基于BP神经网络的单管通信塔损伤识别研究  

Damage Identification of Single-tower Based on the BP Neural Network

作　　者：王清华 荆建中[1] 陈振[1] 唐明波 Wang Qinghua;Jing Jianzhong;Chen Zhen;Tang Mingbo(China Information Technology Design&Consulting Institute Co.Ltd.,Beijing 100048 China)

机构地区：[1]中讯邮电咨询设计院有限公司,北京100048

出　　处：《特种结构》2020年第5期112-116,共5页Special Structures

摘　　要：基于BP神经网络的铁塔损伤识别技术是一种基于铁塔动力参数来识别损伤的新技术,具有对铁塔结构损伤敏感等优点。本文利用有限元分析软件ANSYS首先计算完整铁塔结构的模态频率和位移,通过对不同位置铁塔结构刚度的折减模拟铁塔结构的损伤,分别计算不同位置刚度折减后铁塔结构的模态位移和频率,将模态位移和频率作为BP神经网络的输入参数,利用不同损伤工况下的动力参数训练并识别铁塔的损伤的程度和位置。研究结果表明,基于BP神经网络的铁塔损伤识别技术能够有效的评估出铁塔结构的损伤。Damage identification of tower based on the BP neural network was a new damage identification technique based on dynamic parameter of tower which has some advantages such as damage sensitivity etc.In this study firstly the modal displacement and frequency of non-destructive tower were calculated by the finite element analysis software ANSYS.Then the damage of tower by reducing the rigidity of the tower at different positions was simulated the different modal displacements and frequencies of destructive tower can be obtained.The modal displacements and frequencies in different cases were used as the input parameters of the BP neural network.Finally the dynamic parameters in different cases were used to train and identify the damage degree and location of the tower.The study demonstrates that the proposed damage identification technique of tower is very effective in identifying the damage of tower.

关 键 词：BP 神经网络 铁塔损伤识别 模态位移 频率 

分 类 号：TN929.5[电子电信—通信与信息系统] TP183[电子电信—信息与通信工程]