摘要
为实现基坑沉降变形的高精度预测,在总结传统组合预测模型不足基础上,先进行单项预测模型筛选,并对各单项预测模型进行参数优化处理;其次,进一步进行非线性组合及节点间距离的优化处理和考虑相邻节点的优化处理,以实现基坑沉降变形的优化组合预测。结果表明:ELM模型、GM(1,1)模型及SVR模型具有很好的互补性,可将其作为基础预测模型,再通过在沉降变形预测过程中的优化组合处理,所得预测结果的平均相对误差介于1.56%~1.65%,具较高的预测精度及稳健性,并将预测结果与传统模型预测结果比较,得出预测模型具有相对更优的预测精度,验证其构建思路是有效的,适用于基坑沉降变形预测。通过研究,以期为基坑变形规律研究提供理论指导。
In order to achieve high-precision prediction of foundation pit settlement and deformation,based on the summary of the shortcomings of traditional combination prediction models,a single prediction model screening is first carried out,and the parameters of each single prediction model are optimized;Secondly,further optimization of nonlinear combination and distance between nodes,as well as consideration of optimization of adjacent nodes,is carried out to achieve optimized combination prediction of foundation pit settlement deformation.Example analysis shows that:The ELM model,GM(1,1)model,and SVR model have good complementarity and can be used as basic prediction models.Through optimized combination processing in the settlement deformation prediction process,the average relative error of the predicted results ranges from 1.56%to 1.65%,with high prediction accuracy and robustness.The prediction results are compared with those of traditional models,and it is concluded that the prediction model has relatively better prediction accuracy,which verifies the effectiveness of its construction idea and is suitable for predicting foundation pit settlement deformation.Through research,we aim to provide theoretical guidance for the study of deformation laws in foundation pits.
作者
白霖
毛红梅
王安东
李斌
BAI Lin;MAO Hongmei;WANG Andong;LI Bin(Shaanxi Railway Institute,Weinan 714000,China)
出处
《粉煤灰综合利用》
2025年第3期122-127,共6页
Fly Ash Comprehensive Utilization
关键词
基坑
沉降变形
单项预测模型
优化处理
组合预测
foundation pit
settlement deformation
single prediction model
optimization treatment
combination prediction
