摘要
文中采用麻雀搜索算法优化BP神经网络,对深圳市某在建地铁车站深基坑周围地表沉降监测点进行变形预测。通过对基坑地表变形监测点DBC16-4的118期监测数据进行训练学习,并与粒子群算法优化BP神经网络、遗传算法优化BP神经网络和标准BP神经网络横向对比,验证了训练效果。结果表明,麻雀搜索算法对BP神经网络权重寻优速度较快,收敛精度更高,麻雀搜索算法优化BP神经网络模型预测平均相对误差仅为1.72%,拟合精度较其他算法更高,预测效果良好。
In this study,the sparrow search algorithm was applied to optimize the BP neural network for predicting deformations in monitoring points around a metro station's deep foundation pit in Shenzhen.The training and learning were conducted using 118 periods of monitoring data from monitoring point DBC16-4.The training performance was compared with particle swarm optimization-based BP neural network,genetic algorithm-based BP neural network,and standard BP neural network.The results indicate that the sparrow search algorithm optimizes the weights of the BP neural network with faster speed and higher convergence accuracy.The sparrow search algorithm optimized BP neural network model exhibits an average relative error of only 1.72%,which is lower than other algorithms,demonstrating superior fitting accuracy and excel-lent prediction performance.
作者
石强
程泷
杨展
赵嘉
Shi Qiang;Cheng Long;Yang Zhan;Zhao Jia(Power China Sinohydro Bureau 7 Co.Ltd.,Chengdu,Sichuan 610213;Faculty of Engineering,China University of Geosciences,Wuhan,Hubei 430074)
出处
《江西建材》
2024年第6期174-176,179,共4页
关键词
深基坑
地表沉降
变形预测
BP神经网络
麻雀搜索算法
Deep foundation pits
Ground settlement
Deformation prediction
BP neural network
Sparrow search algorithm
