Machine Learning Model for Vibration Prediction of Deep-Sea Hydrate Mining Risers Based on CLR-ZOA-RF
-
摘要:
为解决深海天然气水合物开采管柱在复杂环境载荷下振动预测困难的问题，将斑马优化算法（ZOA）与随机森林算法（RF）相结合，同时借助混沌映射、莱维飞行和随机扰动变异策略（CLR）来提升斑马算法的性能，提出一种深海水合物开采管柱振动预测机器学习模型（CLR-ZOA-RF模型），实现多因素作用下的开采管柱振动响应实时预测；并采用相似原理，研发深海水合物开采管柱振动模拟实验装置，开展多因素下管柱振动实验，构建了实验数据集. 研究结果表明：所提出的CLR-ZOA-RF模型在处理多维非线性数据时展现出优异的泛化性能，其顺流向、横流向振动的决定系数（
R 2）分别由ZOA-RF模型的0.9359 和0.9264 提升至0.9719 和0.9667 ；相较于ZOA-RF模型，CLR-ZOA-RF模型对于顺流向（横流向）振动预测的平均绝对误差和均方根误差分别下降了26.83%、33.72% （45.15%、29.49%），可见其预测精度高、泛化性好，证实了混合智能算法的有效性.Abstract:To address the challenges of vibration prediction for deep-sea natural gas hydrate mining risers under complex environmental loads, a machine learning model named CLR-ZOA-RF was proposed. This model enhanced the performance of the zebra optimization algorithm (ZOA) by incorporating chaotic mapping, Levy flight, and random perturbation mutation (CLR) strategies, which are then integrated with the random forest (RF) algorithm to achieve real-time vibration response prediction of mining risers under multi-factor influences. Based on the similarity principle, a vibration simulation experimental device for the mining riser was developed to conduct multi-factor riser vibration experiments and construct an experimental dataset. The research results indicate that the proposed CLR-ZOA-RF model exhibits superior generalization performance in processing multi-dimensional non-linear data. Specifically, the determination coefficients (
R 2) for in-line and cross-flow vibrations are improved from 0.935 9 and 0.926 4 of the ZOA-RF model to 0.971 9 and 0.966 7, respectively. Compared with those of the ZOA-RF model, the mean absolute error and root mean square error of the CLR-ZOA-RF model decrease by 26.83% and 33.72% for in-line vibration and by 45.15% and 29.49% for cross-flow vibration. It can be seen that the model achieves high prediction accuracy and good generalization, confirming the effectiveness of the hybrid intelligent algorithm. -
图 1 基于RF算法的开采管柱振动预测模型
Figure 1. Vibration prediction model for mining riser based on RF algorithm
图 2 斑马优化算法优化RF模型参数的流程
Figure 2. Flowchart of optimizing RF model parameters using zebra optimization algorithm
图 3 深海水合物开采管柱振动模拟实验系统结构示意
Figure 3. Schematic diagram of vibration simulation experimental system for deep-sea hydrate mining risers
图 4 深海水合物开采管柱振动模拟实验系统
Figure 4. Vibration simulation experimental system for deep-sea hydrate mining risers
图 5 不同实验工况下RF模型对振动位移的预测效果
Figure 5. Vibration displacement predicted by RF model under experimental conditions
图 7 不同实验工况下ZOA-RF模型对振动位移的预测效果
Figure 7. Vibration displacement predicted by ZOA-RF model under experimental conditions
图 8 顺流向实验数据集消融实验适应度
Figure 8. Fitness of ablation study on in-line flow experimental dataset
图 9 横流向实验数据集消融实验适应度
Figure 9. Fitness of ablation study on cross-flow experimental dataset
