ICS-Elman神经网络在边坡稳定性分析中的应用  

Application of ICS-Elman Neural Network on Slope Stability Analysis

作　　者：荣光旭 马清清 高延超 范小倩[1] 梁俊俊 RONG Guangru;MA Qingqing;GAO Yanchao;FAN Xiaoqian;LIANG Junjun(School of Geology and Construction Engineering,Anhui Technical College of Industry and Economy,Hefei,Anhui 230051,China;School of Engineering and Technology,Guangdong Polytechnic Institute,Zhongshan,Guangdong 528400,China;Chengdu Center,China Geological Survey(Geosciences Innovation Center of Southwest China),Chengdu,Sichuan 610081,China;China Water Resources Pearl River Planning,Surveying&Designing Co.,Ltd.,Guangzhou,Guangdong 510610,China)

机构地区：[1]安徽工业经济职业技术学院地质与建筑工程学院,安徽合肥230051 [2]广东理工职业学院工程技术学院,广东中山市528400 [3]中国地质调查局成都地质调查中心(西南地质科技创新中心),四川成都610081 [4]中水珠江规划勘测设计有限公司,广东广州510610

出　　处：《矿业研究与开发》2025年第3期213-221,共9页Mining Research and Development

基　　金：四川省自然科学基金项目(23NSFSC0297);安徽省高校自然科学研究项目(2022AH052670,2023AH052675)。

摘　　要：为提高边坡稳定性预测准确率,提出了一种基于多策略改进布谷鸟(ICS)算法优化Elman神经网络的边坡稳定预测模型。首先,利用Sinusoidal混沌初始化种群,引入动态步长控制量和对发现概率进行1%步进离散化等策略,改进布谷鸟算法收敛速度慢、精度较差的缺陷,在7个基准函数上的仿真结果表明,与布谷鸟(CS)算法相比,ICS算法的寻优能力和收敛速度有较大提高。然后,利用ICS算法对Elman神经网络的权值和阈值进行优化,构建最优的边坡稳定性预测模型。结果表明,ICS-Elman模型对边坡稳定性安全系数预测结果相对误差范围为-2.81%~6.98%,均方根误差、平均绝对误差分别为0.2750,0.3422,与Elman、CS-Elman模型相比,ICS-Elman模型具有较好的稳定性和精度;与CPSO-BP神经网络相比,ICS-Elman神经网络预测值相对误差范围为-1.57%~1.25%,预测精度更好。In order to improve the accuracy of slope stability prediction,a slope stability prediction model based on multi strategy improved cuckoo search algorithm(ICS)optimized Elman neural network was proposed.Firstly,the population was initialized using Sinusoidal chaos,and strategies such as dynamic step size control and 1%step discretization of discovery probability were introduced to improve the slow convergence speed and poor accuracy of the cuckoo search algorithm.Simulation results on 7 benchmark functions show that compared with the CS algorithm,the optimization ability and convergence speed of the ICS algorithm are significantly improved.Then,the ICS algorithm was used to optimize the weights and thresholds of the Elman neural network,and the optimal slope stability prediction model was constructed.The results show that the relative error range of the ICS-Elman model for predicting the safety factor of slope stability is-2.81%-6.98%,with root mean square error and mean absolute error of 0.2750 and 0.3422,respectively.Compared with the Elman and CS-Elman models,the ICS-Elman model has better stability and accuracy.Compared with the CPSO-BP neural network,the ICS-Elman neural network has a relative error range of-1.57%-1.25%in predicting values,indicating better prediction accuracy.

关 键 词：边坡稳定性 ELMAN神经网络 改进布谷鸟算法 安全系数 

分 类 号：TD854.6[矿业工程—金属矿开采] TP183[矿业工程—矿山开采]