基于神经网络技术的填充墙裂缝开展预测研究.pdfVIP

建筑科嗲 第 32 卷第9 期 Vo l. 32 ,No. 9 2016 年 9 月 BUILDING SCIENCE Sep. 2016 [文章编号] 1002-8528(2016)09-0069-05 DOI: 10. 13614/j. cnki. 11-1962/tu. 2016. 09. 012 基于神经网络技术的填充墙裂缝开展预测研究 苏启旺1 气陈 鹏1 ，彭祖日在1 ，罗道儒1(1. 西南交通大学土木工程学院，四川成都 610031; 2. 西南交通大学抗震 技术四川省重点实验室，四川成都 610031) [摘 要]为探索框架填充墙在地震作用下开裂模式，提出了墙体高宽比和梁柱线刚度比作为预测参数，基于巴收集的 试验研究成果，运用了神经网络技术对上述参数不同组合下的填充墙裂缝开展进行了预测研究，通过所定义的评价预测结果 的指标，比较了不同预测参数组合时框架填充墙裂缝开展模式，并在此基础上对预测结果进行验证。分析结果表明:在使用 墙体高宽比和梁柱线刚度比的双参数组合时，填充墙裂缝预测的开裂模式与实际裂缝发展吻合较好，它能较准确地预测框架 填充墙开裂模式，可为设计提供参考。 [关键词]框架结构;填充墙;裂缝;开裂模式;神经网络 [中图分类号] TU362 [文献标识码] A Research on Cracking Development of Infilled Wall in RC Frame Based on Neural Network 1 2 1 1 Su Qiωαng • ， Chen Peng , Peng Zuzhα0 ， Luo Daoru 1 (1. School of Civil Engineering , Southwest Jiαotong University Chengdu 610031 , Chinα ， 2. Key Laboratory of Seismic Engineering of Sichuan Province , Chengdu 610031 , Chinα) Abstract: In order to know the cracking pattern of frame infilled wall , wall s aspect ratio and line stiffness ratio of beam and column as prediction parameters are proposed. The appearance of cracking in different parameter combination is evaluated by neural network based on existing experiment results: According to the defined indicators which are used to evaluate crack , the crack patterns in different parameter combination are compared. In the same time , the experimental verification for 巳rack pattern of the infills in RC frame is done. The results show that when wall s aspect ratio and line stiffness ratio of beam and column are combined as prediction parameters , the predicted result appears to be the most matched with the actual experiment. It can well predict the