基于BP神经网络-GSL&PS-PGSA的多塔斜拉桥成桥索力优化研究  

Research on Optimization of Cable Force of Multi-tower Cable-stayed Bridge Based on BP Neural Network-GSL&PS-PGSA

作　　者：张冬 肖军[2,3] 周彬 袁晟[4] ZHANG Dong;XIAO Jun;ZHOU Bin;YUAN Sheng(Overseas Business Department(International Company)of CCCC Second Public Bureau,Xi'an,Shaanxi 710061,China;CCCC Second Highway Engineering Bureau Co.,Ltd.,Xi'an,Shaanxi 710065,China;CCCC Highway Long Bridge Construction National Engineering Research Center Co.,Ltd.,Beijing 100011,China;School of Civil Engineering,Changsha University of Science&Technology,Changsha,Hunan 410114,China)

机构地区：[1]中交二公局海外事业部(国际公司),陕西西安710061 [2]中交第二公路工程局有限公司,陕西西安710065 [3]中交公路长大桥建设国家工程研究中心有限公司,北京100011 [4]长沙理工大学土木工程学院,湖南长沙410114

出　　处：《公路工程》2022年第4期34-39,共6页Highway Engineering

基　　金：中国交建特大科技研发项目(2019-ZJKJ-07);国家重点基础研究发展计划(973项目)(2015CB057705);国家自然科学基金项目(51678068)。

摘　　要：为研究多塔斜拉桥成桥索力优化问题,以克罗地亚佩列沙茨大桥6塔中央单索面钢箱梁矮塔斜拉桥为研究背景,基于BP神经网络,引入了GSL&PS-PGSA(生长空间限定与并行搜索的算法组合),建立了以主梁最小弯曲应变能为目标函数的优化模型,提出了一种基于BP神经网络-GSL&PS-PGSA的多塔斜拉桥成桥索力优化方法,解决了易陷入局部最优的缺点,提高了搜索速率、迭代收敛效率和全局寻优能力,对背景桥梁成桥索力进行了优化研究。研究结果表明,优化前后索力最大变化幅度为10.4%,优化后各相邻索力差更为均匀,主梁受力分布更为合理;优化前后主梁变形最大变化幅度为-27.8%,优化后主梁线形更为平缓;验证了基于BP神经网络-GSL&PS-PGSA的多塔斜拉桥成桥索力优化方法的可靠性与精确性。In order to study the optimization of the cable tension of a multi-tower cable-stayed bridge,the 6-tower central single-cable-plane steel box girder and low-tower cable-stayed bridge of the Croatian PeljeBridge is used as the research background.Based on the BP neural network,GSL&PS-PGSA(algorithm combination of growth space limitation and parallel search),established an optimization model with the minimum bending strain energy of the main girder as the objective function,and proposed a multi-tower cable-stayed bridge based on BP neural network-GSL&PS-PGSA The force optimization method solves the shortcomings of being easy to fall into the local optimum,improves the search rate,iterative convergence efficiency and the global optimization ability,and optimizes the background bridge cable force.The research results show that the maximum change range of cable force before and after optimization is 10.4%.After optimization,the difference between adjacent cable forces is more uniform and the force distribution of the main beam is more reasonable;the maximum change range of main beam deformation before and after optimization is-27.8%.The beam alignment is more gentle;the reliability and accuracy of the method for optimizing the cable force of a multi-tower cable-stayed bridge based on BP neural network-GSL&PS-PGSA is verified.

关 键 词：桥梁工程 BP神经网络 GSL&PS-PGSA 斜拉桥 索力优化 

分 类 号：U448.27[建筑科学—桥梁与隧道工程]