表明遗传算法神经网络对深基坑沉降监测数据预测较为有效,同时对模型参数的选取提出了建议。
Neural network prediction provides an effective way for deep foundation pit prediction. Which model is better, how to choose the input layer, output layer and hidden layer parameters have certain influence on the prediction results. Combined with the actual case of rail transit project, taking the settlement monitoring data of deep foundation pit as an example, Several neural network prediction models were compared and analyzed. The residuals, root mean square error (RMSE), absolute average error (MAE) and convergence times of several models were compared. The results show that genetic algorithm neural network The prediction of settlement monitoring data of deep foundation pit is more effective, and some suggestions are put forward for the selection of model parameters.