EMD降噪在高速铁路路基沉降预测中的应用

高速铁路路基实测沉降曲线通常含有噪声，不宜直接进行曲线拟合。将EMD降噪法引入到沉降预测中，首先对路基沉降观测数据进行滤波降噪，然后将降噪后的沉降曲线作为原始数据进行曲线拟合。结合数学模拟实验和京沪高速铁路典型路基沉降板观测数据，分析了EMD降噪效果和降噪后的沉降预测结果。分析结果表明，EMD降噪很好地消除了混杂在实测沉降曲线中的随机误差和异常的波动成分；降噪后的沉降速率曲线更为平缓，能直观地反映出沉降趋于稳定的时间和过程；降噪后的曲线拟合相关系数高，误差平方和小，沉降预测精确度得到了显著提高且能够满足评估指南的要求。

Application of EMD denoising method in subgrade settlement prediction of high-speed railway

In the presence of noise, most measured subgrade settlement curves of high-speed railway shouldn’t be used to predict settlement directly. In this paper, measured settlement curves of subgrade were filtered by EMD denoising method, and then filtered settlement curves were used to predict settlement. Combining mathematical experiment and measured data of subgrade in Beijing-Shanghai high-speed railway, effect of noise reduction and results of settlement prediction have been analyzed. The analysis shows that random errors and abnormal fluctuation components mixed in measured settlement curves has been eliminated well by EMD denoising method. Settlement rate curves of filtered settlement are smoother, and they can reflect directly when and how settlement curves get into stable situation. Settlement prediction accuracy has been significantly improved with high correlation coefficients and low square errors.