| 引用本文: | 陈熹俊,韩小雷,张垒,吴智辉,季静.基于随机森林算法的型钢混凝土梁柱纤维单元参数优化[J].哈尔滨工业大学学报,2025,57(5):127.DOI:10.11918/202401009 |
| CHEN Xijun,HAN Xiaolei,ZHANG Lei,WU Zhihui,JI Jing.Parameter optimization of steel reinforced concrete beam-column fiber elements based on random forest algorithm[J].Journal of Harbin Institute of Technology,2025,57(5):127.DOI:10.11918/202401009 |
|
| 本文已被：浏览 15次 下载 18次 |
码上扫一扫！ |
|
|
| 基于随机森林算法的型钢混凝土梁柱纤维单元参数优化 |
|
陈熹俊1,韩小雷1,2,张垒1,吴智辉1,季静1,2
|
|
(1.华南理工大学 土木与交通学院,广州 510641;2.亚热带建筑与城市科学全国重点实验室(华南理工大学),广州 510641)
|
|
| 摘要: |
| 为提高型钢混凝土（steel reinforced concrete,SRC）梁柱构件的弹塑性分析精度,提出基于随机森林算法的SRC梁柱纤维单元参数预测模型（PMFEP-SRC）,对SRC梁柱纤维单元的滞回曲线拟合效果进行优化。基于收集的153根SRC梁柱试件试验数据,以极限承载力比值RF和耗能能力比值RE为目标参数,承载力调整系数CF和刚度调整系数CS为调整参数,爬山算法作为寻优手段,实现纤维单元最优调整参数的求解。通过随机森林算法,以SRC梁柱试件的试验控制参数作为特征参数,纤维单元最优调整参数（求解的承载力调整系数CF与刚度调整系数CS）作为标签,训练并建立PMFEP-SRC。设计并完成一批不同剪跨比的SRC柱低周往复加载试验,通过试验数据进一步验证了PMFEP-SRC的准确性与可靠性。结果表明:PMFEP-SRC能够较好地拟合不同破坏形态SRC梁柱试件的滞回曲线,对SRC试件的极限承载力和耗能面积的拟合精度远高于未经参数优化的纤维单元。 |
| 关键词: 型钢混凝土梁柱 纤维单元 低周往复加载试验 弹塑性分析 随机森林算法 |
| DOI：10.11918/202401009 |
| 分类号:TU392.1 |
| 文献标识码:A |
| 基金项目:国家自然科学基金(3,5) |
|
| Parameter optimization of steel reinforced concrete beam-column fiber elements based on random forest algorithm |
|
CHEN Xijun1,HAN Xiaolei1,2,ZHANG Lei1,WU Zhihui1,JI Jing1,2
|
|
(1.School of Civil Engineering and Transportation, South China University of Technology, Guangzhou 510641, China; 2.State Key Laboratory of Subtropical Building and Urban Science(South China University of Technology), Guangzhou 510641, China)
|
| Abstract: |
| In order to improve the accuracy of elastic-plastic analysis for SRC (steel reinforced concrete) beam-column members, the prediction model for fiber element parameters of SRC members (PMFEP-SRC) based on random forest algorithm is proposed to optimize the hysteresis curve fitting of SRC beam-column fiber elements. Based on 153 collected SRC beam-column specimens, with the peak load capacity ratio RF and energy dissipation capacity ratio RE are taken as target parameters, and the load capacity adjustment factor CF and stiffness adjustment factor CS are taken as adjustment parameters, the hill climbing algorithm is employed to determine the optimal parameters of the fiber element. PMFEP-SRC was trained and established by the random forest algorithm with the test control parameters of SRC beam-column specimens as the characteristic parameters and the optimal adjustment parameters of the fiber element (the solved load capacity adjustment factor CF and stiffness adjustment factor CS) as the labels. Finally, a batch of SRC columns with different shear-to-span ratios were designed and completed for low-cyclic loading tests, and the accuracy and reliability of PMFEP-SRC was further validated using the test data. The results show that PMFEP-SRC can effectively fit the hysteresis curves of SRC beam-column specimens with different failure modes, and the fitting accuracy of peak load capacity and energy dissipation of SRC specimens is significantly higher than that of the fiber elements without parameter optimization. |
| Key words: steel reinforced concrete(SRC) beam-column fiber element low-cyclic loading tests elastic-plastic analysis random forest algorithm |
