SSA-小波神经网络支持下的地铁沉降变形预测

Prediction of subway settlement and deformation supported by SSA wvelet neural network

沉降变形监测中,研究如何对监测的沉降数据进行处理,预测沉降量,对可能出现的安全隐患做出预判有着很重要的实际意义。本文基于神经网络模型、小波分析和奇异谱分析(singular spectrum analysis,SSA)的相关理论,构建起SSA-小波神经网络变形预测模型,并将模型应用于地铁工程沉降预测中。针对地铁监测数据非平稳性、非线性特征,首先,使用SSA方法提取数据序列中的趋势项与周期项,提高序列信噪比;其次,使用小波神经网络模型对趋势项与周期项分别进行预测与重构,得到最终的预测值。通过对地铁累计沉降量观测数据进行预测,结果表明相比单独的小波神经网络模型,SSA-小波神经网络模型的预测效果更佳稳定,且随着训练样本的增加,预测结果与实际情况更加符合。

In settlement deformation monitoring,it is of great practical significance to study how to process the monitored settlement data,predict the settlement,and predict the potential safety hazards.Based on the relevant theories of neural network model,wavelet analysis and singular spectrum analysis(SSA),this paper constructed the deformation prediction model of SSA wavelet neural network,and applied the model to the settlement prediction of subway engineering.According to the nonstationary and nonlinear characteristics of subway monitoring data,firstly,the trend term and periodic term in the data series were extracted by SSA method to improve the signal-to-noise ratio of the series;Secondly,the wavelet neural network model was explored to predict and reconstruct the trend term and periodic term respectively to obtain the final prediction value.Through the prediction of the observed data of subway cumulative settlement,the results showed that compared with the single wavelet neural network model,the prediction effect of SSA wavelet neural network model was better and stable,and the prediction results were more consistent with the actual situation with the increase of training samples.