节点文献

泄流结构水力拍振机理及动态健康监测技术研究

Research on Hydraulic Beat Mechanism and Dynamic Health Monitoring Method of Flood Discharge Structures

【作者】 李成业

【导师】 练继建;

【作者基本信息】 天津大学 , 水利水电工程, 2013, 博士

【摘要】 我国重大水利水电工程的坝高、泄量、泄洪功率均居世界领先水平。由于高速水流、水气二相流、水流——结构相互作用的复杂性,使得泄流结构在长期泄流荷载作用下极易发生疲劳或振动破坏。因此,开展水流诱发结构的振动机理、泄流结构流激振动响应的时频分析、滤波降噪、工作性态识别以及泄流结构动态健康监测技术等领域的研究,对泄流结构的设计和安全运行具有重要的指导作用。本文主要开展以下三个方面的研究:(一)泄流结构水力拍振机理研究。在总结分析泄流结构水流荷载特性的基础上,应用时域分析、频谱分析以及时频分析方法,研究了冲击荷载、谐波激励荷载、宽带随机荷载以及白噪声荷载激励下结构的响应特性。通过开展泄流结构原型观测、流激振动系列模型试验以及有限元计算,分析了水流荷载、泄流结构自身特性以及流激振动响应三者之间的关系,首次提出并解释了泄流结构的水力拍振现象,总结了泄流结构流激振动的响应机制。(1)建立一系列室内流激振动模型试验,将相关性分析、谱分析以及时频分析方法应用到试验数据的分析过程中。通过分析不同固有频率的试件在水流荷载作用下的动态振动特性,研究了泄流结构在水流荷载作用下的随机共振形式,提出水力拍振的概念;(2)分析二滩拱坝在泄流荷载作用下,坝身不同部位的振动响应机理,解释响应中出现的水力拍振现象以及在此基础上形成的拍振现象,为高拱坝体型设计和运行管理提供依据;(3)根据水力拍振理论和二滩、蜀河、炳灵等水电站的原型观测分析,明确了水力拍振现象发生的各种情况,较全面地总结了泄流结构流激振动响应的形成机制。(二)基于时频分析的泄流结构滤波降噪及工作性态识别方法研究。针对环境激励下的模态辨识方法要求激励荷载必须满足白噪声的特性、且不能在时频域同时进行识别的缺点,本文开展了基于时频分析的泄流结构滤波降噪和工作性态识别方法研究。(1)根据HHT时频分析的特点,研究了泄流结构动力响应中与模态信息相关的共振分量的提取方法,直接应用共振分量进行工作性态识别,能够避免受迫振动分量对识别结果的影响;(2)通过研究EMD与小波阈值联合的滤波降噪方法,提出了含噪信号EMD分解后各分量中噪声标准差的确定方法,解决了小波阈值计算中噪声标准差计算难的问题;(3)研究了基于时频分析的泄流结构工作性态识别方法,并将该方法应用于二滩水电站拱坝和映秀湾水电站拦河闸结构的模态参数辨识中,取得了较好的效果。(三)泄流结构动态健康监测技术研究。针对现有健康监测方法存在的不足,本文开展了泄流结构损伤敏感特征指标以及融合决策方法研究。(1)研究和总结了基于波形、标准差、时间序列模型、正则化频率变化率,时频域指标、坐标模态准则、模态振型与频率、模态曲率等损伤敏感指标,建立泄流结构动态健康监测指标体系;(2)研究了基于数据融合技术的决策计算方法,针对D-S证据理论合成规则存在的不足,引入Jousselme距离和证据可信度,改进了D-S证据理论的组合规则。泄流激励下结构的损伤模型试验结果表明,文中方法避免了单一损伤指标由于传感器自身条件、噪声等因素所引起的误判断,有效地降低了错误证据对最终融合结果的影响,适用于泄流结构动态健康监测过程中的损伤定位;(3)结合导墙结构的工作特点,研究了基于HHT时频域指标的稳定性、敏感性和抗噪性,提出了一种适合导墙结构损伤在线监测的新方法。

