基于EMD和RBF的隧道围岩监测数据分析  

Analysis of monitoring data of wall-rock in expressway tunnel based on EMD method and RBF neural network

作　　者：潘龙[1] 王建国[1] 

机构地区：[1]合肥工业大学土木与水利工程学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2011年第3期395-398,402,共5页Journal of Hefei University of Technology：Natural Science

基　　金：土木工程防灾减灾安徽省工程技术研究中心资助项目(2007368)

摘　　要：文章针对隧道围岩监测数据中含有大量随机误差的问题,基于经验模态分解和奇异谱分析的基本思想,提出了一种对监测数据进行降噪处理的方法,该方法首先将监测时间序列分解为多个固有模态函数,并形成特征矢量矩阵,然后根据该矩阵的奇异谱选择最优的重构阶数对固有模态函数进行重构,从而消除或削弱随机误差的干扰。以某隧道的围岩监测数据为例,用该方法进行降噪处理,并用RBF神经网络对降噪前后的结果进行预测比较;计算结果表明,该降噪方法合理有效,能有效识别噪声和有用信息,适合于隧道围岩的监测数据分析。Considering that many random errors are contained in the monitoring data of tunnel wall-rock, a denoising method for the monitoring data is proposed based on the theory of empirical mode decomposition （EMD） and singular spectrum analysis. The EMD method is used to decompose the time series into some intrinsic mode function components, by which the feature vector matrixes are formed. Then some important intrinsic mode functions are reconstructed by the detem^ined optimal orders according to the result of singular spectrum analysis. As a result, the interference of random error is reduced or eliminated. As an example, the monitoring data of a tunnel wall-rock is denoised by the proposed method. Pressure prediction for the original data and the denoised data are given by using radial basis function（RBF） neural network. The results show that this reasonable method can effectively distinguish the noise and useful information and it is particularly suitable for analyzing the monitoring data of tunnel wall-rock.

关 键 词：经验模态分解 围岩监测 降噪 径向基函数神经网络 

分 类 号：U456.3[建筑科学—桥梁与隧道工程]