摘要
为提高边坡稳定性的预测精度,提出一种基于多策略改进的麻雀搜索算法(MISSA)优化支持向量机(SVM)的边坡稳定性预测模型.选取容重γ、黏聚力c、内摩擦角Ф、边坡角φf、边坡高度H、孔隙压力比ru等6 个代表性特征作为模型的预测指标.针对麻雀优化算法(SSA)存在的收敛速度慢、精确度不高、易陷入局部最优等问题,引入一维复合混沌映射、正余弦算法(SCA)、Levy飞行机制和步长因子动态调整等策略进行优化改进,构建基于MISSA-SVM的边坡稳定性预测模型.将MISSA-SVM模型应用到大溪滑坡等9 组边坡工程实例进行验证.结果表明:MISSA-SVM模型的准确率、精确率、召回率、F1 分数、均方误差(MSE)和曲线下面积(AUC)分别达到 96.29%、92.3%、100%、0.96、0.016 和0.967,均优于SSA优化的SVM模型和BP 模型,预测结果与实际边坡状况完全吻合,表明MISSA-SVM模型具有较强的泛化能力.
Abstract
In order to further improve the prediction accuracy of slope stability,a slope stability prediction model based on MISSA optimized SVM was proposed.Six representative indexes,including bulk density(γ),cohesion(c),internal friction angle(Ф),slope angle(φf),slope height(H)and pore pressure ratio(ru)were selected as the prediction indexes of the model.In response to the problems of slow convergence speed,low accuracy,and susceptibility to local optima in the sparrow optimization algorithm(SSA),strategies such as one-dimensional composite chaotic mapping,SCA,Levy flight mechanism,and dynamic adjustment of step size factor are introduced for optimization and improvement.A slope stability prediction model based on MISSA-SVM was constructed.The MISSA-SVM model was applied to 9 groups of slope engineering examples,such as the Daxi landslide,for verification.The results show that the accuracy,precision,recall,F1 score,mean square error(MSE)and area under the curve(AUC)of the MISSA-SVM model reach 96.29%,92.3%,100%,0.96,0.016 and 0.967,respectively,which are better than the SSA-optimized SVM model and BP model,and the prediction results are completely consistent with the actual slope conditions,indicating that the MISSA-SVM model has strong generalization ability.
基金项目
云南省科技重大专项(202202AG050014)
云南省创新团队项目(202105AE160023)
NETL
NSTL
万方数据