基于ANN的RECFST短柱轴压承载力预测
Prediction of Axial Compression Bearing Capacity of RECFST Short Columns Based on Artificial Neural Network
杜运兴 1刁俊杰1
作者信息
- 1. 湖南大学土木工程学院,湖南长沙 410082
- 折叠
摘要
目的 针对相关设计规范和文献在计算圆端形截面钢管混凝土短柱轴压承载力上的局限性,开发高精高效的轴压承载力预测模型.方法 首先,基于国内外已有的RECFST短柱轴压试验研究结果建立有限元模型,并通过验证;其次,基于Python脚本批量生成有限元模型,建立涵盖广泛输入参数的数据集;然后,利用数据集开发高精度的ANN模型并与相关规范和文献结果进行比较;最后,基于ANN模型开发GUI图形用户界面工具.结果 ANN模型预测值与试验结果之比的平均值NANN/Nu=0.98,模型预测误差远低于相关规范和文献公式预测误差;ANN模型的均方误差KMSE=7.373 4 × 10-7,总数据样本回归值R=0.999 63,表明了ANN模型的有效性以及预测结果的精确性.结论 ANN模型可以准确预测RECFST短柱的轴压承载力,基于模型开发的GUI工具简便实用.
Abstract
In view of the limitations of relevant design specifications and literature formulas in calculating the axial compression bearing capacity of round-ended concrete-filled steel tubular(RECFST)short columns,a high-precision and wide-ranging axial compression bearing capacity prediction model was developed.Firstly,the axial compression test of RECFST short columns completed at home and abroad is investigated,and the finite element model which is strictly verified by the test data is established.Secondly,the finite element model is generated in batches based on Python scripts,and a data set covering a wide range of input parameters is established.Then,the data set is used to develop a high-precision ANN model and compare it with relevant specifications and formulas.Finally,a graphical user interface(GUI)tool is developed based on the ANN model for practical engineering applications.The average value of the ratio of the predicted value of the ANN model to the experimental result is NANN/Nu=0.98,and the prediction error of the model is much lower than that of the relevant specifications and formulas.The mean square error of the ANN model is KMSE=7.373 4 x 10-7,and the regression value of the total data sample is R=0.999 63,indicating the validity of the ANN model and the accuracy of the prediction results.The ANN model can accurately predict the axial compression bearing capacity of RECFST short columns.The GUI tool developed based on the model is simple and practical,which provides a new idea and method for studying the axial compression bearing capacity of RECFST short columns.
关键词
ANN/RECFST短柱/轴压承载力/图形用户界面工具Key words
artificial neural network/RECFST short column/axial bearing capacity/graphical user interface tools引用本文复制引用
出版年
2024
NETL
NSTL
万方数据