【作者】 辜晋德；

【导师】 练继建；

【作者基本信息】 天津大学 ， 水利水电工程， 2010， 博士

【摘要】 消力池作为高坝泄洪时下游河床的防护结构,其自身的稳定性不论对消能防冲还是整体水工结构的安全都具有重大意义。消力池防护结构的工作和破坏机理十分复杂,相关的影响因素很多,因此消力池的稳定性一直是一个重要的研究课题。本文结合模型试验和数值分析,研究高坝消力塘水动力荷载特性。对缝隙内水流脉动压力传播进行了数值和试验研究,基于试验数据建立了简化二维数值模型,并基于此模型对脉动上举力各项形成因素进行了分析。利用支持向量机基于结构风险最小化原则的特点,建立了板块上举力预测模型。主要成果包括:(1)介绍了底流和挑流这两种主要消能形式的特点,分析了两种消能形式下消力池的典型脉动压力特性,包括幅值特性、频谱分布以及空间积分尺度沿程变化规律。并结合工程实例,介绍了几种新型消能工和消力池防护措施,并通过实测数据对比了采用消能工前后消力池的脉动特性变化,从而直观的表现了新型消能工及防护措施的作用。(2)对几个常见的研究缝隙水脉动压力的模型进行了分析,基于水体振荡模型对缝隙内脉动压力的分布规律进行了推导;并以单一余弦波为算例,运用瞬变流模型计算了缝隙中点脉动压力的变化过程。分析了通缝形成后的缝隙水脉动压力传播规律,并将瞬变流模型的计算结果与水体振荡模型计算结果进行了对比,分析了两种方法各自的特点。通过水槽内的底流试验,对板块上下表面脉动压力进行了分析。基于试验数据计算了脉动压力的概率分布以及频谱分布,总结了缝隙内水流脉动压力特性。通过小波分解,将实测脉动信号高低频分离,研究了低频范围内脉动压力的相关关系以及缝隙内脉动压力的转换系数Cg。(3)基于一维水体振荡模型的计算结果,提出了一种简化的二维模型。这个模型认为板块底部脉动压力由其上表面的几个点脉动压力决定,并考虑脉动压力在进入缝隙过程中的衰减,引入一个衰减系数k。通过实测数据反算验证了k的数值范围。进一步应用简化模型对上下表面中点脉动压力的相关系数的验证。基于该简化模型,对板块脉动上举力的相关要素进行了推导,并尝试仅通过上表面多点脉动压力及其相关关系估算板块可能承受的脉动上举力。通过在水槽中的二元水跃实验对理论分析进行了验证,对比结果表明简化二维模型基本能够反映上举力的变化特性。(4)将机器学习理论中的神经网络和支持向量机应用到板块的脉动上举力预测中来,分别构建了上举力的预测模型,通过实测数据检验了模型的预测能力,并基于预测结果对两种模型进行了比较。比较结果发现,两者均达到了较好的效果,说明基于机器学习方法的上举力预测系统有较好的应用前景。基于官地水电站的模型试验数据构建了BP神经网络上举力预测模型,经过数据的检验,模型对上举力的预测具有一定的精度。更多还原

【Abstract】 The safety of protecting structure in high dam plunge pool is the key to success or failure of energy dissipation, there is are many destructive examples in and abroad. Working and destruction mechanism of protecting structure in high dam plunge pool is very complicated, mechanism of transmission of fluctuating pressure in gap is invoved ,also corelationship of dynamic loads on up and down surface of slabs, so the uplift force is very hard to forecast. Model test and numerical analysis were studied in this paper, and hydrodynamic characteristics in high dam plunge pool is researched.fluctuating pressure in gap is studied by model test and numerical analysis, a simplified two-dimension model is build to reveal the characteristics of fluctuatingpressure in gap by model experiment data, then factors of uplift force are calculated using this simplified model. making the best use of the support vector machine characteristic of structure risk minimization principle, the upport vector machine safety monitoring model of slab-upliftforce is established. Main achievement is included as follows:(1)Two kinds of energy dissipation mode and their characteristics were introduced, fluctuating pressure in stilling basins was analyzed, including their amplitude characteristics, spectrum distribution and change law of spatial integral scales . some new kinds of energy dissipator and protective measures of stilling basins were introduced combining with engineering examples, by comparing data from model test,effect of these dissipater was showed.(2)Research models of fluctuating pressure in gap were studied, distribution law of pulsating pressure in gap was deduced based on oscilation model. Use cosine wave as an example, process of fluctuating pressure was caculated using transient flow model. Canculation results of these two different model were compared, characters of two method were analyzed.Fluctuations on upsurface and down surface were achieved by experiments of hydraulic jump in flume tank. Probability distribution and spectrum of fluctuating pressure were studied basing on the achieved data, and characters of fluctuations in gap was summarized. High and low frequency of experiment data were separated using wavelet decomposition. Corelationship between fluctuations in two ends of gap and conversion factor of fluctuations in gap Cg were researched.(3)Basing on conclusions of one dimension water oscillation model, a new simplified two dimension model is presented in this paper. It is considered that pressure fluctuation under slab can be determined by several points on the up surface, considering attenuation of pressure during transmission in gap,a attenuation coefficient K was introduced,and value range of K is calculated by experiment data. Corelation coefficient between fluctuations on the middle of up and down surface is verified using the simplified two dimension model. factors about uplift force were deduced using this model, and it is a new attempt to forecast uplift force by measuring fluctuating pressure on upsurface of slab using this model. Theoretical results and experiment results are compared by model test of hydraulic jump in flume tank.(4)BP neural network and support vector machine, which belong to machine learning theory, were used to forecast fluctuating uplift force by constructing forecast system, ability of the systems was test and compared by model test data. It is said that both these two uplift force forecast systems has good applied foreground for their good effect . a uplift force forecast model was construct by data from Guandi model test, and this model is precise to some extent in uplift force forecasting.更多还原