基于人工神经网络及优化方法的岩体力学参数反分析法综述  

Review of Back Analysis of Rock Mass Mechanic Parameters Based on Artificial Neural Network and Optimization Method

作　　者：刘毅聪 刘祚秋[1] Liu Yicong;Liu Zuoqiu(Department of Applied Mechanics and Engineering,School of Engineering,Sun Yat-Sen University Guangzhou 510006,China)

机构地区：[1]中山大学工学院应用力学与工程系,广州510006

出　　处：《广东土木与建筑》2018年第9期31-35,共5页Guangdong Architecture Civil Engineering

摘　　要：岩体力学参数的合理选取对岩体工程的设计计算具有重要意义。选取岩体力学参数的方法有很多,基于人工神经网络及优化方法对岩体力学参数的反分析法不需要事先建立数学方程并且可以更好地反映输入因子与岩体力学参数的非线性关系,具有较高的研究价值。首先对岩体力学参数的基本概念及其选取方法以及BP神经网络、RBF神经网络的基本原理进行概述,然后对神经网络样本组织及优化过程、神经网络的输入与输出(即岩体力学参数的影响因子与岩体力学参数)、优化神经网络的方法以及反分析结果的检验进行具体的说明,并且对该方法进行分析与评价,最后提出该方法的一些不足以及对该方法研究的展望。The reasonable selection of rock mass mechanic parameters is significant for the rock mass engineering. There are many methods to select rock mass mechanic parameters,among which the back analysis of rock mass mechanic parameters based on artificial neural network and optimization method deserves research because this method dose not need previous math equations and can reflect the non-linear relation- ship between the input and the rock mass mechanic parameters. In the passage,the concepts and the selection of rock mass mechanic parame- ters and the neural network will be summarized at first. Next,the samples of neural network,the input and the output of neural network and the optimization method of neural network as well as the test of the back analysis results will be introduced in detail. In the end,the discus- sion,disadvantages and the expectation of the method will be put forward.

关 键 词：岩体力学参数 反分析 人工神经网络 优化方法 

分 类 号：TU452[建筑科学—岩土工程]