边坡位移的EEMD-PSO-ELM模型预测方法  

Slope displacement prediction method based on EEMD-PSO-ELM model

作　　者：谢博 施富强[1,2] 廖学燕[2] 马胜 杨伟 路祥祥 XIE Bo;SHI Fuqiang;LIAO Xueyan;MA Sheng;YANG Wei;LU Xiangxiang(School of Mechanical Engineering,Southwest Jiaotong University,Chengdu Sichuan 610031,China;Sichuan Academy of Safety Science and Technology,Chengdu Sichuan 610045,China)

机构地区：[1]西南交通大学机械工程学院,四川成都610031 [2]四川省安全科学技术研究院,四川成都610045

出　　处：《中国安全科学学报》2020年第3期157-162,共6页China Safety Science Journal

基　　金：四川省省级科研院所基本科研业务费项目(2018YSKY0038);四川省科技计划项目(〔2016〕8号)。

摘　　要：为解决边坡变形位移预测难度大的问题,利用北斗监测系统获取边坡位移数据,引入集合经验模态分解(EEMD)法、粒子群优化(PSO)和极限学习机(ELM),建立边坡位移预测的EEMDPSO-ELM模型;以攀钢集团石灰石矿5号监测点为例,对原始数据小波去噪,采用EEMD法将位移时间序列分解为波动项位移和趋势项位移;利用PSO-ELM优化模型预测下一时段位移,叠加2项位移预测结果,得到边坡累计位移预测值,并对比分析预测结果。结果表明:EEMD-PSO-ELM模型位移预测方法的平均相对误差(MRE)为0. 15%,均方根误差(RMSE)为0. 03,拟合优度为0. 999 9,该模型具有一定的精确性和适用性。In order to solve the difficulty in predicting slope deformation and displacement,Beidou monitoring system was used to obtain slope displacement data,and EEMD method,PSO and ELM were introduced to build an EEMD-PSO-ELM model for displacement prediction. Then,with No. 5 monitoring point of Pangang Group limestone mine as an example,original data was denoised wavelet by,and displacement time series were decomposed into fluctuating displacement and trending displacement by EEMD method. Displacement in next period was predicted using PSO-ELM optimization model. Finally,cumulative displacement prediction of slope was obtained by combining the two results,and they were compared and analyzed. The research shows that the mean relative error( MRE),root mean square error( RMSE) and goodness of fit of EEMD-PSO-ELM model are 0. 15%,0. 03 and 0. 9999 respectively,indicating the model has certain accuracy and applicability.

关 键 词：矿山边坡位移预测 集合经验模态分解(EEMD)法 粒子群优化(PSO) 极限学习机(ELM) 小波去噪 

分 类 号：X928.03[环境科学与工程—安全科学]