【Abstract】 The height, discharge, flood-discharge power of hydraulic engineering in China isin a leading position of the world. With the influence by high velocity flow, water-airtwo-phase flow and waterflow-structure coupling, the flow pulsating pressure is verylikely to result in structural vibration or even fatigue destruction of flood dischargestructures. It is an important guidance for design and safe operation of flood dischargestructures to research flow-induced vibration mechanism, time-frequency analysismethod for vibration responses, filtering method, working state identification methodand dynamic health monitoring method. The following three main areas are studied:Firstly, research on hydraulic beat mechanism of flood discharge structures. Thecharacteristics of the flow pulsating pressure on the flood discharge structures aresummarized, and then time-domain analysis, spectrum analysis, time-frequencyanalysis method are utilized to study the structural vibration responses induced byimpact load, harmonic load, wind-band random load and white noise load,respectively. Based on the prototype tests, flow-induced vibration model and finiteelement analysis, the relationship among flow discharge excitation, hydraulicstructures’ dynamic characteristics, and vibration response has been studied. Then thehydraulic beat phenomena of flood discharge structures has been put forward andexplained.(1) Series model tests of flow-induced vibration have been built, andcorrelation analysis, spectrum analysis, and the time-frequency analysis method areapplied to study the vibration testing data. According to vibration characteristics ofdifferent models induced by flow load, this random resonance responses phenomenonhas been researched, and a new concept “hydraulic beat” has been put farward.(2)The flow-induced vibration responses from each measuring point arranged on Ertandam have been analyzed. It can be concluded that the hydraulic beat phenomena, andeven beat vibration will occur in the closely frequency structures like high arch dam.(3) Based on the hydraulic beat theory and prototype tests for Ertan, Shuhe, Binglinghydropower station, the formation mechanism of hydraulic beat and flow-inducedvibration for flood discharge structures have been summarized.Secondly, study on the methods for filtering and working state identification offlood discharge structures based on time-frequency analysis. There are two difficultproblems for modal parameters identification under ambient loading: the first one isthat the excitation load must satisfy the characteristics of white noise, and the second is that the identification range is restricted in time domain or frequency domain.Aimed at these issues, the methods for filtering and working state identification basedon time-frequency analysis are provided.(1) The time-frequency analysis methodbased on HHT is utilized to extract the resonant response, which is closely related tothe structural modal parameters. Then resonant responses are used to working stateidentification of flood discharge structures, and the adverse effect caused by forcedvibration response will be avoided.(2) According to the characteristics of IMFsdecomposed from white noise, the calculation method of noise standard deviation ineach IMF is studied, and the filtering method based on EMD and Wavelet-threshold isimproved.(3) The methods based on time-frequency analysis studied in the paper areapplied to the operational modal parameter identification of Ertan arch dam andYingxiuwan hydropower station sluice respectively. The results show that thesemethods have good applicability in vibration responses analysis of flood dischargestructures.Thirdly, the dynamic health monitoring method for flood discharge structures. Inview of some shortcomings existed in health monitoring method, dynamic damageindexes and data fusion decision method are researched in this paper.(1) The dynamicdamage indexes based on vibration waveform, standard deviation, time series model,normalized change ratio of frequency, time-frequency analysis, coordinate modalcriterion, mode shapes and frequencies, mode curvature have been summarized, theindex system of dynamic health monitoring has been established for flood dischargestructures.(2) Due to some issues in the combination rule of D-S evidence theory, animproved method based on Jousselme distance and evidence reliability is put forward.The results of damage model test under flood discharge excitation demonstrates thatthe accurate decision can be obtained based on few evidences, the errors originatedfrom sensor and noise are avoided, and the influence by the wrong evidence iseffectively reduced. The method mentioned in this paper is suitable for damagelocation of dynamic health monitoring.(3) According to working characteristics ofguide walls, and the analysis results for stability, sensitivity, anti-noise of HHTmonitoring indexes, a new monitoring method for guide wall has been put forward.

  • 【网络出版投稿人】 天津大学
  • 【网络出版年期】2014年 12期
节点文献中: 

本文链接的文献网络图示:

本文的引文网络