混合机器学习预测空心钢管混凝土柱承载力  

Prediction of Axial Load Capacity of Hollow Steel Tube Concrete Columns Using Hybrid Machine Learning

作　　者：刘亚南 LIU Yanan(China Railway First Survey and Design Institute Group Co.Ltd,Xi'an,Shaanxi 710043,China)

机构地区：[1]中铁第一勘察设计院集团有限公司,西安710043

出　　处：《铁道工程学报》2024年第12期23-28,39,共7页Journal of Railway Engineering Society

摘　　要：研究目的:本文针对空心钢管混凝土柱性能预测过程中存在的不确定性和误差问题,基于模型试验、有限元模拟及已有研究成果,构建包含构件几何参数、材料特性等在内的综合数据库。利用实数编码遗传算法(RCGA)与自适应神经模糊推理系统(ANFIS)相结合的混合机器学习模型,提出一种创新算法,用于精准预测CFST柱的极限轴压承载力。研究结论:(1)RCGA-ANFIS模型在处理高维、非线性和噪声数据方面表现优异,显著提升了预测的准确性和可靠性;(2)通过蒙特卡罗模拟和敏感性分析,验证了模型的稳健性和有效性,确保了其在不同条件下的高性能表现;(3)该模型不仅为CFST柱的极限承载力预测提供了精确、高效的工具,还为建筑设计与施工提供了技术支持,有助于实现更安全、经济和可持续的建筑设计决策;(4)本研究不仅为CFST柱的承载力预测提供了新的方法,还为CFST构件的设计与优化提供了有效的保障,具有工程应用价值。Research purposes:This paper addresses the uncertainty and error issues in the performance prediction process of hollow steel tube concrete columns.Based on model experiments,finite element simulations,and existing research results,a comprehensive da-tabase including component geometric parameters and material properties,is constructed.A hybrid machine learning model combining real coded genetic algorithm(RCGA)and adaptive neuro fuzzy inference system(ANFIS)is proposed to accurately predict the ultimate axial compressive bearing capacity of CFST columns.Research conclusions:(1)The RCGA-ANFIS model performs well in handling high-dimensional,nonlinear,and noisy data,signifi-cantly improving the accuracy and reliability of predictions.(2)The robustness and effectiveness of the model were verified through Monte Carlo simulation and sensitivity analysis,ensuring its high-performance performance under different conditions.(3)This model not only provides an accurate and efficient tool for predicting the ultimate bearing capacity of CFST columns,but also provides technical support for building design and construction,helping to achieve safer,more economical,and sustainable building design decisions.(4)This study not only provides a new method for predicting the bearing capacity of CFST columns,but also provides effective guarantees for the design and optimization of CFST components,with significant engineering application value.

关 键 词：空心钢管混凝土 极限承载力 遗传算法 模糊推理 混合机器学习 

分 类 号：TU398.9[建筑科学—结构工程] TP181[自动化与计算机技术—控制理论与控制工程]