摘要
精确评估土体的抗剪强度参数对于确保工程安全至关重要,目前尚未有黏泥型泥化夹层抗剪强度指标的预测模型。为得出黏泥型泥化夹层抗剪强度预测最优模型,通过引入牛顿-拉夫逊算法优化神经网络,显著提高了模型的收敛速度和预测精度。同时,将模型与布谷鸟优化BP神经网络模型(CS-BP)、猎人猎物优化BP神经网络模型(HPO-BP)和BP神经网络模型进行了对比。结果表明,无论对于内摩擦角还是黏聚力,NRBO-BP模型都表现出最好的预测能力。在内摩擦角预测中,NRBO-BP模型在训练集和测试集的决定系数分别达到了0.9595、0.9301;在黏聚力预测中,NRBO-BP模型在训练集和测试集的决定系数分别达到了0.9684、0.9341。同时,在内摩擦角和黏聚力预测中NRBO-BP模型的精度在众多对比模型中均为最高。NRBO-BP模型有作为黏泥型泥化夹层抗剪强度指标预测的标准模型使用的潜力。
Accurate assessment of the shear strength parameters of the soil is crucial to ensure the safety of the project,and there is no prediction model for the shear strength index of slime-type mudded intercalations.In order to derive the optimal model for predicting the shear strength of slime-type mudded intercalations,the model was built by optimizing the neural network through the Newton-Raphson algorithm,and was compared with the Cuckoo optimization BP neural network model(CS-BP),the hunter prey optimization BP neural network model(HPO-BP),and the BP neural network model.The results show that the NRBO-BP model exhibits the best predictive ability in the prediction of both internal friction angle and cohesion.In the prediction of internal friction angle,the coefficient of determination of the NRBO-BP model reached 0.9595 and 0.9301 in the training and test sets,respectively,and in the prediction of cohesion,0.9684 and 0.9341 in the training and test sets,respectively.In both of the aforementioned predictions,the NRBO-BP model achieved the highest accuracy amonst all the comparatory models.It could be concluded that the NRBO-BP model has the potential of being used as a standard model for the prediction of shear strength index of slime-type mudded intercalations.
作者
鲁明智
张家明
邱培城
高宇
LU Mingzhi;ZHANG Jiaming;QIU Peicheng;GAO Yu(Faculty of Civil Engineering and Mechanics,Kunming University of Science and Technology,Kunming 650500,China)
出处
《材料导报》
EI
CAS
CSCD
北大核心
2024年第S02期277-281,共5页
Materials Reports
关键词
黏泥型泥化夹层
抗剪强度
牛顿-拉夫逊算法
BP神经网络
预测
slime-type mudded intercalation
shear strength
Newton-Raphson algorithm
BP neural network
prediction
