【作者】 阮永芬；

【导师】 叶燎原；

【作者基本信息】 昆明理工大学 ， 地质工程， 2011， 博士

【摘要】 边坡动力稳定的影响因素是复杂多变的。为了更好的了解动力稳定对各种影响因素的敏感度,进行了多参数影响下的动力稳定分析。同时对高路堤填土进行大量室内及现场试验,来研究环境、荷载、岩土结构特征等因素对岩土工程可靠度的影响。应力和位移是描述物体变形形态的重要物理量,依据监测的应力和位移数据,分析岩土工程的演化规律,预测其发展趋势,及时掌握其稳定情况在工程上具有重要的意义。分析小湾饮水沟堆积体边坡的变形及其特殊的地质情况发现,研究其变形的发生和发展,就要考虑开挖和降雨的影响,即仔细分析堆积体边坡实际的监测资料,寻找出分析研究问题的着手点。要全面了解其地质情况的特殊性,应从空间变异方面着手。所以针对以上问题,本文的研究工作主要包括以下部分：(1)针对高填土路堤边坡,考虑路堤填土材料的物理力学指标、填土高度、坡比、路面宽度、路堤弹性模量、有无地下水以及不同的地震烈度等参数的变化,分别输入美国EL Centro也震波和按规范反应谱拟合的人工合成的地震波,采用静力分析法、拟静力分析法和动力有限元分析法,对高填土路堤边坡稳定性系数的变化规律、动力响应特征及地震荷载作用下边坡内孔压的分布规律进行分析；深入研究地震动峰值加速度沿路堤各高度放大系数的取值,为拟静力法分析计算公路路堤抗震稳定的作用效应和抗力时,如何计入其影响提供计算方法,为山区高填土路堤边坡的抗震稳定性分析提供理论计算依据。(2)对高路堤边坡的填土进行大量的现场及室内试验,分析其抗剪强度指标的影响因素,以及抗剪强度与影响因素之间的相关关系。参数确定后,采用不平衡力传递法和分块极限平衡法,对同一工程边坡所处的三种状态进行稳定性计算,比较分析两种方法的适用特点及存在的问题。(3)由于堆积体空间变异的特殊性,用加权多项式回归法进行球状模型的理论变异函数方法拟合,误差很大,有时甚至无法拟合出变异函数。针对这一问题,结合遗传算法原理,充分利用遗传算法在求解非线性问题时具有全局寻优的特点,拟合出球状模型的理论变异函数参数,并以小湾水电站左岸堆积体边坡的变形监测数据为例进行计算。结果表明用遗传算法来拟合变异函数是可行的,计算结果值与实验变异函数值比较接近,具有较高拟合精度,是非线性、不连续可微模型参数估计的好方法。(4)克立格估值是地质统计学在空间变异上的主要应用,但长期以来,克立格估值权系数一直都没有很好的解决负权值问题,这不但影响着地质统计学的发展,还给实际估值带来非常大的误差,甚至是错误。针对这一问题,采用熵权理论从克立格方程组出发,解决负权值问题,并用小湾坝前堆积体边坡监测资料对其可行性进行了验证。对负权值进行改进时,即考虑了权值系数的修正也考虑了样品点距离的修正。(5)为能充分的体现出堆积体边坡的方向变异性,对所研究堆积体的空间变异性做出定量评价,本文在区域化变量空间分析的基础上,构造出方向综合变异指标Ov来定量描述堆积体边坡的方向变异性。(6)对小湾堆积体边坡进行空间分析后发现,可用克立格法来估计边坡待测点位移。这种方法在没有监测点布置或很难施工监测点时非常有用,不但可以节约建设成本,还可以少受地理位置的限制。可在工程中给予推广(7)基于遗传算法,对改进的BP神经网络的结构参数进行优化,用实际监测的变形资料,对建立的遗传-改进BP神经网络预测模型进行学习训练,用训练好的模型对滑坡变形进行预测。此模型避免了仅用神经网络预测时易陷入局部极小值的缺陷,提高了预测的精度,并考虑了降雨对边坡变形的影响,编制了相应的计算程序,使边坡变形的研究在一定程度上达到智能化。用提出的计算模型对三峡永久船闸三闸首中隔墩岩体的开挖变形、小湾电站边坡变形以及新滩的滑坡变形的预测中进行应用,预测结果均与工程实际稳定性状况和变形实测值相符合,预测的相对误差基本小于5%,从而验证了所提理论方法的可靠性和有效性,能满足中长期的边坡变形预报。(8)为了能很好的研究降雨对大型堆积体边坡变形的影响,应用信息熵理论,通过信息的传递性和相关性原理来分析降雨对边坡变形速率的影响。并以小湾电站左岸引水沟堆积体边坡变形详实丰富的表面监测数据为例进行计算分析,建立了降雨量、时间因子和变形速率之间的信息传递函数。实例分析表明,该方法简便适用,且信息熵在分析边坡变形这类多因素影响的复杂问题时有很大的优势,方法本身对问题的处理不作任何的假定。更多还原

