| 梁忠豪,秦楠*,纪沛志,周彤彤,葛强.高温作用后砂岩蠕变试验及PSO-BP神经网络单轴蠕变长期强度预测研究[J].实验力学,2022,37(4):573~584 |
| 高温作用后砂岩蠕变试验及PSO-BP神经网络单轴蠕变长期强度预测研究 |
| Research on sandstone creep test after high temperature and PSO-BP neural network uniaxial creep long-term strength prediction |
| 投稿时间：2021-05-22 修订日期：2021-09-28 |
| DOI：10.7520/1001-4888-21-126 |
| 中文关键词: 黄砂岩 单轴蠕变长期强度 蠕变变形 PSO-BP神经网络预测 |
| 英文关键词:yellow sandstone uniaxial creep experiment after high temperature long-term uniaxial creep strength creep deformation PSO-BP neural network prediction |
| 基金项目:矿山灾害预防控制教育部重点实验室开放研究基金(MDPC201915)；山东省自然科学基金(ZR2021QE202) |
|
| 摘要点击次数: 427 |
| 全文下载次数: 78 |
| 中文摘要: |
| 为了研究高温作用后黄砂岩的蠕变强度及变形特征，对高温作用后的黄砂岩开展单轴蠕变试验，系统地分析了高温损伤、轴压对黄砂岩蠕变变形特征、蠕变强度、蠕变速率的影响。利用PSO-BP神经网络算法对不同力学参数进行训练，预测高温作用后黄砂岩的单轴蠕变长期强度。研究结果发现：高温作用后黄砂岩存在蠕变应力阈值，低于阈值时仅发生稳定蠕变，高于阈值后发生不稳定蠕变；蠕变试验中试件处于低应力状态时，随着温度的增加，蠕变变形程度与稳态蠕变率呈线性变化关系。处于高应力状态时，温度对二者影响程度增大。使用PSO-BP神经网络预测高温作用后黄砂岩蠕变长期强度，发现比传统BP神经网络模型训练速度快、预测精度高。本文研究成果可为地下岩体工程高温后灾变重建提供一定的技术支撑和借鉴。 |
| 英文摘要: |
| In order to study the creep strength and deformation characteristics of yellow sandstone after high temperature action, uniaxial creep experiments were carried out on yellow sandstone after high temperature action, and the effects of high temperature damage and axial pressure on the creep deformation characteristics, creep strength and creep rate of yellow sandstone were systematically analyzed. The results demonstrate that there is a creep stress threshold for yellow sandstone after high-temperature action. Below the creep stress threshold, only stable creep occurs, while above the creep stress threshold, unstable creep occurs. Moreover, the degree of creep deformation and steady-state creep rate are linearly related with increasing temperature when the specimens are at low stress in the creep test. In the high stress state, the influence of temperature on the degree of creep deformation and steady-state creep rate increases at the same time. PSO-BP neural network is also applied to predict the long-term strength of creep in yellow sandstone after high temperature action. It is found that the actual fit is better, the training speed is faster and the prediction accuracy is higher than the traditional BP neural network model. The research results can provide some technical support and reference for the reconstruction of subsurface rock projects after high temperature catastrophic changes. |
| 查看全文 下载PDF阅读器 |
| 关闭 |
