摘要
为指导中等跨径钢箱梁桥智能选型设计,将钢桥模块化理论与随机森林算法相结合用于初步选型,采用半监督学习的方法增强数据集,提出一种中等跨径钢箱梁桥模块化智能选型方法。该方法首先选择桥梁10个特征作为自变量,将钢箱梁的9种横截面形式作为因变量,收集75条专家评价数据作为训练集;其次通过算法适配和对比分析,采用遗传-随机森林的组合算法对结果的分类精度相较于原始的随机森林算法模型准确率提升了约10%;然后采用半监督学习算法对训练数据集进行增强,在不改变算法的同时进一步提高了模型准确率,整体模型准确率接近0.9;最后基于上述算法和前后端技术实现在Web平台上的快速部署和三维结果展示。
A modularized intelligent method used to select types for medium-sized steel box girder bridge is proposed,which combines the modularized theory and random forest calculation of steel bridges in the preliminary bridge type selection.The method takes the semi-supervised learning to enhance data sets.Firstly,10 eigenfactors are selected as independent variables,9 cross-section patterns as the results of dependent variables and 75 expert evaluation data as training set.Secondly,compared with algorithm adaptation and comparative analysis,the combined algorithm of generic algorithm and random forest can generate more accurate results classification,about 10%higher than the accuracy of the original random forest algorithm.Thirdly,the semi-supervised learning is used to enhance data sets,aiming to improve the accuracy of the model without changing the algorithm,as a result,the overall accuracy of the model approximates 90%.Finally,based on the above algorithms and front-end and back-end technologies,the rapid deployment and 3D results display on the Web platform are realized.
作者
徐秀丽
陈宇文
严锴
李雪红
吴军华
XU Xiu-li;CHEN Yu-wen;YAN Kai;LI Xue-hong;WU Jun-hua(School of Civil Engineering,Nanjing Tech University,Nanjing 211816,China;Suzhou Port and Shipping Development Center,Suzhou 215000,China;School of Computer Science and Technology,Nanjing Tech University,Nanjing 211800,China)
出处
《世界桥梁》
北大核心
2023年第3期58-65,共8页
World Bridges
基金
江苏省交通运输科技项目(2021Y17-2,2019Y18)。
关键词
钢桥
钢箱梁
模块化
智能选型
随机森林
半监督学习
平台开发
steel bridge
steel box girder
modularization
intelligent type selection
random forest
semi-supervised learning
platform development
