高速铁路路基沉降预测分析与软件应用

Prediction Analysis and Software Application of High-speed Railway Subgrade Settlement

引入数据预处理理论保障原始监测数据集的质量,采用SVR算法针对预处理后的沉降数据进行精准预测,进一步整合开发了一种高速铁路路基沉降预测分析软件,并结合现场实例进行了验证分析。结果表明:高速铁路路基沉降精准预测软件不仅可提供数据的预处理操作,还可提供高精度的预测分析功能;软件内嵌了数据预处理的算法,可对原始监测数据集进行缺失值的处理和预先过滤,为后续高速铁路路基沉降预测分析提供了数据质量保证;软件内嵌的SVR算法很好地反映了路基沉降的走势与上下波动,所得的沉降实测值与预测值的相关性极高,两测点降噪后的R^(2)分别为0.904、0.898,而MSE分别为0.025、0.031,体现出较高的泛化能力,整体预测精度高、预测误差小。

In order to integrate the development of high-speed railway subgrade settlement prediction software,the data preprocessing theory is introduced to guarantee the quality of the original monitoring data set,and then SVR algorithm is used to accurately predict the preprocessed settlement data,further integrating and developing a kind of high-speed railway subgrade settlement prediction and analysis software.The field example is combined to verify the analysis.The results show that the accurate prediction software of high-speed railway subgrade settlement can not only provide data preprocessing operation,but also provide high-precision prediction and analysis function.Data preprocessing algorithm is embedded in the software,which can process and pre-filter the missing values of the original monitoring data set,and provide data quality assurance for the subsequent prediction and analysis of high-speed railway subgrade settlement.The SVR algorithm embedded in the software can well reflect the trend and fluctuation of subgrade settlement,and the measured settlement value is highly correlated with the predicted settlement value.The R^(2)values at two measurement points after noise reduction are 0.904 and 0.898 respectively,and the MSE values are 0.025 and 0.031 respectively,which reflect the greater capability of generalization.The overall prediction accuracy is high and the prediction error is small.