深海混输立管作业期间持续遭受外部风浪流耦合荷载及内部矿液两相流体磨蚀作用，长期作用下结构损伤逐渐积累。立管结构呈高长细比、柔性状态，传统结构损伤识别过程中存在模态参数识别困难、单测点响应损伤敏感度低等诸多问题。针对上述问题，本文提出了基于数据融合及一维残差卷积自编码器（One Dimension Residual Convolution Autoencoder，1D-RCAE）的深海混输立管结构损伤识别方法，以结构损伤敏感的应变动态响应为输入，使用主成分分析（Principal Component Analysis, PCA）进行多测点应变响应特征融合，进一步利用1D-RCAE自动提取损伤敏感特征，以结构损伤前后敏感特征间的马氏距离构建结构损伤判定指标，实现混输立管结构健康状态监测。通过500米深海混输立管结构数值模拟和立管物理缩尺模型实验对本文提出的方法进行验证，结果表明：有限测量信息下能够有效实现深海混输立管结构损伤识别，其中数值模拟验证损伤识别准确率高于99%，物理模型验证损伤识别准确率高于98%。同时探究了噪声污染、海洋环境因素变化对本文提出方法损伤识别性能的影响规律。
Abstract
Deep-sea mining riser continuously suffer from internal Solid-liquid two-phase fluid abrasion and external wind-wave-current coupling load during operation, and structure damage gradually accumulates. Because of high slenderness ratio and flexibility, it is difficult to identify modal parameters, and the damage sensitivity of single measuring point response is low when applying traditional damage detection method to deep-sea mining riser. To solve the problems, a data fusion and one dimension residual convolutional auto-encoder (1D-RCAE) based method is proposed for deep-sea mining riser damage identification. Firstly, PCA is used to fuse bending strain responses from multiple measuring points into one variable. Then, the 1D-RCAE is used to extract the damage sensitive feature (DSF) from the fused variable. Lastly, the Mahalanobis distance between the extracted DSF under the currently testing and the baseline conditions is selected as the damage index. The damage detection effectiveness is verified on a 500m numerical model and a laboratory model of deep-sea mining riser, and the result shows that the accuracy of damage identification is higher than 98%. At the same time, the effects of noise pollution and changing marine environment is explored.
关键词
深海混输立管 /
结构损伤识别 /
一维残差卷积自编码器 /
主成分分析 /
数据融合
{{custom_keyword}} /
Key words
Deep-sea mining riser /
structural damage identification /
one dimensional residual convolutional auto-encoder (1D-RCAE) /
principal component analysis (PCA) /
data fusion
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1]. Miller K A, Thompson K F, Johnston P, et al. An overview of seabed mining including the current state of development, Environmental Impacts, and Knowledge Gaps [J]. Frontiers in Marine Science, 2018, 4:418.
[2]. Wang S Q, Xu M Q. Modal strain energy-based structural damage identification: a review and comparative study [J]. Structural Engineering International, 2018, 29(2):1-15.
[3]. 张兆德，王德禹. 基于模态参数的海洋平台损伤检测[J]. 振动与冲击，2004，3(02)：7-12+135.
Zhang Z D, Wang D Y. Study on crack detection using modal parameters of a damaged jacket platform [J]. Journal of Vibration and Shock, 2004, 3(02):7-12+135.
[4]. Min C, Kim H, Yeu T, et al. Sensitivity-based damage detection in deep water risers using modal parameters: numerical study [J]. Smart Structures & Systems, 2015, 15(2):315-334.
[5]. Graves J R, Dareing D W. Direct method for determining natural frequencies of marine risers in deep water [J]. Journal of Energy Resources Technology, 2004, 126(1):47-53.
[6]. Je H M, Park S Y. A study on damage detection of production riser [J]. Journal of Navigation and Port Research, 2015, 39(3):179-184.
[7]. Riveros C, Utsunomiya T, Maeda K, et al. Vibration-based damage detection in flexible risers using time series analysis [J]. Doboku Gakkai Ronbunshuu A, 2007, 63(3):423-433.
[8]. Liu H, Yang H Z, Liu F S, et al. Damage localization of marine risers using time series of vibration signals [J]. Journal of Ocean University of China, 2014, 13(5):777-781.
[9]. 胡晨. 基于深度学习理论的桥梁结构损伤识别研究[D]. 南昌市: 华东交通大学，2020.
Hu C. Research on bridge infrastructures damage identification based on deep learning. Nanchang, East China Jiaotong University, 2020.
[10]. Masci J, Meier U, Cireşan D, et al. Stacked convolutional auto-encoders for hierarchical feature extraction [C]// Honkela, T. In: International Conference on Artificial Neural Networks, Berlin, Springer Berlin Heidelberg, 2011, 52-59.
[11]. Yang Q, Shen D J, Du W C, et al. A Deep learning-based framework for damage detection with time series [J]. IEEE Access, 2021, 9: 66570-66586.
[12]. Li A, Vafa S, Paul Z. Evaluation of ASR in concrete using acoustic emission and deep learning [J]. Nuclear Engineering and Design, 2021, 380:111328.
[13]. Yu J B, Zhou X K. One-dimensional residual convolutional autoencoder based feature learning for gearbox fault diagnosis [J]. IEEE Transactions on Industrial Informatics, 2020, 16(10):6347-6358.
[14]. 王硕，王培良. 基于一维卷积自编码器-高斯混合模型的间歇过程故障检测[J]. 信息与控制，2019，48(03): 285-292.
Wang S, Wang P L. Fault detection method for a batch process based on a one-dimensional convolution autoencoder and gaussian mixture model [J]. Information and Control, 2019, 48(03):285-292.
[15]. Wang S Q, Wang H Y, Xu M Q, et al. Identifying the presence of structural damage: A statistical hypothesis testing approach combined with residual strain energy [J]. Mechanical Systems and Signal Processing, 2020, 140:106655.
[16]. Xu M Q, Wang S Q, Li H J. A residual strain energy based damage localisation method for offshore platforms under environmental variations [J]. Ships and Offshore Structures, 2019, 14(7): 747-754.
[17]. Deraemaeker A, Reynders E, Roeck G D, et al. Vibration-based structural health monitoring using output-only measurements under changing environment [J]. Mechanical Systems and Signal Processing, 2008, 22(1):34-56.
[18]. 董绍江，吴文亮，贺坤，等. 基于性能衰退评估的轴承寿命状态识别方法研究[J]. 振动与冲击，2021，40(05): 186-192+210.
Dong S J, Wu W L, He K, et al. Bearing life state recognition method based on performance degradation evaluation [J]. Journal of Vibration and Shock, 2021, 40(05):186-192+210.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(1891 KB)
