基于神经网络的矩形钢管混凝土破坏模态研究

Study on the failure mode of concrete-filled rectangular steel tubular columns based on artificial neural network

为了研究矩形钢管混凝土柱轴压破坏模态,总结了大量文献的试验数据,并采用Abaqus进行有限元模拟,有限元模拟结果作为研究破坏模态的补充数据.分析结果表明矩形钢管混凝土柱轴压的破坏模态主要分为受压和受弯2种,长细比是影响破坏模态类型的最主要的因素.通过神经网络的预测方法研究受压及受弯2种破坏模态的长细比临界值,同时对矩形钢管混凝土柱的轴压受压受弯破坏模态长细比临界值进行了参数化分析.结果表明:在本文研究参数范围内,受压与受弯2种破坏模态的长细比临界值为32左右,长宽比(长/宽)越小,含钢率越大,约束效应系数越大,临界值越大;混凝土强度越大,临界值越大;钢管厚度越大,约束效应系数越大,临界值越小;钢材强度越大,约束效应系数越大,临界值越小.

In order to study on the failure mode of concrete-filled rectangular steel tubular columns （CFF） under axial compression, the article summarizes a lot of test data and uses Abaqus to simulate the failure mode, the results of simulation are used as supplemental data. The results show that the major failure mode of CFRT on axial compression can be divided into two kinds called compression and flexural, and slenderness ratio is the most important faetor affecting the type of failure mode. We also studied on the threshold of slenderness ratio based on forecasting method neural network. At the same time, parameter analysis of slenderness ratio threshold was conducted with the neural network. The results show that the slenderness ratio of two types of failure mode is about 32.The smaller Aspeet ratio （length/width） is, the greater steel ratio is, the greater confining factor is, the greater slenderness ratio is;the greater strength of concrete is, the greater slenderness ratio is; the greater steel thickness is, the greater confining factor is ,the smaller slenderness ratio is; the greater strength of steel is, the greater confining factor is, the smaller slenderness ratio is.