摘要
针对反向传播(back propagation,BP)神经网络在训练过程中存在的易过度拟合、收敛速度慢和易陷入局部最优等问题,引入天牛须搜索(beetle antennae search,BAS)算法优化传统BP神经网络中的权值和阈值,建立了BAS-BP神经网络模型。利用深圳市某深基坑开挖的周围道路地表沉降监测数据进行BAS-BP模型仿真测试。实验结果表明,BAS-BP模型在均方误差(mean square error,MSE)、平均绝对误差(mean absolute error,MAE)和平均绝对百分比误差(mean absolute percentage error,MAPE)精度指标上均优于BP神经网络模型,预测精度更高。
To solve the problems of back propagation(BP)neural networks in the training process,such as being easy to over-fit,slow convergence and being easy to fall into local optimum,we optimize the weights and thresholds of traditional BP neural networks by beetle antennae search(BAS)algorithm,and establish a neural network model called BAS-BP.The simulation test of BAS-BP model is conducted by the monitoring data of surface subsidence of a deep excavation in Shenzhen.The experimental results show that the mean square error(MSE),the mean absolute error(MAE)and the mean absolute percentage error(MAPE)of BAS-BP model are better than those of BP neural network model,which indicates that it has higher prediction accuracy.
作者
杨帆
黄超
YANG Fan;HUANG Chao(School of Surveying and Geoscience,Liaoning Technical University,Fuxin 123000,China)
出处
《测绘地理信息》
CSCD
2022年第5期47-50,共4页
Journal of Geomatics
基金
辽宁省教育厅重点实验室基础研究(LJZS001)
作者简介
第一作者:杨帆,教授,主要从事变形监测与预报方面的研究工作。E-mail:yangfan2008beijing@126.com;通讯作者:黄超,硕士生,主要从事变形监测、InSAR技术研究工作E-mail:1527678872@qq.com
