基于神经网络的钢绞线加固RC柱抗震性能评估

Evaluation of Seismic Behavior of RC Columns Strengthened Steel Wire Based on Neural Network

为评估钢绞线加固RC柱的抗震性能,基于27根钢绞线加固柱在低周反复荷载作用下的试验结果,在利用灰色关联理论进行模型输入变量选择的基础上,建立了基于径向神经网络模型的延性分析模型并进行了预测,研究了轴压比、钢绞线间距、预应力水平因素对加固柱延性的影响规律。结果表明:该方法能够反映延性与影响因素间的非线性变化规律;当剪跨比较小时,预应力水平以0.6为分界点,预应力水平小于0.6时提高对延性是有利的,预应力水平大于0.6时再提高将对延性不利;评估结果可为工程实际和抗震与优化设计提供参考。

For evaluating the seismic peffomlance of RC columns strengthened steel wire, based on the test results of 27 circular RC columns strengthened with prestressed high strength steel wire under low reversed cyclic loading, and on the basis of selection of model input variable using the them7 of grey conrrelation, the ductility of the radial neural network analysis model is established. Through prediction and research, the influence law of reinforement column ductility is known about the axial compression ratio, steel wire spacing, prestressinglevel. The results show that the method can reflect the nonlinear variation between ductility and impact factors. When the shear span ratio is small, prestressing level in 0. 6 as the cut-off point. When prestressinglevel is less than 0.6, the increase of the prestressing level is good for ductility. On the other hand, when prestressing level is gloater than 0. 6, the increase of the prestressing level will be adverse to the ductility. Evaluation results provide reference for engineering and seismic and optimization design.