目的 针对深水导管架平台在复杂海洋环境荷载作用下振动响应非平稳、噪声干扰强及阻尼比识别稳定性不足等问题,提出一种适用于现场监测数据的运行模态参数识别方法。方法 建立融合自适应随机减量技术、Hilbert包络分析和卷积神经网络的RDT-CNN模态识别方法,构建由RDT信号、Hilbert包络及对数包络组成的多通道输入特征,并利用CNN建立信号特征与固有频率、阻尼比之间的非线性映射关系。结果 数值算例表明,与传统RDT方法相比,RDT-CNN方法可将固有频率识别误差由20.0%降低至4.8%,阻尼比识别误差由44.3%降低至12.0%。工程实测数据验证表明,该方法在复杂环境振动条件下具有较好的识别稳定性。结论 自适应RDT-CNN方法能够有效提取非平稳振动响应中的衰减特征,降低传统RDT方法对理想指数衰减假设的依赖,提高深水导管架结构模态参数识别精度,可为海洋平台结构健康监测和运行模态分析提供参考。
Abstract
The work aims to address the challenges of non-stationary vibration responses, strong noise interference, and insufficient stability in damping ratio identification for deep-water jacket platforms under complex marine environmental loads, by proposing an operational modal parameter identification method suitable for field monitoring data. A novel RDT-CNN modal identification framework integrating Adaptive Random Decrement Technique (RDT), Hilbert envelope analysis, and Convolutional Neural Networks (CNN) was established. Firstly, an adaptive triggering strategy was employed to extract random decrement signatures from ambient vibration acceleration responses. Secondly, multi-channel input features comprising the RDT signals, Hilbert envelopes, and logarithmic envelopes were constructed. Subsequently, a CNN was utilized to establish the nonlinear mapping relationship between these signal features and the structural modal parameters (natural frequency and damping ratio). The numerical results demonstrated that, compared with traditional RDTs, the proposed RDT-CNN method significantly reduced identification errors: the error for natural frequency decreased from 20.0% to 4.8%, and the damping ratio error decreased from 44.3% to 12.0%. Field measurement validation further confirmed that the method exhibited superior identification stability under complex environmental vibration conditions. The adaptive RDT-CNN method effectively extracts decay characteristics from non-stationary responses, mitigates the dependence of traditional RDT on idealized exponential decay assumptions, and improves the precision of modal identification. This study provides valuable references for structural health monitoring and operational modal analysis of offshore platforms.
关键词
深水导管架 /
随机减量技术 /
卷积神经网络 /
运行模态识别 /
阻尼估计
Key words
deep-water jacket /
random decrement technique /
convolutional neural network /
operational modal identification /
damping estimation
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 伊廷华. 结构健康监测教程[M]. 北京: 高等教育出版社, 2021.
YI T H.Structural Health Monitoring[M]. Beijing: Higher Education Press, 2021.
[2] 杨和振, 李华军, 王国兴. 海洋平台结构模态参数识别的仿真研究[J]. 海洋工程, 2003, 21(4): 75-80.
YANG H Z, LI H J, WANG G X.A Simulation Study of Modal Parameters Identification for Offshore Platform Structures[J]. The Ocean Engineering, 2003, 21(4): 75-80.
[3] 吕中亮, 杨昌棋, 安培文, 等. 多点激励模态参数识别方法研究进展[J]. 振动与冲击, 2011, 30(1): 197-203.
LYU Z L, YANG C Q, AN P W, et al.Progress on Modal Parameter Identification with Multiple-Excitation[J]. Journal of Vibration and Shock, 2011, 30(1): 197-203.
[4] IBRAHIM S R.Random Decrement Technique for Modal Identification of Structures[J]. Journal of Spacecraft and Rockets, 1977, 14(11): 696-700.
[5] 陈太聪, 沈文杰. 模态辨识中随机减量技术的实用改进[J]. 振动测试与诊断, 2019, 39(6): 1153-1159.
CHEN T C, SHEN W J.Practical Improvement of the Random Decrement Technique in Modal Identification[J]. Journal of Vibration, Measurement & Diagnosis, 2019, 39(6): 1153-1159.
[6] 吴经纬. 卷积神经网络在土木工程结构损伤中的应用[J]. 中国建筑金属结构, 2025, 24(23): 147-149.
WU J W.Application of Convolutional Neural Network in Structural Damage of Civil Engineering[J]. China Construction Metal Structure, 2025, 24(23): 147-149.
[7] CUI M J, KHODAYAR M, CHEN C, et al.Deep Learning-Based Time-Varying Parameter Identification for System-Wide Load Modeling[J]. IEEE Transactions on Smart Grid, 2019, 10(6): 6102-6114.
[8] AZIMI M, PEKCAN G.Structural Health Monitoring Using Extremely Compressed Data through Deep Learning[J]. Computer-Aided Civil and Infrastructure Engineering, 2020, 35(6): 597-614.
[9] 万小朋, 王军强, 赵美英. 基于模态频率和神经网络的结构损伤检测[J]. 西北工业大学学报, 2003, 21(2): 156-159.
WAN X P, WANG J Q, ZHAO M Y.Structural Damage Monitoring Based on Vibration Modal Analysis and Neural Network Technique[J]. Journal of Northwestern Polytechnical University, 2003, 21(2): 156-159.
[10] 高志斌, 黎雯洋, 车鑫. 一种基于CNN-LSTM深度神经网络的智能发动机模型设计方法[J]. 计算机时代, 2026(2): 57-60.
GAO Z B, LI W Y, CHE X.A Method for Designing an Intelligent Engine Model Based on CNN-LSTM Deep Neural Networks[J]. Computer Era, 2026(2): 57-60.
