基于BP神经网络法的岩质边坡稳定性评价

发布时间：2018-06-17 07:02

本文选题：BP神经网络 + 岩质边坡　；参考：《南昌航空大学》2017年硕士论文

【摘要】：本文针对公路岩质边坡稳定性问题,利用BP人工神经网络建立岩质边坡稳定坡角预测模型,对稳定边坡角进行预测。同时结合RMR岩体质量分级系统,对岩体稳定性进行分级评价。通过调研,收集大量已建公路边坡工程的工程数据,以其为训练样本建立BP人工神经网络模型。该模型的输入层影响因子采用RMR岩体质量分级系统的参数,并根据参数的评分标准来量化边坡工程勘测数据。对于存在多组节理的岩质边坡,将初始参数值分别进行算术平均调整,加权平均调整以及考虑楔形破坏加权平均调整,将调整后的参数值分别输入到BP神经网络模型中预测稳定边坡角,并将预测值与已建公路的边坡角实际值进行对比分析。如果出现边坡角预测值大于实际值的情况,那么无法直接根据对比结果判断边坡是否稳定,此时利用有限元软件对预测值进行建模分析,验证预测值的可靠性。本研究取得的主要研究成果如下:(1)对RMR岩体质量分级系统参数值进行考虑楔形破坏加权平均调整,得到的岩体质量分级评价结果与已建工程实际情况最接近。(2)将BP神经网络预测结果与已建工程实际数据对比,结果表明利用BP人工神经网络方法预测公路岩质边坡稳定坡角是可行的。(3)有限元软件分析表明,利用BP神经网络预测得到的结果均是正确的,验证了BP神经网络预测稳定边坡角的可靠性。(4)对RMR岩体质量分级系统参数值进行考虑楔形破坏加权平均调整,经过BP神经网络预测得到的稳定边坡角预测值与已建边坡工程边坡角实际值最接近。
[Abstract]:Aiming at the stability of rock slope of highway, a prediction model of stable slope angle of rock slope is established by using BP artificial neural network, and the angle of stable slope is predicted. At the same time, the stability of rock mass is evaluated by RMR classification system. Through investigation and investigation, a large number of engineering data of highway slope engineering have been collected, and BP artificial neural network model has been established with it as the training sample. The parameters of RMR rock mass quality classification system are used for the influence factors of the input layer of the model, and the slope engineering survey data are quantified according to the scoring criteria of the parameters. For rock slopes with multiple joints, the initial parameter values are adjusted respectively by arithmetic average, weighted average adjustment and weighted average adjustment considering wedge failure. The adjusted parameters are input into the BP neural network model to predict the stable slope angle, and the predicted value is compared with the actual value of the slope angle of the built highway. If the prediction value of slope angle is larger than the actual value, it is impossible to judge whether the slope is stable or not directly based on the comparison results. At this time, the finite element software is used to model and analyze the predicted value to verify the reliability of the prediction value. The main research results obtained in this study are as follows: (1) adjusting the RMR rock mass quality classification system parameters considering the wedge failure weighted average. The obtained rock mass quality grading evaluation results are closest to the actual situation of the existing projects.) the BP neural network prediction results are compared with the actual data of the existing projects. The results show that it is feasible to predict the stable slope angle of highway rock slope by BP artificial neural network method. The finite element software analysis shows that the results obtained by BP neural network prediction are all correct. The reliability of BP neural network for predicting stable slope angle is verified. The predicted value of the stable slope angle obtained by BP neural network is the closest to the actual value of the slope angle of the existing slope engineering.
【学位授予单位】：南昌航空大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U416.14

【参考文献】