基于神经网络的深基坑沉降预测模型比较

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。

Comparison of Models for Settlement Prediction Based on Neural Network for Deep Foundation Pit

Neutral network is a good method for deformation prediction of deep foundation. However, which one is the best model among different kinds of neutral network model, how to choose the parameters for input layer, output layer and hidden layer, also have some impact on the prediction results. Based on the analysis of a certain practical case, we analyzed several common neural network prediction models, compared the residuals, RMSE, MAE and convergence times of several models. It is found that the neural network based on improvement genetic algorithm is most suitable for the prediction of settlement monitoring data. The parameter choice for the model is suggested.