摘要
建立了基于多层前馈神经网络的钢筋混凝土梁柱节点受剪承载力计算模型。同时,结合MATLAB程序,完成了国内外94个构件的试验结果与我国GB50010-2010、ACI-ASCE 352等现有规范计算结果及BP-ANN计算结果的对比研究。研究表明:经神经网络模型所得计算结果与试验值吻合良好,离散程度较小,且较各国规范建议的设计方法的计算结果更接近试验值。因此,该神经网络模型可用于计算及预测钢筋混凝土梁柱节点的受剪承载力。
Based on multi-layer feedforward neural network （BP-ANN）, the calculation model of shear capacity of RC beam-column join was established. At the same time, combining with program MATLAB the contrast study of 94 components at home and abroad between experimental results and calculated results from current codes GBSO010 - 2010,ACI-ASCE 352 and from BP-ANN was carried out. Researches show that the reinforced concrete beam-col- umn joint shear strengths obtained by the simplified model based on BP-ANN are in good agreement with test re- sults, and they are closer to the experimental values than the results from design codes. The BP- ANN model can be used to the shear capacity prediction and design of reinforced concrete beam-column joint.
出处
《世界地震工程》
CSCD
北大核心
2015年第4期7-14,共8页
World Earthquake Engineering
基金
国家自然科学基金项目资助(51308065)
陕西省自然科学基金项目资助(2012JM7011)
长安大学创新团队资助项目(2013G3282015)
高等学校博士学科点专项科研基金资助课题(优先发展领域)(20130205130001)
