论文部分内容阅读
提出一种有效的框架结构螺栓连接半刚性节点损伤的分步识别方法.首先采用仅与损伤位置有关、而与损伤程度无关的模态参数损伤指标作为神经网络训练输入,以确定损伤节点位置,极大地简化了神经网络结构,提高了损伤定位效率;然后通过模态敏感性方法对前步识别出的可能损伤节点进行损伤程度识别,有效地克服了传统模态敏感性方法在同时进行损伤位置及程度识别时,损伤识别动力学反问题结果的准确性与可靠性受可用模态信息量影响较大的问题.通过对一个8层半刚性连接框架的数值仿真与一个实验室两层螺栓连接钢框架的实验研究,验证了所提方法的可行性与有效性.“,”For the majority of existing damage detection methods,it is generally concentrated on damage along members of the structure and neglects the damage at its connections.This paper develops an efficient method for identifying the semi-rigid connection damage for steel frame structures with bolted joints.In the proposed method,the artificial neural network is firstly introduced to indicate the damage locations,where the modal parameter based damage index depending only on the damage location is utilized as training data.This process greatly simplifies the structure of neural network and significantly improves the efficiency of damage localization.After that,the modal sensitivity-based method is employed to identify the damage extent for the identified potential damaged connections.The difficulty encountered for the convenient sensitivity-based damage detection method,i.e.,the significant influence of the lack of available modal measurement information on the accuracy and reliability of damage detection results,is avoided in the proposed method.The feasibility and efficiency of the proposed method is verified through both numerical simulations conducted for an eight-storey steel frame with semi-rigid connections as well as experimental verifications for a laboratory two-storey steel frame with bolted-connections.