基于MEA-BP神经网络的自锚式悬索桥施工阶段吊索索力预测分析  

Prediction and Analysis of Sling Force of Self-anchored Suspension Bridge in Construction Stage Based on MEA-BP Neural Network

作　　者：赵立财 ZHAO Licai(Department of Civil and Construction Engineering,National Taiwan University of Science and Technology,Taipei 10607,China;Third Engineering Co.,Ltd.,China Railway 19th Bureau Group Corporation Limited,Shenyang 110136,China)

机构地区：[1]台湾科技大学营建工程系,中国台湾台北10607 [2]中铁十九局集团有限公司第三工程有限公司,辽宁沈阳110136

出　　处：《铁道学报》2023年第3期55-61,共7页Journal of the China Railway Society

基　　金：辽宁省“兴辽英才计划”青年拔尖人才项目(XLYC2007146);中国铁建股份有限公司科技研究开发计划(2020-C20)。

摘　　要：为减小自锚式悬索桥在施工过程中吊索索力偏差对桥梁线形的影响程度,提高有限元模型的计算效率,提出一种基于思维进化(MEA)算法优化BP神经网络的吊索索力预测方法,以实现对桥梁各施工阶段的高精度逼近与吊索索力的快速反馈。在考虑施工过程中材料参数、荷载参数和环境温度等因素的不确定性基础上,结合有限元模型得到神经网络训练样本集。通过MEA算法实现BP神经网络权值与阈值的寻优,从而提高BP神经网络的预测精度。以某空间索面自锚式悬索桥为工程背景,建立该座桥梁的MEA-BP神经网络预测模型。结果表明,MEA-BP神经网络较传统BP神经网络具有更强的泛化能力与预测精度,MEA-BP神经网络的预测值与现场实测值的误差在10%以内,MEA-BP神经网络模型在索力预测方面具有较好的适用性。In order to reduce the influence of sling force deviation on the bridge alignment during the construction of self-anchored suspension bridge and improve the calculation efficiency of finite element model,a sling force prediction method based on mind evaluation algorithm(MEA)optimized BP neural network was proposed to achieve high-precision approximation and fast feedback of sling force in each construction stage of the bridge.Considering the uncertainty of material parameters,load parameters and ambient temperature during construction,the training sample set of neural network was obtained based on the finite element model.The weight and threshold of BP neural network were optimized by MEA for improving the prediction accuracy of the BP neural network.Based on the case study of a self-anchored suspension bridge,the MEA-BP neural network prediction model of the bridge was established.The results show that the MEA-BP neural network has stronger generalization ability and prediction accuracy than the traditional BP neural network.The error between the predicted value of the MEA-BP neural network and the field measured value is less than 10[WTB4]%.The MEA-BP neural network model shows good applicability in cable force prediction.

关 键 词：自锚式悬索桥 吊索索力 BP神经网络 思维进化算法 

分 类 号：U448.25[建筑科学—桥梁与隧道工程]