首页 | 本学科首页   官方微博 | 高级检索  
     

基于神经网络优化算法的海洋动态缓波型立管设计
引用本文:黄 松,黄宇杰,何杨烨. 基于神经网络优化算法的海洋动态缓波型立管设计[J]. 海洋工程, 2024, 0(3): 83-94
作者姓名:黄 松  黄宇杰  何杨烨
作者单位:1. 中国石油大学(北京) 安全与海洋工程学院,北京 102249;2. 清华大学 能源与动力工程系,北京 100084
基金项目:中国石油大学(北京)基金资助项目(2462023QNXZ007);北京市科学技术协会青年人才托举工程(BYESS2023462)
摘    要:柔性立管连接海上平台和水下生产系统,缓波型构型可以减轻其受到的顶部张力和疲劳损伤,触地点处系链对于固定水下立管起着重要作用,然而加装系链的缓波型立管系统设计更为复杂。通过缓波型基本理论计算出合理的立管初始状态,之后使用立管分析软件OrcaFlex,在充分考虑环境因素、船体结构、立管材料和线型以及重力块、浮力块、系链等配置参数的情况下,建立缓波型立管系统有限元模型,通过静动态分析验证其满足设定的关于立管构型、张力、弯曲半径,系链张力和FPSO偏移量5个约束条件。结合神经网络优化算法和遗传算法制定出针对文中立管系统的优化算法并通过MATLAB编程将浮力块数量优化至最少,之后基于L-M算法构建神经网络模型,通过迭代训练提升精度,得到最终参数优化结果。通过对比优化前后静动态分析结果可知:优化后浮力块数量大幅减少,立管最大有效张力大幅减小,而悬挂段最低点深度有一定程度的增大,整体构型更趋向于合理的缓波型构型。

关 键 词:缓波型立管;浮力块;动态响应;神经网络;优化设计
收稿时间:2023-09-12
修稿时间:2024-01-29

Design of marine dynamic lazy wave riser based on neural network optimization algorithm
HUANG Song,HUANG Yujie,HE Yangye. Design of marine dynamic lazy wave riser based on neural network optimization algorithm[J]. The Ocean Engineering, 2024, 0(3): 83-94
Authors:HUANG Song  HUANG Yujie  HE Yangye
Affiliation:1. College of Safety and Ocean Engineering, China University of Petroleum, Beijing 102249, China; 2. Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China
Abstract:Flexible risers connect offshore platforms with subsea production systems. Lazy wave configurations can alleviate top tension and fatigue damage. The tether in the riser touch down area plays an important role in fixed subsea risers. However, designing lazy wave riser systems with added anchor chains is more complex. Through basic theory calculations of lazy-wave type principles, a reasonable initial riser state is determined. Subsequently, utilizing the OrcaFlex riser analysis software and considering environmental factors, vessel structure, riser materials, line types, gravity blocks, buoyancy blocks, anchor chains, and other configuration parameters, a finite element model of the lazy wave riser system is established. Static and dynamic analyses validate compliance with five predetermined constraints regarding riser configuration, tension, bending radius, anchor chain tension, and FPSO offset. Combining neural network optimization algorithms and genetic algorithms, an optimization algorithm tailored to the riser system is devised. MATLAB programming minimizes the number of buoyancy blocks. Then, based on the L-M algorithm, a neural network model is constructed, enhancing precision through iterative training to obtain the final parameter optimization results. Comparison of static and dynamic analysis results before and after optimization reveals a significant reduction in the number of buoyancy blocks and a substantial decrease in maximum effective tension in the riser. However, the depth of the lowest point in the suspended section increases to some extent, indicating a more reasonable lazy wave configuration overall.
Keywords:lazy wave riser; buoyancy blocks; dynamic response; neural networks; optimization design
点击此处可从《海洋工程》浏览原始摘要信息
点击此处可从《海洋工程》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号