独立桩海洋平台基础冲刷深度智能识别方法  

Intelligent Recognition Method for Scour Depth of Independent Pile Offshore Platform Foundation

作　　者：逄志浩 刘康[1] 王书冰 袁征 张权 朱渊[1] PANG Zhihao;LIU Kang;WANG Shubing;YUAN Zheng;ZHANG Quan;ZHU Yuan(Centre for Offshore Engineering and Safety Technology,China University of Petroleum(East China),Qingdao 266580,Shandong,China;Hainan Branch,CNOOC Enertech-QHSE Services Co.,Ltd.,Haikou 570105,Hianan,China)

机构地区：[1]中国石油大学(华东)海洋油气装备与安全技术研究中心,山东青岛266580 [2]中海油安全技术服务有限公司海南分公司,海南海口570105

出　　处：《中国海洋平台》2025年第2期37-44,86,共9页China offshore Platform

摘　　要：为应对独立桩海洋平台桩基在复杂海洋环境冲刷作用下产生的入土深度减小、承载力下降,严重影响平台稳定性的问题,提出一种基于变分模态分解(Variational Mode Decomposition,VMD)和深度学习算法的智能识别方法。构建独立桩海洋平台数字仿真模型,运用动力时程分析法模拟不同冲刷深度下平台的动力响应,采用VMD处理动力响应信号,提取关键特征参数,并以特征参数为输入,以冲刷深度为样本输出,结合卷积神经网络(Convolutional Neural Network,CNN)和双向长短时记忆(Bidirectional Long Short-Term Memory,BiLSTM)网络构建冲刷识别模型,进行冲刷深度工况的智能识别。使用试验测量数据对该冲刷智能识别方法的准确性进行验证。结果显示,该模型在仿真条件下的识别准确率达97.22%,在室内试验中的识别准确率达99.17%。In order to address the issue of reduced penetration depth and diminished bearing capacity of independent pile offshore platform foundation,due to scour in complex marine environments that critically impacts platform stability,an intelligent recognition method leveraging Variational Mode Decomposition(VMD)and deep learning techniques is introduced.A digital simulation model of an independent pile offshore platform is established,and the dynamic response of the platform under various scour depths is simulated via the dynamic time-history analysis method.The dynamic response signals are processed through VMD to extract the critical feature parameters.The extracted feature parameters serve as inputs,while the scour depth functions as the output to construct a scour recognition model.This model integrates Convolutional Neural Network(CNN)and Bidirectional Long Short-Term Memory(BiLSTM)network to intelligently recognize scour depth conditions.The accuracy of this scour recognition method is validated against test data.The results indicate that the model achieves a recognition accuracy of 97.22%under the simulation conditions and 99.17%in the laboratory tests.

关 键 词：海洋平台 冲刷深度 动力响应 智能识别 变分模态分解 卷积神经网络 双向长短时记忆网络 

分 类 号：TE95[石油与天然气工程—石油机械设备]