基于改进IAVOA-BP算法的GFRP布加固角钢极限承载力预测模型研究  

Prediction algorithm of ultimate bearing capacity of angle steel reinforced by GFRP sheet based on improved IAVOA-BP algorithm

作　　者：王彦海[1,2] 李书炀 邓德慧 李梦源 尹恒伟 Wang Yanhai;Li Shuyang;Deng Dehui;Li Mengyuan;Yin Hengwei(College of Electrical&New Energy,China Three Gorges University,Yichang 443000,China;Hubei Provincial Engineering Technology Research Center for Power Transmission Line,Yichang 443000,China;State Grid Hubei Jingmen Electric Power Supply Company,Jingmen 448000,China)

机构地区：[1]三峡大学电气与新能源学院,宜昌443000 [2]湖北省输电线路工程技术研究中心,宜昌443000 [3]国网湖北省电力有限公司荆门供电公司,荆门448000

出　　处：《国外电子测量技术》2023年第11期65-73,共9页Foreign Electronic Measurement Technology

基　　金：国家自然科学基金(52079070)项目资助。

摘　　要：针对现役输电铁塔在边坡变形等自然灾害作用下可能存在承载力不足的问题,提出采用玻璃纤维布(GFRP)对铁塔角钢进行加固,而现有复合材料加固钢结构承载力理论分析预测精度低、普适性不高。因此本文提出一种改进非洲秃鹫智能算法(AVOA)优化BP神经网络的GFRP布加固角钢极限承载力预测模型,首先引入Sobol序列、指数变换策略、多点Levy飞行策略及柯西变异扰动4种方法对原始AVOA算法优化;之后将得到的IAVOA算法优化BP算法的权值、阈值,得到IAVOA-BP预测模型;最后将角钢长细比、GFRP布层数、铺设角度、铺设长度作为输入量,加固后角钢极限承载力作为预测值并进行对比。结果表明,IAVOA-BP与AVOA-BP预测模型相比,平均相对误差下降47.61%、绝对平均误差下降47.04%、均方根误差值下降47.83%,改进后的IAVOA-BP预测模型能够较为准确的预测GFRP布加固角钢后的承载力大小。In response to the potential issue of insufficient bearing capacity of existing transmission towers under the influence of natural disasters such as slope deformation,it has been proposed to reinforce the tower angle steel with GFRP.However,the most theoretical analysis results of the bearing capacity of steel structures reinforced with composite materials are low prediction accuracy and low universality.Therefore,this paper proposes an improved african vulture optimization algorithm(AVOA)optimized BP neural network model for predicting the ultimate bearing capacity of GFRP-reinforced angle steel.Firstly,four methods,namely Sobol sequence,exponential transformation strategy,cauchy variation and multi-point Levy flight strategy,are introduced to optimize the original AVOA algorithm.Furthermore,the optimized weights and thresholds of the BP algorithm are obtained using the IAVOA,and the IAVOA-BP prediction model is obtained.Finally,by taking the aspect ratio of the angle steel,the number of GFRP cloth layers,the laying angle,and the laying length as input variables and the ultimate bearing capacity of the angle steel after reinforcement as the predicted value,a comparison was made.The results showed that,compared with the AVOABP prediction model,the IAVOA-BP prediction model with improvements had a decrease of 47.61%in average relative error,a decrease of 47.04%in absolute mean error,and a decrease of 47.83%in root mean square error.This proves that the improved IAVOA-BP prediction model can accurately predict the bearing capacity of GFRP-strengthened angle steel.

关 键 词：改进非洲秃鹫算法 复合材料 极限承载力 BP神经网络 Sobol序列 柯西变异 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TU391[自动化与计算机技术—控制科学与工程]