图 10 顺流向实验数据集多种算法适应度
Figure 10. Fitness of multiple algorithms on in-line flow experimental dataset
图 11 横流向实验数据集多种算法适应度
Figure 11. Fitness of multiple algorithms on cross-flow experimental dataset
图 12 不同实验工况下CLR-ZOA-RF模型对振动位移的预测效果
Figure 12. Vibration displacement predicted by CLR-ZOA-RF model under experimental conditions
图 13 不同实验工况下多种算法优化RF模型对振动位移的预测效果
Figure 13. Vibration displacement prediction under experimental conditions using RF models optimized by multiple algorithms
表 1 开采管柱实际与实验参数对比
Table 1. Comparison of actual and experimental parameters of mining riser
项目 长度/m 外径/mm 内径/mm 材料类型 密度/（kg•m−3） 弹性模量/GPa 实际值 2000 440 400 13Cr-L80 7850 207.0 实验值 3 22 20 PVC 1600 2.1 
下载: 导出CSV
表 3 水合物颗粒实际与实验参数对比
Table 3. Comparison of actual and experimental parameters of hydrate particles
项目 直径/mm 材料类型 密度/（kg•m−3） 实际值 3～12 天然气水合物 880～ 1350 实验值 0.6（30目） 橡胶颗粒 900
下载: 导出CSV
表 4 顺流向振动实验数据集
Table 4. Experimental dataset of in-line vibration
序号 管柱
位置/mm目数/目 固液流量/（m3•h−1） 气体流量/
（L•min−1）外场流速/
（m•min−1）位移
均方根/mm1 1400 10 0.81 5.84 10 0.108145443 2 1400 20 0.85 3.46 30 1.135495117 3 1400 30 0.88 7.90 20 0.531436397 4 900 10 0.39 8.31 20 2.120023976 5 900 20 0.85 3.46 30 0.858739601 6 900 30 0.79 6.72 10 0.060340720 7 400 10 0.61 4.40 30 1.784474676 8 400 20 0.56 6.24 20 3.582649959 $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ $\vdots $ 675 400 30 0.87 3.09 30 7.789864799
下载: 导出CSV
表 5 预测模型的预测误差指标
Table 5. Average error of prediction model
模型 MAE RMSE R2 顺流向 横流向 顺流向 横流向 顺流向 横流向 ZOA-RF 0.2911 0.8460 0.4318 1.8278 0.9359 0.9264 PSO-RF 0.4100 0.7992 0.5500 2.1719 0.8960 0.8961 SSA-RF 0.4181 0.7839 0.5275 2.0676 0.9044 0.9058 CLR-ZOA-RF 0.2130 0.4640 0.2862 1.2887 0.9719 0.9667
下载: 导出CSV
-
[1] 国家发展改革委, 国家能源局. “十四五”现代能源体系规划[EB/OL]. 发改能源〔2022〕210号, 2022-01-29. [2] 吴西顺, 黄文斌, 刘文超, 等. 全球天然气水合物资源潜力评价及勘查试采进展[J]. 海洋地质前沿, 2017, 33(7): 63-78. doi: 10.16028/j.1009-2722.2017.07007Wu Xishun, Huang Wenbin, Liu Wenchao, et al. World-wide progress of resource potential assessment, exploration and production test of natural gas hydrate[J]. Marine Geology Frontiers, 2017, 33(7): 63-78. doi: 10.16028/j.1009-2722.2017.07007 [3] 国家发展和改革委、能源局. 《能源技术革命创新行动计划(2016—2030年)》[J]. 设备监理, 2016(3): 5. [4] 杨进, 李磊, 宋宇, 等. 中国海洋油气钻井技术发展现状及展望[J]. 石油学报, 2023, 44(12): 2308-2318.Yang Jin, Li Lei, Song Yu, et al. Current status and prospects of offshore oil and gas drilling technology development in China[J]. Acta Petrolei Sinica, 2023, 44(12): 2308-2318. [5] 周守为, 陈伟, 李清平, 等. 深水浅层非成岩天然气水合物固态流化试采技术研究及进展[J]. 中国海上油气, 2017, 29(4): 1-8.Zhou Shouwei, Chen Wei, Li Qingping, et al. Research on the solid fluidization well testing and production for shallow non-diagenetic natural gas hydrate in deep water area[J]. China Offshore Oil and Gas, 2017, 29(4): 1-8. [6] Aitken J. XIV. An account of some experiments on rigidity produced by centrifugal force[J]. