针对传统数值仿真计算方法耗时长、占用计算机资源多等缺点，提出基于LightGBM算法的拖曳系统动力响应进行评估的回归预测模型，以已有的OrcaFlex数值模拟得到的数据为样本，以拖曳系统上的海洋环境条件、拖船航速和下放缆长为特征，以动力响应为目标，引入LightGBM算法，对拖曳缆顶端张力最大值等动力响应进行预测分析。与传统数值模拟方法相比，LightGBM算法在保证结果准确性的同时大幅度提高了计算效率。通过与随机森林（RF）、极限梯度提升（XGBoost）算法相比，其准确度和计算效率的表现更好。最后提出了贝叶斯参数优化的LightGBM算法，准确度进一步提高，为提前采取措施保障拖曳系统的作业安全提供了一条高效的技术途径，同时为建立拖曳系统数字孪生体提供了有力的技术支撑。
In view of the shortcomings of traditional numerical simulation calculation methods, such as time-consuming and occupying more computer resources, a regression prediction model for evaluating the dynamic response of the towing system based on LightGBM algorithm is proposed. The data obtained from the existing OrcaFlex numerical simulation are taken as samples, and the marine environmental conditions on the towing system, the speed of the tugboat and the length of the cable are taken as characteristics. Aiming at the dynamic response, the LightGBM algorithm is introduced to predict and analyze the dynamic response such as the maximum tension at the top of the towing cable. Compared with traditional numerical simulation methods, LightGBM algorithm can ensure the accuracy of results and greatly improve the computational efficiency. Meanwhile, compared with Random Forest (RF) and limit gradient Boosting (XGBoost) algorithm, its accuracy and computational efficiency are better. Finally, LightGBM algorithm with Bayesian parameter optimization is proposed, which further improves the accuracy, provides an efficient technical approach for taking measures to ensure the safety of the towing system in advance, and provides a strong technical support for the establishment of the digital twin of the towing system.
2024,46(3): 34-40 收稿日期：2023-01-18
DOI：10.3404/j.issn.1672-7649.2024.03.006
分类号：U661.4
基金项目：国家重点研发计划资助项目（2018YFC0310502）
作者简介：董磊磊(1985-),男,博士,副教授,研究方向为非粘合柔性软管系统设计及分析
参考文献：
[1] 张大朋, 白勇, 朱克强, 等. 不同模式下拖缆对水下拖体运动姿态的影响研究[J]. 船舶力学, 2018, 22(8): 967-976.
ZHANG D M, BAI Y, ZHU K Q, et al. Study on the influence of towing cable on the motion attitude of underwater towing body under different modes[J]. Ship Mechanics, 2018, 22(8): 967-976.
[2] 刘铭, 李家旺, 朱克强. 基于集中质量法的水下拖曳缆索动力响应分析[J]. 水道港口, 2017, 38(4): 405-411.
LIU M, LI J W, ZHU K Q. Dynamic response analysis of underwater towing cable based on concentrated mass method[J]. Journal of Waterway and Harbour, 2017, 38(4): 405-411.
[3] LIU, CHUANG, LI, et al, Dynamic analysis of semi-submersible production platform under the failure of mooring lines[J]. China Ocean Engineering, 2021, 35(1): 84-95.
[4] YUAN Y, ZHENG M, XUE H, et al. Nonlinear riser-seabed interaction response among touchdown zone of a steel catenary riser in consideration of vortex-induced vibration[J]. Ocean Engineering, 2021, 227: 108891.
[5] 王琰, 胥美美, 童俞嘉, 等. 基于机器学习的环境监测数据对循环系统疾病死亡影响及预测预警模型构建[J]. 数据分析与知识发现, 2022, 6(10): 79-92.
WANG Y, XU M M, TONG Y J, et al. The impact of environmental monitoring data on death of circulatory system diseases and the construction of prediction and early warning model based on machine learning[J]. Data Analysis and Knowledge Discovery, 2022, 6(10): 79-92.
[6] WANG Z, QIAO D, YAN J, et al. A new approach to predict dynamic mooring tension using LSTM neural network based on responses of floating structure[J]. Ocean Engineering, 2022, 249: 110905.
[7] 张书旭, 李峰, 谢建平, 等. 基于融合神经网络模型的药物分子性质预测[J]. 计算机科学, 2021, 48(9): 251-256.
ZHANG S X, LI F, XIE J P, et al. Prediction of drug molecular properties based on fusion neural network model[J]. Computer Science, 2021, 48(9): 251-256.
[8] 吕国华, 胡学先, 杨明, 等. 基于联邦随机森林的船舶AIS轨迹分类算法[J]. 信息网络安全, 2022, 22(4): 67-76.
LV G H, HU X X, YANG M, et al. AIS trajectory classification algorithm for ships based on Federal Random Forest[J]. Netinfo Security, 2022, 22(4): 67-76.
[9] 罗恒, 刘杭. 基于LightGBM-LSTM组合模型的商业建筑能耗预测[J]. 计算机应用与软件, 2022, 39(11): 36-42+65.
LUO H, LIU H. Energy consumption forecast of commercial buildings based on LightGBM-LSTM combination model[J]. Computer Applications and Software, 2022, 39(11): 36-42+65.
[10] 贾子锌, 柳淑学, 李金宣, 等. 基于BP神经网络对横浪作用下系泊油船的运动量预测分析[J]. 水道港口, 2022, 43(4): 430-436.
JIA Z X, LIU S X. LI J X, et al. Prediction and analysis of motion of mooring tanker under transverse waves based on BP neural network[J]. Journal of Waterway and Harbor, 2022, 43(4): 430-436.
[11] 韩宇, 黄国良, 李鹏. 基于LSTM的软刚臂单点系泊力预测方法研究[J]. 天津科技, 2020, 47(5): 74-80.
HAN Y, HUANG G L, LI P, et al. Research on soft rigid arm single point mooring prediction method based on LSTM[J]. Tianjin Science & Technology, 2020, 47(5): 74-80.
[12] 李世峰, 邱占芝. 基于改进GA-BP网络的系泊缆力预测建模与仿真[J]. 系统仿真学报, 2017, 29(7): 1457-1463.
LI S F, QIU Z Z. Prediction modeling and simulation of mooring cable force based on improved GA-BP network[J]. Journal of System Simulation, 2017, 29(7): 1457-1463.
[13] 孙晔, 武文华, 樊哲良, 等. 基于RBF神经网络的FPSO系泊力预测方法和原型应用[C]//第十六届中国海洋(岸)工程学术讨论会论文集(上册), 2013: 306-312.
SUN Y, WU W H, FAN Z L, et al. Prediction method and prototype application of FPSO mooring based on RBF neural network [C]//Proceedings of the 16th China Ocean (Coastal) Engineering Symposium (Volume 1), 2013: 306-312.
[14] DIETTERICH T G. Machine learning research: four current directions[J]. AI Magazine, 1997, 18(4): 97-136.
[15] DNV. GL-RP-C205 Environmental conditions and envirormental loads[S].
[16] CARTER D J T. Prediction of wave height and period for a constant wind velocity using the JONSWAP results[J]. Ocean Enginerring, 1982, 9(1): 17-33.
[17] 张雯月. 风浪流联合作用对船舶随机横摇的影响研究[D]. 大连: 大连理工大学, 2019.
[18] 许亮. 基于贝叶斯优化算法的参数化结构设计[J]. 建筑技术开发, 2022, 49(17): 13-15.
XU L. Parametric structure design based on Bayesian optimization algorithm[J]. Building Technology Development, 2022, 49(17): 13-15.
