【作者】 秦远田；

【导师】 陈国平；

【作者基本信息】 南京航空航天大学 ， 固体力学， 2007， 博士

【摘要】 由于结构的复杂性和动载荷本身的复杂性,很多情况下不能直接测量动载荷,因此由结构的动响应和结构的动态特性来反演动载荷是非常必要的,这方面的研究已经取得了令人瞩目的进展。单点和多点动载荷的识别技术逐渐成熟,但是对于结构上连续分布的动载荷的识别技术几乎处于空白。本文主要研究分布动载荷的识别问题,建立了矩量法和小波方法识别动载荷的识别模型。运用矩量法正交基函数的收敛特性,将待识别的复杂动载荷转化为在广义正交域中求解基函数的拟合系数,从而使复杂问题简单化。运用有限元分析软件并基于线性系统的叠加性质,建立了该方法的通用识别模型,该模型适用于工程中复杂的结构和任意的边界条件。该方法的识别精度高,有很强的抗干扰性能,且对测量点位置的要求较宽松,适合工程运用。通过仿真算例和实验验证了方法的可行性和正确性。本文的主要研究内容如下:(一)将矩量法引入到移动载荷的识别中。文中首先以车桥系统为工程背景,讨论了几种车桥耦合系统简化的建模方法,并用动响应实验数据进行比较,得出了适用于工程的二自由度车桥简化模型。在此基础上研究了一类特殊的分布载荷—移动载荷的识别,建立了基于矩量法的移动载荷识别的通用模型,解决了移动载荷识别受限于结构的复杂程度和结构边界条件的缺点。(二)将矩量法正交基函数扩展成二维,并应用在频域二维分布动载荷的识别中。用二维正交基函数拟合待识别载荷函数中的两个未知量,以有限元分析软件为平台,建立了基于复杂结构的有限元模型的二维分布动载荷识别模型的频域方法,并用算例进行了验证。(三)运用分布动载荷识别频域方法的思想,用三维正交基函数拟合待识别二维分布动载荷函数中的三个待识别量,直接在时域中识别二维分布动载荷。(四)讨论了矩量法中如何用尽量少的测量点来识别分布动载荷的问题。在有限元模型中,将二维分布动载荷等效成有限元节点载荷,将节点载荷以某一方式排列在一假定的坐标轴上,则二维分布动载荷就成了该坐标轴上的节点载荷,运用一维分布动载荷的识别技术来识别这些点载荷,从而达到降维的目的。优化节点载荷序列在假想一维坐标轴上的排列顺序就可以降低拟合阶数,从而达到减少分布动载荷识别中需求的测量点数的目标。(五)引入小波理论到动载荷的识别中,建立了基于结构动力学有限元方程小波方法的动载荷识别模型,用仿真算例验证了方法的正确性。(六)分别用移动载荷和分布载荷试验来验证文中方法的工程可行性。更多还原

【Abstract】 Load cannot be measured in most conditions because of complexity of the structure and the load.So, it is necessary to inverse it from the response and the characteristic of the structure.This study has acquired outstanding achievements.The technology of points-load-identification are very better and better. However the technology of identification for structurally distributed load in succession is seldom.In this dissertation, The main work is to study the technology for distributed load identification.and establish identification models based on the moment method and wavelet method. Based on the vonvergence of basis function for moment method, A complex dynamic load to be identified is simplized by solving the coefficients of the orthogonal basis function. According to the linear principle, A general identification model is established in the software for finite element anaylsis. It includes all complex structure under arbitrary boundary conditions. The method possesses of the abilities of high identification precision, anti-interference and nonrestraint of the measure position. The feasibility and correctness of these methods are verified by testing instance and experiment. The primary contents are as follows:1. A moment method is applied to moving load identification. First three methods for establishing simple model are discussed for the coupling interaction system between vehicle and bridge .These simulating results generated by three methods are contrasted with each other and with experiment data, A conclusion is drawn that the two degrees of freedom model is best.Then the identification of moving load, A special distributed load, is studied in this dissertation, A general model for moving load identification is established on a base of the moment method and some shorcomings are overcomed.2. The basis functions for the moment method are defined in a two-dimension orthogonal space and used for identifying two-dimension distributed load. The two variables of the unknown load are fitted by the basis functions.A model in frequence domain is established for identifying two-dimension distributed load, acting on a complex constructure, in the finite-element-analysis software and it is verified by testing instance.3. Based on the ideas of the frequence domain method of a distributed load identificaition, A two-dimension distributed load can be identified directly in time domain by using three-dimension basis function to fit the three vaiable of unknown load.4. A problem is dicussed that how to reduce the number of measure points to identify the distributed load equally. A two-dimension distributed load can be equaled to the node loads in finite element model.when these node loads are arranged on a supposed axis, they can be identified by a one-dimension method, so the aim for reducing dimension can be reached. When these node loads on the supposed axes are arrayed in an optimized order, the fitting order can be reduced and the object of reducing the number of measure points can be reached. 5. The wavelet theory is introduced into dynamic identification and a model based on dynamic finite element is established for wavelet method whose correctness is verified by testing instance.6. The practical feasibility for these methods is verified by moving and distruted load experiment.更多还原