摘要
确定岩体物理力学参数的方法有很多,近年来最常用的是非线性分析法。鉴于各种非线性分析法的局限性,基于BP神经网络,利用遗传算法(Genetic Algorithm,GA)对神经网络的初始权值和阈值进行优化从而改进预测方法,并利用工程实例验证其有效性。以前坪水库右坝肩碎裂结构岩体为研究对象,利用52组岩体的物理力学参数试验结果,基于改进的神经网络程序,建立了岩体力学参数预测模型,并将岩体力学参数的预测结果与实测结果相比较,以验证模型的有效性。结果表明:GA-BP神经网络模型可以解决常规预测方法中的结构不合理、个别结果误差较大等问题;预测结果的精度可达到0.95以上,能够有效预测前坪水库坝址区岩体的力学参数。该方法可为碎裂结构岩体力学参数的优化选取提供一种新途径。
There are many methods to determine the physical and mechanical parameters of rock mass,and the most commonly used method is nonlinear analysis in recent years.In view of the limitations of various nonlinear analysis methods,based on the BP neural network,this paper adopts the Genetic Algorithm(GA)to optimize the initial weights and thresholds of the neural network to improve the prediction method,and uses engineering examples to verify its effectiveness.The rock mass with fragmented structure on the right abutment of Qianping reservoir is used as the research object.Using the test results of 52 sets of rock masses,based on the improved neural network program,a prediction model of rock mass mechanical parameters is established.The predicted results of rock mass mechanical parameters are compared with the measured results to verify the validity of the model.The results are as follows.The GA-BP neural network model can solve the problems of unreasonable structure and large error in individual results in conventional prediction methods.The accuracy of the prediction results can reach more than 0.95,which can effectively predict the mechanical parameters of the rock mass in the Qianping reservoir dam site area.This method can provide a new approach to optimize the selection of mechanical parameters of fractured rock mass.
作者
刘汉东
张世英
LIU Handong;ZHANG Shiying(Institute of Geotechnical Engineering and Hydraulic Structure,Henan Key Laboratory of Geotechnical and Structural Engineering,North China University of Water Resources and Electric Power,Zhengzhou 450046,China)
出处
《华北水利水电大学学报（自然科学版）》
2020年第6期78-84,共7页
Journal of North China University of Water Resources and Electric Power：Natural Science Edition
基金
国家重点研发计划资助项目(2019YFC1509704)。
关键词
碎裂结构岩体
BP神经网络
遗传算法
岩体力学参数
fractured rock mass
BP neural network model
Genetic Algorithm
rock mass mechanical parameters
作者简介
刘汉东(1963-),男,山东菏泽人,教授,博导,博士,从事岩土力学方面的研究。E-mail:liuhandong@ncwu.edu.cn;张世英(1997-),女,河南南阳人,硕士研究生,从事地质工程方面的研究。E-mail:1908299746@qq.com。
