基于支持向量机的土石坝坝坡稳定可靠度分析

发布时间：2018-01-13 19:19

本文关键词：基于支持向量机的土石坝坝坡稳定可靠度分析　出处：《广西大学》2015年硕士论文　论文类型：学位论文

【摘要】：传统的土石坝坝坡稳定可靠度分析大多是基于极限平衡法得到的功能函数来实现的。然而这样得到的功能函数即不能真实的反映土体的受力状态,又未能考虑坝体内的渗流作用。这样就会导致计算结果偏大。为解决这些问题,本文采用替代模型来逼近真实的功能函数,从而利用替代模型来计算土石坝坝坡稳定可靠度。由于支持向量机具有出色的小样本学习能力,能够利用少量的样本构造逼真的替代模型,因此本文将支持向量回归机与传统的可靠度计算方法结合提出SVR-MC法。将SVR-MC法与传统的可靠度分析方法对比发现其计算精度能够满足工程需求,且计算效率高。将SVR-MC法应用于土石坝可靠度计算时需要利用少量样本对支持向量机进行训练。由于拉丁超立方抽样能够均匀、随机的抽取样本,因此利用其对基本随机变量进行抽样,并利用Abaqus软件,使用强度折减法计算出相应的坝坡安全系数,从而得到分布较均匀的训练样本。采用有限元强度折减法计算坝坡安全系数时能够考虑坝体内渗流场和应力场的耦合作用,因此利用支持向量机得到的代替模型能更好的反映真实的功能函数。最后,把提出的基于支持向量机的土石坝可靠度计算方法应用于澄碧河水库土石坝可靠度分析中。以澄碧河水库大坝加固前为研究对象,以大坝上游水位为185m,下游水位为132为计算工况。将计算结果与传统的基于极限平衡法的可靠度计算结果进行比较,发现考虑渗流作用计算得到的可靠指标要比不考虑渗流作用的结果小一些。结果表明：基于支持向量机的土石坝可靠度分析方法计算精度满足工程要求,计算结果合理、准确,对土体分布复杂的土石坝进行可靠度分析具有很大潜力,值得推广使用。
[Abstract]:The traditional stability reliability analysis of earth-rock dam slope is based on the function function obtained by the limit equilibrium method. However, the function function can not truly reflect the stress state of soil. In order to solve these problems, the substitution model is used to approximate the real function function. Therefore, the stability reliability of earth-rockfill dam slope can be calculated by using the substitution model. Because of the excellent learning ability of small sample, the support vector machine can construct a realistic replacement model with a small number of samples. In this paper, the support vector regression machine is combined with the traditional reliability calculation method and the SVR-MC method is put forward. Comparing the SVR-MC method with the traditional reliability analysis method, it is found that the accuracy of the method can meet the needs of engineering. Beg. When applying SVR-MC method to the reliability calculation of earth-rockfill dam, a small number of samples should be used to train the support vector machine. Because the Latin hypercube sampling can be uniform, the samples can be taken randomly. Therefore, the basic random variables are sampled, and the corresponding safety factor of dam slope is calculated by using the strength reduction method and Abaqus software. Thus, the uniform distribution of training samples can be obtained. The coupling effect of seepage field and stress field in dam body can be considered when the safety factor of dam slope is calculated by using the finite element strength reduction method. Therefore, the substitute model obtained by support vector machine can better reflect the real function. Finally. The reliability calculation method of earth-rock dam based on support vector machine is applied to the reliability analysis of earth-rock dam in Chengbihe Reservoir. Before the dam is strengthened Chengbihe Reservoir is taken as the research object and the upstream water level of the dam is 185m. The downstream water level is 132 as the calculation condition. The calculation results are compared with the traditional reliability calculation results based on limit equilibrium method. It is found that the reliability index calculated by considering seepage action is smaller than that without seepage effect. The results show that the reliability analysis method based on support vector machine can meet the requirements of engineering. The calculation results are reasonable and accurate, and the reliability analysis of earth and rock dams with complex soil distribution has great potential and is worth popularizing.
【学位授予单位】：广西大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TV641

【参考文献】