公路梁桥支座脱空度预测模型及优化算法应用研究  

Prediction model and optimization algorithm application of support void degree for highway beam bridges

作　　者：马世纪 乔兰[1,2] 邓乃夫 李庆文[1,2] 陈璐 MA Shiji;QIAO Lan;DENG Naifu;LI Qingwen;CHEN Lu(Beijing Key Laboratory of Urban Underground Space Engineering,University of Science and Technology Beijing,Beijing 100083,China;School of Civil and Resource Engineering,University of Science and Technology Beijing,Beijing 100083,China;BGRIMM Technology Group,Beijing 102628,China)

机构地区：[1]北京科技大学城市地下空间工程北京市重点实验室,北京100083 [2]北京科技大学土木与资源工程学院,北京100083 [3]矿冶科技集团有限公司,北京102628

出　　处：《振动与冲击》2024年第15期218-227,243,共11页Journal of Vibration and Shock

基　　金：国家自然科学基金面上项目(52274107,52204091,52374113);中央高校优秀青年团队培育项目(FRF-EYIT-23-01);北京市科技新星计划资助(20230484242)。

摘　　要：桥梁支座脱空度是桥梁结构的常见病害,对桥梁的安全性和使用寿命产生严重影响。因此,准确获取支座的脱空信息成为一个重要的研究课题。该研究提出了一种两阶段的支座脱空检测方法,第一阶段,基于柔度矩阵对角差值比指标(flexibility diagonal matrix change rate,FDMCR)实现脱空支座定位,第二阶段采用BP(back propagation)神经网络进行支座脱空度的预测。研究针对单跨简支梁桥和三跨连续梁桥,通过有限元模拟实现了梁桥理论振型、自振频率的获取,并利用激光多普勒测振系统对室内单跨梁桥开展模态测试,获取实测数据,验证了所提方法的可行性。此外,针对单跨梁桥,分析了单目标粒子群优化算法、混合蛙跳算法、人工蜂群算法(artificial bee colony,ABC)优化算法和多目标非支配排序遗传算法(NSGA-II)优化所构建BP模型的效果。研究结果表明,两阶段的方法可以有效地实现脱空定位和脱空度预测,而优化算法则能够提高预测模型的预测效果,尤其是ABC算法在单跨梁桥脱空度预测模型上表现出更低的预测误差,同时基于支座位置属性构建多目标函数能够缓解单目标中各支座预测效果不均匀的问题。该研究对于预测公路梁桥支座脱空度具有重要的实际意义,并为类似问题的解决提供了新的思路和方法。Bridge support void degree is a common disease in bridge structures,it has a serious impact on the safety and service life of bridges.Therefore,correctly obtaining information of bridge support void degree has become an important study topic.Here,a two-stage support void detection method was proposed.In the first stage,based on the flexibility diagonal matrix change rate(FDMCR)index,support void positioning was realized.In the second stage,a BP(back propagation)neural network was used to predict support void degree.Aiming at a single-span simply supported beam bridge and a 3-span continuous beam bridge,the bridges’theoretical vibration modes and natural frequencies were obtained with finite element simulation.Modal testing was conducted for indoor single-span beam bridge using laser Doppler vibration measurement system to obtain measured data and verify the feasibility of the proposed method.In addition,for single-span beam bridge,the effectiveness of the BP model constructed by using the single objective PSO/SFLA/ABC optimization algorithm and the multi-objective non-dominated sorting genetic algorithm(NSGA-II)for optimization was analyzed.The study results showed that the two-stage method can effectively realize void positioning and void degree prediction,while optimization algorithms can improve the prediction effect of prediction model,especially,ABC algorithm has lower prediction errors for the void degree prediction model of single-span beam bridge;meanwhile,building a multi-objective function based on position attributes of supports can relieve the problem of uneven prediction effects of various supports in single targets;this study has important practical significance for predicting void degree of highway beam bridge supports,and provides new ideas and methods for solving similar problems.

关 键 词：桥梁支座 支座脱空度预测 柔度矩阵 BP神经网络 优化算法 

分 类 号：TU311.3[建筑科学—结构工程] TN911.6[电子电信—通信与信息系统]