摘要
安全系数是用来评估边坡稳定性的重要指标之一,复杂的边坡系统导致安全系数预测存在不确定性。因此,为了获得更加可靠的安全系数,同时解决鹈鹕算法(POA)随着迭代次数的增加易陷入局部最优的缺点,提出了一种融合多策略的鹈鹕算法(IPOA)与支持向量机(SVR)结合的回归模型来预测边坡安全系数。首先,融合多策略将原始的鹈鹕算法进行改进;再运用改进的鹈鹕算法与支持向量机结合,选取六个影响因素作为IPOA-SVR模型的输入层指标并对模型进行训练,得到IPOA-SVR边坡稳定性预测模型;最后,分别与KNN、RF和Adaboost模型对比,并计算各个模型在训练集和测试集上的均方误差(MSE),以此来验证IPOA-SVR模型的优越性。实验结果显示:与其他模型相比,IPOA-SVR模型寻优性能强,在测试集上的均方误差为0.030 9、相关系数为0.91,说明本文对POA算法所用策略的有效性,IPOA-SVR模型可以为边坡失稳灾害的相关预测提供坚实的技术基础。
The safety factor is one of the important indicators used to evaluate slope stability.The complex slope system leads to the uncertainty of safety factor prediction.Therefore,in order to obtain a more reliable safety factor and solve the shortcoming that the Pelican Algorithm(POA)would easily fall into a local optimum as the number of iterations increases,the regression model combined with multi-strategy Pelican Algorithm(IPOA)and support vectors machine(SVR)was proposed to predict the slope safety factor.First,the original Pelican algorithm was improved by integrating multiple strategies.Then the improved Pelican algorithm was combined with the support vector machine to select six influencing factors as input layer indicators of the IPOA-SVR model and train the model to obtain IPOA-SVR slope stability prediction model.Finally,compared with KNN,RF and Adaboost models,the mean square errors(MSE)of each model on the training set and test set were calculated to verify the superiority of the IPOA-SVR model.The experimental results showed that compared with other models,the IPOA-SVR model had better optimization performance.The mean square error on the test set was 0.0309,and the correlation coefficient was 0.91.This illustrates the effectiveness of the strategy used in this article for the POA algorithm.And the IPOA-SVR model can provide a solid technical foundation for the prediction of slope instability disasters.
作者
张佳琳
王孝东
吴雅菡
水宽
张玉
程玥淞
杜青文
ZHANG Jialin;WANG Xiaodong;WU Yahan;SHUI Kuan;ZHANG Yu;CHENG Yuesong;DU Qingwen(Faculty of Land Resources Engineering,Kunming University of Science and Technology,Kunming 650093,China;Institute of Resource Utilization and Rare Earth Development,Guangdong Academy of Sciences,Guangzhou 510651,China;National Key Laboratory of Rare Metal Separation and Comprehensive Utilization,Guangzhou 510651,China;Key Points of Mineral Resources Development and Comprehensive Utilization of Guangdong Province,Guangdong Laboratory,Guangzhou 510651,China;Faculty of Public Safety and Accident Management,Kunming University of Science and Technology,Kunming 650093,China)
出处
《有色金属(矿山部分)》
2025年第1期115-123,共9页
NONFERROUS METALS(Mining Section)
基金
云南省重点研发计划资助项目(202003AC100002)
昆明理工大学引进人才科研启动基金资助项目(KKSY201721032)。
关键词
安全系数
鹈鹕算法
支持向量机
边坡稳定性
均方误差
safety factor
The Pelican Algorithm
Support Vector Regression
slope stability prediction
Mean Square Error
