摘要
变形是造成基坑事故的最主要因素,为准确分析变形特性,实现变形的精准动态预测,保证基坑的安全施工。提出一种将经验模态分解(EMD)、粒子群算法(PSO)和极限学习机(ELM)组合的深基坑多维度时变预测模型。经EMD将基坑变形时变序列进行分解,获得多尺度本征模态函数(IMF);采用PSO-ELM对各IMF时变序列进行预测,等权叠加各预测值,得到模型最终预测结果,同时利用PSO-ELM模型对未经处理的时变序列进行预测。以南宁市某深基坑为例,结果表明:经EMD分解的模型预测相对误差为0.22%0.42%,平均相对误差值仅为0.32%;未经EMD分解的模型预测相对误差为0.31%0.75%,平均相对误差值为0.64%,经EMD分解后模型预测精度明显高于未经分解的模型精度,能较好地应用于非平稳时变预测,为深基坑变形预测提供一种新的方法。
Deformation is the most important factor causing foundation pit accidents.For accurate analysis of deformation characteristics,precise and dynamic prediction of deformation,and safe construction of foundation pits,a multi-dimensional time-varying prediction model integrated by empirical mode decomposition(EMD),particle swarm optimization(PSO)and extreme learning machine(ELM)for deep foundation pits is proposed.The time-varying sequence of the foundation pit deformation is decomposed by EMD to obtain a multi-scale intrinsic mode function(IMF).The PSO-ELM is used to predict each IMF time-varying sequence,and the predicted values are superimposed to obtain the final prediction result of the model.The PSO-ELM model is employed to predict the unprocessed time-varying sequence.Taking a deep foundation pit in Nanning as an example,the results show that the relative error of model prediction by EMD decomposition is 0.22%0.42%,the average relative error value is only 0.32%;the relative error of model prediction without EMD decomposition is 0.31%0.75%The average relative error value is 0.64%.The model prediction accuracy is significantly higher than that of the undecomposed model after EMD decomposition.It can be applied to non-stationary time-varying prediction and provide a new prediction for deep foundation pit deformation prediction.method.
作者
王景春
宋培林
王炳华
何旭升
WANG Jingchun;SONG Peilin;WANG Binghua;HE Xusheng(School of Civil Engineering,Shijiazhuang Railway University,Shijiazhuang 050043,China;Nanning Rail Transit Group Co.,Ltd.,Nanning 530029,China)
出处
《铁道标准设计》
北大核心
2020年第9期103-108,共6页
Railway Standard Design
基金
河北省重点研发计划项目(19275410D)
国家自然科学青年基金项目(51609138)
南宁轨道交通集团有限责任公司专项研究项目。