摘要
以某导管架平台为模型,对分别存在腐蚀和桩基冲刷影响的结构进行模态分析,表明在腐蚀和桩基冲刷的影响下海洋平台的模态参数有明显的时变效应。基于频率变化和神经网络算法的损伤识别方法有较好的记忆能力和泛化能力,可准确识别结构损伤程度。忽略腐蚀和桩基冲刷的影响,基于神经网络算法的损伤识别结果产生较大误差,难以对损伤程度实现有效识别。考虑腐蚀和桩基冲刷影响,提出基于神经网络算法的海洋平台损伤识别方法的改进方案。数值研究表明,改进方法识别误差很小,可对损伤程度实现有效识别。
Modal analysis of an offshore platform model under corrosion and scour condition respectively indicates that modal parameters are time varying. The damage identification method based on frequency change and neural network algorithm can precisely calculate damage severity for processing favorable recollection and externalization capacity. Neglecting the impact of corrosion and scour condition, valid damage severity calculation is hard to achieve as the results generated by neural network contain considerable errors. To consider the impact of corrosion and scour condition, an improvement for the neural network algorithm based damage identification method of offshore platform is proposed. Numerical analysis shows that the improved method is accurate enough to identify damage severity.
出处
《海洋工程》
CSCD
北大核心
2017年第3期21-28,共8页
The Ocean Engineering
基金
教育部
财政部"船舶数字化智能设计系统"资助项目(201335)
关键词
频率变化
神经网络
腐蚀
冲刷
损伤识别
海洋平台
frequency change
neural network
corrosion
scour
damage identification
offshore platform
