施工技术2017年4月下 16CONSTRUCTIONTECHNOLOCY第46卷第8期 DOI:10. 7672/sgjs2017080016 基于GA-SVM的基坑施工地面沉降 时间序列预测研究*
石祥锋，王丽芬²，沈阳，刘俊娥”，郭章林 （1.华北科技学院土木工程系，北京101601；2.河北工程大学经济管理工程与商学院， 河北邯郸056038；3.北京物资学院信息学院，北京101149) [摘要】基坑施工引发的工程性地面沉降的因素众多，地面沉降数据具有高度的非线性，高精度的地面沉降预测具 有一定的困难。
利用三次样条插值法对观测数据进行预处理，运用遗传算法（GA)寻找支持向量机(SVM）的最优 参数并建立GA-SVM时间序列预测模型，应用于基坑工程信息化施工和动态设计。
用GA-SVM时间序列预测模型 对广州某地铁站基坑工程的地面沉降数据进行预测，并与实测值相和其他预测方法相比较，该模型有明显的优越 性，对基坑施工安全性的提高具有重大的意义。
[关键词】基坑；施工；沉降；三次样条插值法；遗传算法；预测；研究 [中图分类号】TU472.1[文献标识码】A[文章编号】1002-8498(2017）08-0016-04 ResearchonTimeSeriesPredictionofFoundationExcavation ConstructionLandSettlementBasedontheGA-SVM SHI Xiangfeng',WANG Lifen²,SHEN Yang',LIU Jun'e’,GUOZhanglin' (1.Department of Civil Engineering,North China Institute of Science and Technology， Beijing 101601,China; 2. School of Economics Management and Business,Hebei University of Engineering,Handan,Hebei 056038,China;3. Beijing Wuzi University,School of Information,Beijing101149,China) Abstract;Physical land settlement caused by foundation excavation construction has different factors， ground settlement data is highly nonlinear,so high precision ground settement prediction has certain difficulty. This paper processes the observed data by using the cubic spline interpolation method. The genetic algorithm (GA） is used to find the optimal parameters of support vector machine (SVM） and GA - SVM time series prediction model is established,and these methods which are put in the foundation model of a subway station foundation excavation engineering in Guangzhou is used to predict the ground settlement data，which is pared with the measured values and other prediction methods,this model has obvious superiority,which has great significance for the improvement of safety construction of foundation excavation. Key words;foundations;construction; setlement; cubic spline interpolation method;genetic algorithm; forecast;research 0引言其重要。
地面沉降是基坑稳定性和施工安全的重 地下工程和基坑施工导致的地面沉降对基坑要因素之一，不仅会产生基坑塌的风险而且会引 施工的安全影响程度日益突出，尤其是随着我国...
