基于Elman桥梁损伤识别与监测数据处理研究

发布时间：2018-03-12 09:53

本文选题：Elman神经网络　切入点：蚁群算法　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：桥梁结构是现代交通运输系统的重要节点工程,为了保证桥梁的安全运行,目前许多大型桥梁都安装有健康监测系统,但如何从大量的监测数据中获取有用信息,并针对现场采集的信息判断桥梁的损伤状况是目前工程界的研究热点之一。为此本文提出使用Elman神经网络对桥梁的关键节点进行了损伤判别的方法,并将该方法应用到石河大桥实际工程中,本文的主要研究内容如下：(1)选取损伤因子和损伤位置,利用石河大桥梁的有限元模型模拟桥梁未损伤和不同程度损伤状况下损伤状况,获取损伤数据作为神经网络的输入,本文提出使用Elman神经网络作为损伤识别的模型并采用BP神经网络作为参照对比。(2)为了改善Elman神经网络的损伤识别效果,避免神经网络由于初始权值的随机性使得结果陷入局部最优而出现损伤率不高的问题,提出了使用蚁群算法优化神经网络。(3)在监测数据处理方面,对七个测点原始监测数据样本进行了数据分析前处理,本文对石河大桥监测应变和温度数据进行了时间序列分析,实验发现温度与应变都有周期变化趋势,并对石河大桥的应变数据采用多种距离指标进行了聚类分析最终获得两种聚类。本文将Elman神经网络应用于桥梁结构损伤识别,因Elman神经网络特殊的结构特点使其有较好的稳定性,通过仿真实验得出这种方法是可行的,经过蚁群算法优化后的神经网络识别率得到提高。将该模型与石河大桥的监测数据相结合,识别结果未发现石河大桥存在损伤,且桥梁健康监测系统采集回的数据是正常的且桥梁结构无异常。
[Abstract]:Bridge structure is an important node engineering of modern transportation system. In order to ensure the safe operation of bridge, many large bridges are equipped with health monitoring system, but how to obtain useful information from a large number of monitoring data. It is one of the research hotspots in the engineering field to judge the damage condition of the bridge based on the information collected in the field. In this paper, a method to judge the damage of the key nodes of the bridge by using the Elman neural network is put forward. This method is applied to the actual project of Shihe Bridge. The main research contents of this paper are as follows: (1) selecting the damage factors and location, and using the finite element model of Shihe Bridge to simulate the damage condition of the bridge under the condition of undamaged and different degree of damage. In order to improve the damage identification effect of Elman neural network, we propose to use Elman neural network as the model of damage identification and BP neural network as reference. In order to avoid the problem of low damage rate due to the randomness of the initial weights, the paper proposes that the ant colony algorithm is used to optimize the neural network in the aspect of monitoring data processing. In this paper, the data of the monitoring strain and temperature of Shihe Bridge are analyzed in time series. It is found that both temperature and temperature have a tendency to change periodically. The strain data of Shihe Bridge are analyzed by a variety of distance indexes, and finally two kinds of clustering are obtained. In this paper, the Elman neural network is applied to the damage identification of the bridge structure. Because of the special structure characteristics of Elman neural network, it has good stability. The simulation results show that this method is feasible. The recognition rate of neural network optimized by ant colony algorithm is improved. Combining the model with the monitoring data of Shihe Bridge, no damage is found in Shihe Bridge. The data collected by the bridge health monitoring system are normal and the bridge structure is not abnormal.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U446;TP274

【相似文献】