【Abstract】 The influential factors of dynamic stability are very complex and erratic. In order to better understand the sensitivity of dynamic stability to these influential factors, this paper performs a dynamic stability analysis under the influence of multiple parameters. Meanwhile, lots of indoor and field experiments on high embankment filled are carried out to find the influence environment; load and geotechnical structural characteristics have on the reliability of geotechnical engineering.Stress and displacement are two important physical factor used to describe the deformation modality of object. Analyzing the evolutionary rule of geotechnical engineering, forecasting its development trend and knowing its stability according to the monitored data of stress and displacement are of great significance in engineering. This paper, through analyzing the deformation of talus slope at xiaowan watering gully and its special geological information, finds that the effect of excavation and precipitation has to be taken into account when analyzing the happening and development of its deformation. In other word, the actual monitoring data of deformation of talus slope has to be analyzed in detail in order to find a breakthrough to a problem. In order to gain a comprehensive understanding of its special geotechnical situation, one has to approach the issue from the angle of spatial variations. Regarding the above question, this paper includes the following aspects:(1) According to the high embankment filled slope, this paper consider the change of parameters such as the soil index in physical and mechanical properties, embankment height, slope ratio, road width, elastic modulus, having or not having underground water, and different seismic intensity. While doing dynamic comparative analysis, the earthquake wave of EL Centro of Unite State and spectrum synthesize artificial earthquake wave is inputted, we adopts the method of static analysis, pseudostic analysis, static and dynamic finite element analysis method to make a comparison analysis the static and dynamic stability of high embankment filled slope, so as to assess their influence to stable coefficient of the high embankment filled stability ,research dynamic response characteristic and time history curve of pore pressure distributing rule under earthquake load; then this paper deeply discuss how to get the value of seismic acceleration dynamic magnification coefficient along the height of fill while adapting the pseudostic analysis method to calculate the effect and resistance on highway earthquake stability, this has provided the reliable basis for the high embankment filled slope in the mountainous area earthquake resistance stability analysis design.(2) In order to analysis the influence factors of shear strength and correlativity between shear strength and influence factors, we have undergone extensive the indoor and field tests to high embankment filled. While adapting two different slope blocks calculate method of the block limiting equilibrium method and transfer coefficient method to carry out stability calculate and analysis for the three kinds of states slope engineering, and then analysis the features and problems existed.(3) Due to uniqueness of talus slope’s spatial variations, the fitting error of the spherical model variation function is bigger using the weighted polynomial regression method, sometimes it is impossible to fit variation function .In order to solve this problem, the genetic algorithm is applied to fit the parameter of spherical model variation function, because the genetic algorithm have the advantage on seeking priorities in fully exert when use it to solves the non-linear problems, furthermore, the data of the accumulation body with high slope on the left bank in Front of Xiaowan hydro plant Dam is employed to calculate and test as an example. By comparing and analyzing the dates on the example from different angles, it demonstrates that the new method is convenient, universal and precise. It is an excellent method for estimating parameter of non-liner, incontinuity or differentiable model parameter.(4) The estimation of Kriging is a main application of geostatistics in spatial variability, but for many years, there is a problem that the nonnegative constrains are not taken into consideration in the weights of estimation. This is not only influenced further development of Kriging, but also impacted on the estimation of actual results. To solve this problem, the entropy weight theory models calculating the weights of estimation in which the nonnegative constrains are taken into consideration are proposed to overcome the malpractices existed in the related Kriging equations. Finally, the empirical analysis and verification of the algorithm is provided and the results show that it is feasible by the weight theory algorithm. Both of the correction of coefficient of weight and the correction of sample point of distance are taken into consideration in improving the negative weight.(5) In order to fully in the person of orientation variability and make out quantitative evaluation for the spatial variability of talus slope, based on spatial analysis of regional variable parameters that the directional synthetic variance index Ov are present, then using this index quantitative description the directional variation of talus slope.(6) Through the research of XiaoWan talus slope, it’s discovered that the displacement of slope can estimated with the help of the theory and method of Kriging. This method is very useful when there is not monitoring point or it is difficult to construct monitoring point; It not only can save a great deal of construction cost, but also is hardly restricted by geographic position. It should be generalized in engineering.(7) Based on the genetic algorithm, this article put forward a revised model of BP neural network and uses it to predict and calculate landslide deformation. Its steps are as follows:The genetic algorithm are adopted to optimize the architectures parameters of BP neural network, and then the practical deformation data are employed to test the model of the genetic-improved BP neural network modeling. The modeling reduces the probability of getting into the local minimum while employing neural network to forecast the landslide deformation and it significantly promotes the forecast calculation precision, it consider the influence of rainfall for slope deformation and create the calculate program and making the study of deformation show better artificial intellective. Then we applied the modeling in training for predication calculation and analysis of the landslide deformation to the measurement data of three actual engineering, the results show that the forecasting value and actual measurement value are very closely, and the relative error falls below 5%. It demonstrated the theory and method has high reliability and validity. So it can be concluded that this modeling is of high forecasting precision and can satisfy the request of deformation prediction of landslide in practical projects and in medium and long term.(8) In order to make a good research on the influence of rainfalls precipitation to the high talus slope deformation, the method of information entropy is adopted by using the principle of information transfer and relativity to analyze and calculate the rainfall precipitation influence on the deformation velocity. Moreover, taking the full and accurate data of the accumulation body with high slope on the left bank in front of Xiaowan Dam as an example, the information transfer function between rainfall and deformation is established. The analysis of engineering examples shows that this new method is simple and easy to use. It also demonstrates its great advantage for the analysis of such complex slope deformation problem that has a lot of influencing factors. The method does not make any hypothesis before putting into consideration.更多还原