基于机器学习的隔震结构最大层间位移角预测  

PREDICTION OF MAXIMUM INTER-STOREY DRIFT OF ISOLATED STRUCTURE BASED ON MACHINE LEARNING

作　　者：彭嘉舜 王敦强 PENG Jia-shun;WANG Dun-qiang(Shandong Jianzhu University,250101,Ji'nan,China)

机构地区：[1]山东建筑大学,济南250101

出　　处：《建筑技术》2024年第16期2001-2005,共5页Architecture Technology

摘　　要：如今采用机器学习等智能算法预测结构最大层间位移角已成为趋势,但将其用于隔震体系方面的研究较少。因此,基于1 736个隔震体系在地震作用下的响应数据,研究3种机器学习算法在该数据预测方面的预测性能。结果表明,神经网络、随机森林及支持向量机的总体决定系数均达到0.96以上,其中神经网络算法准确率最高为0.999 8,但其耗时最长约为3 486 s,支持向量机算法的总体预测准确率最低为0.974 4,其耗时最短为108 s。这表明机器学习算法能较好地预测隔震结构的最大层间位移角。Nowadays,it has become a trend to predict the maximum inter-storey drifts of structures by intelligent algorithms such as machine learning.However,there are few studies on the application of such intelligent algorithms in seismic isolation systems.On the basis of response data of 1736 seismic isolation systems in earthquakes,data predication by three machine learning algorithms was studied.Results show that the total determination coefficient of the neural network,random forest and support vector machine is greater than 0.96;the neural network algorithm has the highest prediction accuracy(0.9998)but requires the longest time(approximately 3486 s);and the support vector machine has the lowest prediction accuracy(0.9744)but requires the shortest time(108 s).This indicates that the machine learning algorithm can be applied to better predict the maximum inter-storey drifts of isolated structures.

关 键 词：建筑信息化 隔震结构 层间位移角 机器学习 

分 类 号：TU375[建筑科学—结构工程]