建筑混凝土疲劳寿命智能预测模型设计  

Design of intelligent prediction model for fatigue life of building concrete

作　　者：于皓皓 YU Haohao(School of Architectural Engineering,Jiyuan Vocational and Technical College,Jiyuan 459099,China)

机构地区：[1]济源职业技术学院建筑工程学院,河南济源459099

出　　处：《混凝土》2025年第2期167-171,共5页Concrete

基　　金：2023年度河南省重点研发与推广专项(科技攻关)项目“新型聚合膨润土的耐盐性及其自愈修复性能研究”(232102320089)。

摘　　要：建筑混凝土疲劳寿命预测的精度受到材料参数和微观结构的复杂性、外部环境因素等多个因素的影响,导致预测模型精准设计的难度增加。为了有效解决这一问题,提出了基于深度随机森林的建筑混凝土疲劳寿命预测方法。利用水泥基料与减水剂、河砂细骨料、石子颗粒粗骨料和水制备建筑混凝土试件,测量并确定混凝土试件的剥落损伤度、孔隙率、弹性模量损伤等冻融循环损伤参数。根据损伤参数对混凝土试件进行损伤状态识别。结合材料的损伤状态,筛选出关键的疲劳特征。搭建深度随机森林模型,将选取的关键疲劳特征输入至该模型中进行训练。经过多次迭代运算,模型可以逐渐学习特征与疲劳寿命之间的关系,并预测建筑混凝土的疲劳寿命值。试验结果表明:该方法的混凝土损伤状态识别准确率均值为96.55%,混凝土疲劳寿命预测效果好,时间短,可行性高。The prediction of the fatigue life of building concrete is influenced by multiple factors such as the complexity of material parameters and microstructure,as well as external environmental factors,which increases the difficulty of predicting the fatigue life of building concrete.In order to solve this problem effectively,a fatigue life prediction method of building concrete based on deep random forest was proposed.Using cement-based materials and water reducing agents,river sand fine aggregates,stone particle coarse aggregates,and water to prepare building concrete specimens,measure and determine the freeze-thaw cyclic damage parameters of concrete specimens,such as peeling damage,porosity,and elastic modulus damage.Identify the damage status of concrete specimens based on damage parameters.Based on the damage state of the material,select key fatigue characteristics.Build a deep random forest model,and input the selected key fatigue characteristics into the model for training.After several iterations,the model can gradually learn the relationship between characteristics and fatigue life,and predict the fatigue life of building concrete.The experimental results show that the average accuracy rate of concrete damage state recognition using this method is 96.55%,and the prediction effect of concrete fatigue life is good,with short time and high feasibility.

关 键 词：深度随机森林 建筑混凝土 疲劳寿命预测 混凝土试件 损伤参数 损伤状态 关键疲劳特征 

分 类 号：TU528.07[建筑科学—建筑技术科学]