论文部分内容阅读
时频分解方法局部均值分解(local mean decomposition,LMD)在沉降监测中已经得到了应用,但在使用中会出现模态混叠现象。总体局部均值分解(ensemble local mean decomposition,ELMD)通过添加辅助噪声可以抑制局部均值分解过程中出现的模态混叠现象。提出了一种基于ELMD的并联式组合沉降预测方法,结合高速铁路某桥梁实际监测数据,在对ELMD模型进行仿真分析的基础上,分别使用ELMD和LMD将一组离散非线性信号分解为3个PF分量和1个剩余分量,并利用支持向量机和卡尔曼滤波进行预测验证。结果表明:使用ELMD进行分解的过程中能够很好地抑制LMD方法中出现的模态混叠问题。在预报精度方面,基于ELMD的并联式组合模型的平均相对误差可以达到8.3%,可为沉降监测的预报工作提供参考和借鉴。
Time-frequency decomposition method Local mean decomposition (LMD) has been used in settlement monitoring, but modal aliasing can occur during use. The ensemble local mean decomposition (ELMD) can suppress the modal aliasing phenomenon in the local average decomposition by adding the auxiliary noise. Based on the actual monitoring data of a bridge on a high-speed railway, based on the ELMD model simulation analysis, a set of discrete nonlinear signals is decomposed into three by ELMD and LMD respectively PF component and a residual component, and the use of support vector machine and Kalman filter for predictive verification. The results show that the use of ELMD decomposition process can be very good to suppress the modal aliasing in the LMD method. In terms of forecast accuracy, the average relative error of ELMD-based parallel combination model can reach 8.3%, which can provide reference for the prediction of settlement monitoring.