| 马 琳.基于ICEEMD-ICA准则进行数据处理的基坑变形组合预测研究[J].地质与勘探,2023,59(5):1074-1082 |
| 基于ICEEMD-ICA准则进行数据处理的基坑变形组合预测研究 |
| Combined prediction of foundation pit deformation based on ICEEMD-ICA criterion for data processing |
| 投稿时间：2022-03-07 修订日期：2023-04-26 |
| DOI：10.12134/j.dzykt.2023.05.013 |
| 中文关键词: 基坑变形 经验模态分解 组合预测 相关向量机 对比验证 |
| 英文关键词: foundation pit deformation, empirical mode decomposition, combined prediction, correlation vector machine, comparative verification |
| 基金项目:杨凌职业技术学院2020年院内基金(传统建筑元素在现代商业建筑设计中的应用研究-以天水市“天麟·名城苑”商业设计为例，编号：ZK20-18）资助 |
|
| 摘要点击次数: 666 |
| 全文下载次数: 245 |
| 中文摘要: |
| 为确保基坑变形的高精度预测，基于基坑现场变形监测成果，先利用ICEEMD-ICA准则进行数据处理，即将变形数据分解为趋势项和随机项，再通过ISFLA-RVM-GRNN模型实现其组合预测。最后，再引入若干传统预测模型进行类似预测，通过预测结果对比，验证本文组合预测思路的合理性。实例分析表明：ICEEMD-ICA准则能实现基坑变形数据的合理分解，其分解能力要优于传统分解模型，且在5个监测点的预测结果中，ISFLA-RVM-GRNN模型的预测精度较高，预测结果的平均相对误差间于1.97%~2.07%。经外推预测，得基坑变形趋于稳定方向发展。同时，通过不同模型预测结果的对比性验证，得出ISFLA-RVM-GRNN模型相较传统预测模型具有更优的预测效果，验证了其构建思路是合理有效的。 |
| 英文摘要: |
| To ensure the high-precision prediction of foundation pit deformation, based on the on-site deformation monitoring results of foundation pit, the ICEEMD-ICA criterion was used for data processing, which decomposed the deformation data into trend items and random items, and then combined prediction was realized by the ISFLA-RVM-GRNN model. Finally, several traditional prediction models were introduced to make similar prediction, and the rationality of the combination prediction approach in this paper was verified by comparing the prediction results. The example analysis shows that the ICEEMD-ICA criterion can realize the reasonable decomposition of foundation pit deformation data, and its decomposition ability is superior to the traditional decomposition model. Among the prediction results of five monitoring points, the ISFLA-RVM-GRNN model has higher prediction accuracy, with an average relative error between 1.97% and 2.07%. After extrapolation and prediction, it is found that the foundation pit deformation tends to develop in a stable direction. At the same time, through the comparative verification of the prediction results of different models, it is concluded that the ISFLA-RVM-GRNN model has better prediction performance than the traditional prediction model, which verifies that its construction approach is reasonable and effective. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
