CAJ | 학술논문

基于BP神经网络，以7个结构参数和2个荷载参数作为输入，损伤值作为输出，建立了爆炸作用下钢筋混凝土(RC)柱损伤的快速评估模型。在模型中，采用以精细化有限元模拟得到的1,032组 RC 柱损伤的数据训练网络，采用遗传算法和粒子群算法优化网络的初始权值和阈值，并以216个新损伤样本对模型的预测能力进行测试。研究表明，所建立的评估模型能够较准确地用于爆炸作用下 RC 柱损伤的快速评估；所得到的损伤分区图可以有效并直观地用于RC柱的爆炸风险评估。
기우BP신경망락，이7개결구삼수화2개하재삼수작위수입，손상치작위수출，건립료폭작작용하강근혼응토(RC)주손상적쾌속평고모형。재모형중，채용이정세화유한원모의득도적1,032조 RC 주손상적수거훈련망락，채용유전산법화입자군산법우화망락적초시권치화역치，병이216개신손상양본대모형적예측능력진행측시。연구표명，소건립적평고모형능구교준학지용우폭작작용하 RC 주손상적쾌속평고；소득도적손상분구도가이유효병직관지용우RC주적폭작풍험평고。
A fast assessment model for damage of RC columns under blast loading is proposed based on BP neural networks,by defining 7 structural parameters and 2 loading parameters as input,damage values as output. In the model,the network is trained using 1,032 groups of damage values of RC columns obtained from refined finite ele-ment simulation,and the initial weights and bias are optimized by genetic algorithm and particle swarm optimization. The predictive ability of the model is verified by 216 new samples of damage. The research indicates that the devel-oped model can be used for the fast assessment of damage of RC columns under blast loading with reasonable accu-racy,and the generated partition figure of damage can be effectively and intuitively used for the blasting risk assess-ment of RC columns.