基于健康监测的桥梁结构有限元模型修正方法研究

发布时间：2018-11-19 11:02

【摘要】：桥梁工程是国家生命线工程。桥梁通车运营后,随着时间的推移,会受到环境侵蚀、人为因素、材料老化、自然灾害以及车辆荷载的交互作用,从而受到不同程度的损伤和劣化。桥梁健康监测系统从传感器实时监测的数据中获取结构响应信息和桥位环境状况,并对数据计算分析,判断和评估桥梁结构的受力状态和抗力衰减规律,确保桥梁安全运营。健康监测系统任务的核心是损伤识别,然而损伤识别的前提是建立精确的有限元模型。本文以桥梁健康监测系统监测的动力响应为有限元模型修正的研究目标,以神经网络算法和响应面法在模型修正中的应用为研究对象。采用ANSYS软件对有限元模型进行建模分析。利用遗传算法优化神经网络,采用MATLAB编制算法程序实现模型修正;同时利用MATLAB开发了工具箱,实现了响应面法模型修正的可视化。基于健康监测系统实时监测的数据可以实现模型修正的实时性,能够及时的为桥梁结构计算提供一个准确的有限元模型。主要研究内容如下:1、介绍有限元模型修正的理论,对有限元模型修正的一般过程进行说明;对模型修正中的参数选取,以及模型修正后的误差判别做了详细介绍;2、针对传统加速度传感器优化布置方法的不足,提出了本文方法,结合工程经验完成沮河大桥加速度传感器的优化布置;总结不同传感器性能的优缺点,并介绍全桥传感器的布置方案;3、针对普通广义回归神经网络的缺陷,提出基于遗传算法的优化方法,并将其应用到有限元模型修正中,详细介绍神经网络在模型修正中应用原理及流程;4、在MATLAB软件环境下,将本文提出的优化广义回归神经网络模型修正方法运用到依托工程上。通过对修正结果的对比分析,验证了基于遗传算法优化神经网络模型修正法的准确性、优越性和有效性;5、介绍响应面法进行模型修正的原理及流程,并将其应用到依托工程上。利用MATLAB开发出一个工具箱,将响应面法模型修正做到了可视化。
[Abstract]:Bridge engineering is a national lifeline project. After the bridge is opened to traffic, with the passage of time, it will be affected by environmental erosion, human factors, material aging, natural disasters and the interaction of vehicle loads, so it will be damaged and degraded to varying degrees. The bridge health monitoring system acquires the structure response information and the bridge environment condition from the sensor real-time monitoring data, calculates and analyzes the data, judges and evaluates the stress state and the resistance attenuation rule of the bridge structure, and ensures the bridge safe operation. Damage identification is the core of the task of health monitoring system. However, the premise of damage identification is to establish an accurate finite element model. In this paper, the dynamic response of bridge health monitoring system is taken as the research object of finite element model modification, and the application of neural network algorithm and response surface method in model modification is studied. The finite element model is modeled and analyzed by ANSYS software. Genetic algorithm is used to optimize neural network, MATLAB is used to program the model correction, and MATLAB is used to develop toolbox to realize the visualization of response surface model modification. The real-time monitoring data based on the health monitoring system can realize the real-time correction of the model and can provide an accurate finite element model for bridge structure calculation in time. The main research contents are as follows: 1. The theory of finite element model modification is introduced and the general process of finite element model modification is explained. 2. Aiming at the deficiency of the traditional acceleration sensor optimization arrangement method, this paper puts forward the method of this paper, combining the engineering experience to complete the optimization layout of the acceleration sensor of the Ju River Bridge; The advantages and disadvantages of different sensors are summarized, and the layout scheme of the full bridge sensor is introduced. 3. Aiming at the defects of general generalized regression neural network, an optimization method based on genetic algorithm is proposed and applied to finite element model modification. The application principle and flow chart of neural network in model modification are introduced in detail. 4. In the MATLAB software environment, the optimized generalized regression neural network model modification method proposed in this paper is applied to the supporting engineering. The accuracy, superiority and effectiveness of the neural network model modification method based on genetic algorithm are verified by the comparison and analysis of the modified results. 5. The principle and process of model modification based on response surface method are introduced, and applied to supporting engineering. A toolbox is developed by using MATLAB to visualize the response surface model modification.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U446;U441

【参考文献】