【作者】 张辉东；

【导师】 练继建；

【作者基本信息】 天津大学 ， 水利水电工程， 2007， 博士

【摘要】 随着社会与经济的发展,水力发电作为一种可再生的清洁能源日益得到重视和开发。水电机组的单机容量和引用水头急剧提高,厂房振动问题日益突出。大型复杂水电站厂房结构的动力学问题一直是国内外研究的热点和难点。目前,水电厂房结构在非线性动力学、抗震、流固耦合振动、模态参数识别方法等方面略显不足。因此,该文致力于水电站厂房结构从工程中提炼出的动力学正、反问题的解决方案和结构原型振动观测等方面的研究。主要研究成果如下:⑴传统的水电站厂房结构抗震研究思路主要是基于反应谱方法对结构进行线性分析。本文提出一种新的抗震分析方法,即用地震时程法并考虑材料的非线性因素对水电站厂房结构进行抗震研究,地震时程分析法具有全过程仿真的特点,它使得位移场(应力应变场)描述更细致,包括变形、应力、损伤形态等动力特性,它能准确的反应厂房结构的地震响应;而考虑非线性因素则能够展现裂缝出现的位置及其数量,能够精确细致地暴露结构的薄弱层和薄弱部位。这种方法的提出,对大体积混凝土结构进行基于性能的抗震研究及安全评估具有重要的意义。⑵本文利用强耦合法把有限元理论和声场理论结合起来,建立了流固耦合场下厂房结构与流场的有限元动力学方程,为分析水电站厂房结构动力学特性提供了理论基础,计算了坝后式水电站厂房的流固耦合振动特性,并与河床式水电站流固耦合特性进行比较,阐述了流体-结构两者之间的耦联振动特点,为水电站厂房结构防止共振的产生、确保安全发电奠定基础。⑶水电站厂房结构振动主要是水力、机械和电磁三大类振源引起的,厂房结构与机组之间存在明显的耦联作用,厂房结构和机组振动系统呈明显的动态耦合效应和非线性特征,用传统的力学方法预测厂房结构振动响应较为困难。本文提出的神经网络预测方法是基于实测数据而不考虑结构精确的数学力学模型的前提条件下,能够非线性的映射出研究对象的振动特性,即通过机组的振动和尾水脉动的监测数据预测厂房结构的振动,该方法简单、实用并具有一定预测精度,可以用于大型水电站厂房结构的健康监测与振动控制中。⑷水电站厂房的振动是机械力、电磁力、水力脉动共同作用的结果,其动荷载很难测得,结构模态参数识别的难度不言自明。为解决以上困难,本文提出一种新的水电站厂房结构模态参数时域识别方法:RDT+ARMA方法,即采用随机减量法提取停机过程中的结构振动信号,并用ARMA模型识别方法实现厂房结构模态参数的识别;进而提出一种基于环境激励的厂房结构模态参数识别法:EMD+RDT+ARMA方法,即EMD法对环境激励信号进行分解,随机减量法提取IMFs的随机减量特征后,用ARMA模型方法实现结构模态参数的识别。上述两种方法实用,并具有一定的精度,尤其适应于结构低阶模态参数识别。⑸水电站厂房现场振动测试是监测厂房结构动力安全与否的重要途径。本文分析了三峡水电站厂房结构振动原型观测数据,研究了水电站厂房结构振动位移随负荷变化规律,并引入了水电站厂房结构振动控制标准,评价了某发电负荷工况下厂房结构的振动状况。从理论上分析了厂房结构的各种动荷载,计算了结构在动荷载作用下的振动响应,对实测数据和理论计算数据进行了对比,得到一些有价值的内容。更多还原

【Abstract】 With the development of the society and economy, the waterpower, as a regenerated energy source, is increasingly regarded and rapidly developed. By the increase of the capacity and size of the hydro-generator units,the vibration problems of power house structure are catching people eyes, The dynamics problems are the focus, which the domestic and overseas scholars are studying on. Presently, the non-linear dynamics, the fluid-solid interaction, the anti-seism problem and the modal parameters identification method of power house structure should be improvement.Thus this paper studies on the solution of both the positive and negative problem extracted from the projects and the analysis of vibration detection. The principal achievements of this dissertation are as follows:⑴The conventional method for anti-seism study of power station structure is the linear analyses which bases on the response spectrum method. A new anti-seism analyzing method is proposed, which is using the non-linear method and the seism time series method to study the anti-seism problem of power station structure. With the complete process simulation method, it makes displacement field, deformation, stress, cracking field more precise than the conventional method, which can describe the real response of structure. Also, it can shows the time, position and quantity of cracks and weakness of structure with the method. The analysis is of high value on the anti-seism study based on the capability and the damaging estimation of large scale power house structure.⑵The paper combines FEM with acoustic theory with strong interaction method, finite element dynamic equations are built, which provides the basis for analyzing characteristics of power house structure. Vibration characteristics of power house structure at dam toe are calculated with the method and the comparison of characteristics of power house structure at dam toe and in river channel is done to describe the FSI features, and it makes a foundation for safe production and avoiding resonance.⑶The excitation sources of powerhouse structural vibration in hydropower station includes: hydrodynamic forces, mechanical forces and electromagnetic forces. The coupled dynamic action between powerhouse structure and generating units is strong. Because of the vibration system is distinctly non-linear and of dynamic characteristics, it is difficult for conventional methods to solve vibration response of power house induced by internal fluid, but nerve network technology based on test data can highly and non-linearly reflect integrated characteristics of study objectives, regardless of mathematical and mechanical model and various complicated condition. The vibration response of hydropower station can be predicted. The method is simple, effective and of high precision. It is used in health monitor and vibration control fields.⑷The mechanical forces, magnetic forces and current pulse have a big effect on the vibration of power house, it is very difficult to identify model parameters of the structure. To work out a correct solution to identify model parameters and improve the precision, a new method in time domain is proposed: RDT+ARMA method. The method is that it abstracts free-decay signal contents from the signal with random decrement method, when dynamic load released. Model parameters are identified with ARMA method from the signals. And then an important method on identifying mode parameters is proposed. The method (EMD+RDT+ARMA) is based on environmental excitation, which decomposes original signal, the free decay signals of IMFs are abstracted with RDT method, and then model parameters are identified with ARMA method. The methods are simple, effective and of high precision, it is specially fit for the first modes parameters of a powerhouse structure.⑸The vibration test of powerhouse structure is an important path for detecting structure safety. The test data and vibration response changes under various loads of Three Gorges Powerhouse structure are analyzed. The vibration controlling standard is introduced in this paper. The vibration situation of the structure is evaluated. Various dynamic loads are analyzed theoretically and the structure response is calculated under the loads, and the test data is compared with theoretical data. That gets some valuable contents.更多还原