| 侯少晗,苏翠侠,张茜*,刘尚林,马明慧.基于工程实测数据增量学习的掘进总推力分析与建模[J].实验力学,2024,39(5):599~611 |
| 基于工程实测数据增量学习的掘进总推力分析与建模 |
| Analysis and modeling of total thrust based on incremental learning of engineering measured data |
| 投稿时间：2023-11-15 修订日期：2024-01-02 |
| DOI：10.7520/1001-4888-23-238 |
| 中文关键词: 工程实测数据分析 硬岩隧道掘进机 掘进总推力 模型更新 样本增量宽度学习 |
| 英文关键词:engineering measurement data analysis hard rock TBM total thrust model updating increment-broad learning system |
| 基金项目:国家重点研发计划(2022YFC3802301); 国家自然科学基金项目(12372186，12022205) |
|
| 摘要点击次数: 2287 |
| 全文下载次数: 172 |
| 中文摘要: |
| 随着传感检测技术不断发展，复杂实验或工程实测可以给出丰富的力学量测试信息。如何通过数据分析提取关键力学量的演化规律、建立科学有效的预估模型是实验力学数据分析的重要研究内容。硬岩隧道掘进机(Tunnel Boring Machine，简称TBM)实测数据分析能够为施工过程中掘进参数的合理调控提供决策依据。掘进总推力是贯穿装备施工的关键力学量，由于掘进过程要穿越多种类型地质，装备掘进参数调控要求总推力预测模型具有适应工况变化的能力。针对上述需求，本文建立了基于样本增量宽度学习(Increment-Broad Learning System,简称I-BLS)的总推力建模方法，该方法能够随着掘进过程中工况的变化进行模型更新。结合新疆某隧道工程实例，基于上述方法建立了该工程的总推力预测模型，由独立测试集分析模型的有效性。将本方法的工程计算结果与基于3种常用机器学习方法的计算结果对比分析，结果表明本文方法能够在保持预测精度的同时，有效提高模型更新的计算效率。 |
| 英文摘要: |
| With the continuous development of sensing and testing technology, complex experiments or engineering measurements can provide abundant mechanical measurement information. How to extract the evolution law of key mechanical quantities through data analysis and establish a scientific and effective prediction model is an important research content of experimental mechanical data analysis. The analysis of the measured data of the hard rock Tunnel Boring Machine (TBM) can provide the decision basis for the reasonable control of tunneling parameters in the construction process. The total thrust is the key mechanical quantity through the equipment construction, because the driving process must pass through various types of geology, the control of equipment driving parameters requires the total thrust prediction model to adapt to the change of working conditions. To meet the above requirements, this paper establishes a total thrust modeling method based on Increment-Broad Learning System (I-BLS), which can update the model with the change of working conditions in the process of tunneling. Combined with a tunnel project in Xinjiang, a total thrust prediction model was established based on the above method, and the effectiveness of the model was analyzed by an independent test set. The engineering calculation results of the proposed method were compared with those of three common machine learning methods. The results show that the proposed method can maintain the prediction accuracy and improve the computational efficiency of model updating effectively. |
| 查看全文 下载PDF阅读器 |
| 关闭 |
