基于模态识别的波形钢腹板箱梁损伤位置研究

Study on Damage Location of Corrugated Steel Web Box Girder Based on Mode Recognition

摘要:

为解决波形钢腹板箱梁损伤识别和损伤定位问题，以上海塘大桥工程中波形钢腹板箱梁为研究对象，采用Ansys Workbench有限元建立波形钢腹板箱梁损伤结构计算模型，结合BP神经网络学习和重复训练功能，对波形钢腹板箱梁损伤前后的模态进行重复训练和计算，探索波形钢腹板箱梁损伤程度和位置与模态参数之间的变化规律。研究表明：波形钢腹板箱梁损伤后模态参数呈现下降趋势，不同程度和位置损伤后对模态参数影响存在一定的差异性，结构损伤程度与模态参数下降呈线性关系，训练后的BP神经网络对波形钢腹板箱梁损伤识别误差在3%范围内，验证了训练后的BP神经网络对波形钢腹板箱梁损伤识别的可靠性和准确性。

Abstract:

In order to solve the problem of damage identification and location of corrugated steel web box girder, taking the corrugated steel web box girder in Shanghaitang Bridge Project as the research object, the damage structure calculation model of corrugated steel web box girder is established by finite element software of Ansys Workbench. Combined with BP neural network learning and repeated training function, the modes of corrugated steel web box girder before and after damage ware retrained and calculated to explore the change rule between the damage degree and position of corrugated steel web box girder and the modal parameters. The study shows that the modal parameters of corrugated steel web box girder present a downward trend after damage, and the influence of different degree and location of damage on the modal parameters is different. The structural damage degree has a linear relationship with the decline of modal parameters. The trained BP neural network can identify the damage error of corrugated steel web box girder within 3%. The reliability and accuracy of the trained BP neural network for damage identification of corrugated steel web box girder are verified.