改进BP神经网络钢筋混凝土井壁极限承载力随机分析  

Random analysis of the ultimate bearing capacity of reinforced concrete shaft lining based on improved BP neural network

作　　者：王伟[1] 李永恒 李思琪 姚亚锋[1,2] WANG Wei;LI Yongheng;LI Siqi;YAO Yafeng(School of Architectual Engineering,Nantong Vocational university,Nantong 226001,China;Department of Civil Engineering,Anhui Jianzhu University,Hefei 230022,China)

机构地区：[1]南通职业大学建筑工程学院,江苏南通226001 [2]安徽建筑大学土木工程学院,安徽合肥230022

出　　处：《中国测试》2023年第6期36-41,共6页China Measurement & Test

基　　金：江苏省高等学校大学生创新创业训练计划项目(202111052016Y);南通市级科技计划项目(MS12021028)。

摘　　要：钢筋混凝土井壁模型试验分析发现最能影响极限承载力的是混凝土抗压强度,影响最弱的因素为配筋率。工程勘探发现:在深部井筒工程中,井壁极限承载力具有较大的不确定性。针对传统BP神经网络在解决随机性工程问题中的不足,采用随机因子和层间均方误差修正网络模型评价函数。在此基础上,以混凝土抗压强度、厚径比和配筋率为输入量,以井壁极限承载力为输出值,建立改进BP神经网络的井壁极限承载力随机预测模型。工程实例表明该模型在实际工况中能较好地预测深厚沉积层钢筋混凝土井壁的极限承载能力,且误差控制在4%以内,相比传统的计算公式方法精度有明显的提高。同时,改进的模型具有较小的权值和偏置,比传统网络的响应更趋于平滑,大大减少过度拟合的现象。The analysis of reinforced concrete shaft lining model test shows that the compressive strength of concrete is the most influential factor of ultimate bearing capacity and the ratio of reinforcement is the least influential factor.In deep shaft engineering,the ultimate bearing capacity of shaft lining has great uncertainty.Aiming at the shortcomings of traditional BP neural network in solving stochastic engineering problems,random factor and interlayer mean square error correction network model are used to evaluate the function.On this basis,taking concrete compressive strength,thickness-diameter ratio and reinforcement ratio as input and shaft lining ultimate bearing capacity as output,a random prediction model of shaft lining ultimate bearing capacity is established by improved BP neural network.The engineering example shows that the model can predict the ultimate bearing capacity of reinforced concrete shaft lining in deep sedimentary layer well under actual working conditions,and the error is controlled within 4%,which is obviously better than the traditional calculation formula method.In addition,the improved model has smaller weight and bias,and the response tends to be smoother than the traditional network,which greatly reduces the over-fitting phenomenon.

关 键 词：井壁承载力 钢筋混凝土 改进BP网络 随机预测 

分 类 号：TD26[矿业工程—矿井建设] TB9[一般工业技术—计量学]