基于模态曲率理论及神经网络的简支梁桥损伤识别

Damage identification for simply supported beam bridge based on modal curvature theory and neural network

针对多片简支梁桥损伤识别问题,提出了一种分步识别法。该方法综合运用模态曲率差及神经网络技术对结构损伤进行识别,能够有效地发挥两种方法的优点。最后以一座五片简支T梁桥为算例进行了数值模拟。模拟结果表明,该方法可以准确识别结构的损伤位置及损伤程度,是一种行之有效的简支梁桥损伤识别方法。

A two-step method was proposed to identify the damage of the simply supported beam bridge with multiple girders.The method identifies the structure damage using the modal curvature difference and the neural network technique to avoid the disadvantages of each single method.A numerical example was provided for a simply supported T beam bridge with five girders to verify the feasibility of the method,and the results show that this two-step method is effective to locate damage and evaluate the damage degree.