基于PSO-LSSVM模型的基坑变形时间序列预测.pdfVIP

2015 年5 月 控制 工 程 May 2015 第 22 卷 3 期 Control Engineering ofChina Vo 1.22,No.3 文章编号: 1671-7848(2015)03-0475-06 DOI: 1O.14107/j.c础i.kzgc.l40019 基于PSQ-LSSVM 模型的基坑变形时间序列预测 曹净，丁文云，赵党书，宋志刚，刘海明 (昆明理工大学土木工程学院，昆明白05∞) 摘 要: 现场量测获得的基坑变形资料蕴含了系统内部力学演化信息。针对基坑变形影响 因素的复杂性、监测数据的高度非线性以及人工神经网络方法的过学习问题，利用粒子群 (PSO) 算法优选最小二乘支持向量机(LSSVM)参数，并结合相空间重构理论进行数据预处 理，提出了一种基于PSO-LSSVM 模型的基坑变形时间序列预测方法。利用该方法建立基坑 变形预测模型应用于动态设计和信息化施工，对保证基坑安全具有重要意义。将该方法用于 昆明某基坑工程的深层水平位移预测，不断利用基坑前期工况的最新实测数据建模，对后期 工况变形量进行滚动预测， 获得了令人满意的效果。 关键词: 基坑变形，时间序列预测，最小二乘支持向量机，粒子样优化算法，非目空间重构 中图分类号: TU463 文献标识码:A Time Series Forecast of Fouodatioo Pit Deformatioo Based 00 PSO-LSSVM CAOJing, DING Wen-yun, Zι40Dang-shu, SONG Zhi-gang, LIU Hai-ming σaculty ofCivil Engineering and Mechanics ,Kunming University of Science and technology, Kunming 650500, China) Abstract: The deform data of foundation pits 也quired by field m四surement contains intemal m饵hanics evolutionary infonnation. Aiming at 也e complex.ity ofinfluential 如ωrs of foundation pit defonnation,the highly nonlinear monitoring data and extra learning problem of the ar世ficial neural network, a time series forecast method of foundation pit defonnation based on PSO-LSSVM model is proposed. Wherein,the parameter of least 吨uares support v民tor machine is optimized by particle sw缸mand the phase spa臼reconstruction thωryis 因ωpr叩roc臼s data. Through anal归ing 世le m臼sured data, the defonnation forecast model is 臼tablish时 and appliedω 也e dynarnic design and infonnation construction, which is of great significance to ensure the safety of foundation pit. This method has b臼nsu∞出sfully applied in the