决策树算法在土质边坡稳定性评价中的应用

发布时间：2018-03-27 16:20

本文选题：土质边坡　切入点：决策树　出处：《南华大学》2015年硕士论文

【摘要】：长期以来,边坡稳定性分析与评价始终是岩土工程中的重要研究领域。从极限平衡法到当前广泛使用的不确定性分析法,研究者们针对其进行了大量研究。不同的方法从不同的角度对边坡的稳定性及相关问题进行了研究,如可靠性分析方法从数理统计的角度进行边坡稳定和失效的概率分析;极限平衡法从坡体极限状态的角度,建立静力平衡方程来进行边坡稳定性状态的求解;模糊综合评价法则从考虑信息的随机性和模糊性角度,运用模糊数学进行边坡的稳定性分析。但不同的方法有其各自的局限性。实际上,边坡工程是一个不断变化着的、开放的复杂巨系统。确定性计算不能很好地概括其复杂性,而大多非确定性计算方法,如学习向量量化神经网络,网络的预测精度往往与数据的差异性有关。因此,建立一个对样本数据要求低且通用性能好的边坡稳定性评价模型具有重要意义。针对土质边坡稳定性影响因素易于量化且高度非线性的特点,本文将决策树算法引入到土质边坡稳定性评价中,形成了一种以决策树为基础的综合评价方法。具体包括:(1)简要阐明了本文研究的意义,对国内外的土质边坡稳定性研究进行了总结。系统地阐述了常用的土质边坡稳定性的评价方法,并对其优缺点进行了分析和归纳。(2)简要介绍了决策树的广泛运用。系统地阐述了常用的决策树分类算法,并进行对比分析,指出其各自的适用范围。论证了决策树分类器运用于土质边坡稳定性评价中的可行性。(3)首次将决策树算法运用于土质边坡稳定性评价,建立了基于C4.5决策树算法的土质边坡稳定性评价模型。同时建立了基于LVQ神经网络、基于BP神经网络的土质边坡稳定性评价模型,对多种模型进行对比分析,证明了将决策树算法运用于土质边坡稳定性评价的优越性。
[Abstract]:For a long time, slope stability analysis and evaluation has always been an important research field in geotechnical engineering. Researchers have done a lot of research on it. Different methods have studied slope stability and related problems from different angles, such as reliability analysis method from the point of view of mathematical statistics of slope stability and failure probability analysis; From the angle of the limit state of slope body, the limit equilibrium method establishes the static equilibrium equation to solve the slope stability state, and the fuzzy comprehensive evaluation law considers the randomness and fuzziness of the information. Fuzzy mathematics is used to analyze slope stability. However, different methods have their own limitations. In fact, slope engineering is a changing and open complex giant system. However, the prediction accuracy of most uncertain computing methods, such as learning vector quantization neural network, is often related to the difference of data. It is of great significance to establish a slope stability evaluation model with low sample data requirement and good general performance, aiming at the characteristics that the influencing factors of soil slope stability are easy to quantify and highly nonlinear. In this paper, the decision tree algorithm is introduced into the evaluation of soil slope stability, and a comprehensive evaluation method based on decision tree is formed. The research on the stability of soil slope at home and abroad is summarized, and the commonly used methods for evaluating the stability of soil slope are systematically expounded. The advantages and disadvantages of decision tree are analyzed and summarized. The wide application of decision tree is briefly introduced. The commonly used decision tree classification algorithms are systematically expounded and compared. The feasibility of applying decision tree classifier to the stability evaluation of soil slope is demonstrated. The decision tree algorithm is applied to evaluate the stability of soil slope for the first time. The evaluation model of soil slope stability based on C4.5 decision tree algorithm is established, and the evaluation model of soil slope stability based on LVQ neural network and BP neural network is established. The superiority of applying decision tree algorithm to the evaluation of soil slope stability is proved.
【学位授予单位】：南华大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU43

【相似文献】