[11] BRINCKER R, VENTURA C E.Introduction to Operational Modal Analysis[M]. New York: John Wiley & Sons, 2015.
[12] LECUN Y, BENGIO Y, HINTON G.Deep Learning[J]. Nature, 2015, 521(7553): 436-444.
[13] 周飞燕, 金林鹏, 董军. 卷积神经网络研究综述[J]. 计算机学报, 2017, 40(6): 1229-1251.
ZHOU F Y, JIN L P, DONG J.Review of Convolutional Neural Network[J]. Chinese Journal of Computers, 2017, 40(6): 1229-1251.
[14] 肖志权, 庞国强, 蔡展文, 等. CNN-LSTM在侧倾试验台法整车质心高度预测中的应用[J]. 重庆交通大学学报(自然科学版), 2026, 45(3): 120-127.
XIAO Z Q, PANG G Q, CAI Z W, et al.Application of CNN-LSTM in Predicting CG Height Ofthe Whole Vehicle by the Tilt Table Method[J]. Journal of Chongqing Jiaotong University (Natural Science), 2026, 45(3): 120-127.
[15] 杨观赐, 杨静, 李少波, 等. 基于Dopout与ADAM优化器的改进CNN算法[J]. 华中科技大学学报(自然科学版), 2018, 46(7): 122-127.
YANG G C, YANG J, LI S B, ET Al.Modified CNN Algorithm Based on Dropout and ADAM Optimizer[J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2018, 46(7): 122-127.
[16] 张雄, 王文强, 马亚伟, 等. 基于改进CNN-LSTM模型的滚动轴承多故障分类与诊断[J]. 中国工程机械学报, 2026, 24(1): 128-133.
ZHANG X, WANG W Q, MA Y W, et al.Multi Fault Classification and Diagnosis of Rolling Bearings Based on Improved CNN-LSTM Model[J]. Chinese Journal of Construction Machinery, 2026, 24(1): 128-133.
[17] 王文婷, 叶星宏, 夏利娟. 基于CNN-LSTM的船体板架损伤位置检测研究[J]. 舰船科学技术, 2025, 47(19): 87-93.
WANG W T, YE X H, XIA L J.Research on Hull Plate Frame Damage Detection Based on CNN-LSTM[J]. Ship Science and Technology, 2025, 47(19): 87-93.
[18] 李正夫, 王瑶祺, 周雨农, 等. 基于小波变换与注意力的CNN-LSTM海水浊度预测方法[J]. 环境化学, 2025, 44(12): 4819-4830.
LI Z F, WANG Y Q, ZHOU Y N, et al.Research on CNN-LSTM Seawater Turbidity Prediction Method Based on Wavelet Transform and Attention[J]. Environmental Chemistry, 2025, 44(12): 4819-4830.
[19] LÜ X B, NURMEMET I, XIAO S T, et al.Spatial-Temporal Simulation and Prediction of Root Zone Soil Moisture Based on Hydrus-1D and CNN-LSTM-Attention Models in Yutian Oasis, Southern Xinjiang, China[J]. Pedosphere, 2025, 35(5): 846-857.
[20] 汪靖尧, 冯陈, 张玉全, 等. 基于ICEEMDAN和CNN- LSTM-ATTENTION的抽水蓄能机组振动预测[J]. 水电能源科学, 2026, 44(3): 194-199.
WANG J Y, FENG C, ZHANG Y Q, et al.Vibration Prediction of Pumped Storage Units Based on ICEEMDAN and CNN-LSTM-ATTENTION[J]. Water Resources and Power, 2026, 44(3): 194-199.
[21] 李广耀, 孔新宇, 黄捷, 等. EMD与随机减量技术在飞机颤振模态辨识中的应用[J]. 成都航空职业技术学院学报, 2022, 38(4): 46-49.
LI G Y, KONG X Y, HUANG J, et al.Application of EMD and Random Decrement Technique in Aircraft Flutter Modality Identification[J]. Journal of Chengdu Aeronautic Polytechnic, 2022, 38(4): 46-49.
[22] 杨勇, 宁平华, 王晟, 等. 基于解析模式分解和随机减量技术的桥梁模态参数识别[J]. 铁道科学与工程学报, 2022, 19(5): 1355-1363.
YANG Y, NING P H, WANG S, et al.Analytical Modal Decomposition-Based Random Decrement Technique for Modal Parameters Identification of Bridge[J]. Journal of Railway Science and Engineering, 2022, 19(5): 1355-1363.
[23] 王明甲, 武永顶. 基于随机减量法振动系统模态分析仿真研究[J]. 机电技术, 2017(5): 56-58.
WANG M J, WU Y D.Simulation Study on Modal Analysis of Vibration System Base on Random Decrement Method[J]. Mechanical & Electrical Technology, 2017(5): 56-58.
[24] 徐辰奎, 刘国明, 杨亮. 基于随机减量和时域 ITD 法的重力坝模态参数识别[J]. 水利与建筑工程学报, 2015, 13(6): 178-182.
XU C K, LIU G M, YANG L.Modal Parameter Identification of Gravity Dams Based on ITD Method and Random Decrement Technique[J]. Journal of Water Resources and Architectural Engineering, 2015, 13(6): 178-182.
[25] 熊铃华, 张俊波, 蔡智华. 基于随机减量的船舶主机振动测试[J]. 动力学与控制学报, 2015, 13(1): 48-51.
XIONG L H, ZHANG J B, CAI Z H.Vibration Test of a Ship Host Based on Random Decrement[J]. Journal of Dynamics and Control, 2015, 13(1): 48-51.
PDF(4459 KB)