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 1878, 5(29): 81-105. doi: 10.1080/14786447808639394 [7] 曾志军, 胡卫东, 刘竟成, 等. 高温高压深井天然气测试管柱力学分析[J]. 天然气工业, 2010, 30(2): 85-87, 145.Zeng Zhijun, Hu Weidong, Liu Jingcheng, et al. A mechanical analysis of the gas testing string used in HTHP deep wells[J]. Natural Gas Industry, 2010, 30(2): 85-87, 145. [8] 李子丰. 油气井杆管柱力学研究进展与争论[J]. 石油学报, 2016, 37(4): 531-556.Li Zifeng. Research advances and debates on tubular mechanics in oil and gas wells[J]. Acta Petrolei Sinica, 2016, 37(4): 531-556. [9] 邢誉峰, 梁昆. 梁纵向与横向耦合非线性振动分析[J]. 北京航空航天大学学报, 2015, 41(8): 1359-1366.Xing Yufeng, Liang Kun. Nonlinear vibration analysis of longitudinal-transverse coupled beam[J]. Journal of Beijing University of Aeronautics and Astronautics, 2015, 41(8): 1359-1366. [10] 高德利, 王宴滨. 海洋深水钻井力学与控制技术若干研究进展[J]. 石油学报, 2019, 40(增刊2): 102-115.Gao Deli, Wang Yanbin. Some research progress in deepwater drilling mechanics and control technology[J]. Acta Petrolei Sinica, 2019, 40(S2): 102-115. [11] 张慎颜, 刘秀全, 畅元江, 等. 深水钻井平台-隔水管系统波激疲劳分析[J]. 船舶力学, 2019, 23(7): 843-850.Zhang Shenyan, Liu Xiuquan, Chang Yuanjiang, et al. Wave-loading fatigue analysis on deep-water drilling platform/riser system[J]. Journal of Ship Mechanics, 2019, 23(7): 843-850. [12] 张智奇, 任浩杰, 付世晓, 等. 均匀流作用下双柔性异管径立管响应特性试验研究[J]. 船舶力学, 2021, 25(1): 16-28.Zhang Zhiqi, Ren Haojie, Fu Shixiao, et al. Experimental study on response characteristics of double flexible pipe with different diameters under uniform flow[J]. Journal of Ship Mechanics, 2021, 25(1): 16-28. [13] Lu Y, Yu Z C, Ma Y X, et al. Fatigue damage characteristics of a flexible cylinder under concomitant excitation of time-varying axial tension and VIV[J]. Ocean Engineering, 2023, 288: 116079. doi: 10.1016/j.oceaneng.2023.116079 [14] Miwa S, Liu Y, Hibiki T, et al. Study of unsteady gas-liquid two-phase flow induced force fluctuation (Part 1: Evaluation and modeling of two-phase flow induced force fluctuation)[J]. Transactions of the JSME (in Japanese), 2014, 80(809): FE0005. doi: 10.1299/transjsme.2014fe0005 [15] 沈平川, 刘庆, 齐欢欢, 等. 非均匀两相流传热管流致振动中阻尼模拟研究[J]. 核动力工程, 2020, 41(6): 116-120. doi: 10.13832/j.jnpe.2020.06.0116Shen Pingchuan, Liu Qing, Qi Huanhuan, et al. Study on flow-induced vibration damping simulation of heat exchanger tube in non-uniform two-phase flow[J]. Nuclear Power Engineering, 2020, 41(6): 116-120. doi: 10.13832/j.jnpe.2020.06.0116 [16] Liang W X, Lou M. Numerical simulation of vortex-induced vibration of a marine riser with a multiphase internal flow considering hydrate phase transition[J]. Ocean Engineering, 2020, 216: 107758. doi: 10.1016/j.oceaneng.2020.107758 [17] Liang W X, Lou M, Zhang C, et al. Experimental investigation and phenomenological modeling of fatigue crack growth in X80 pipeline steel under random loading[J]. International Journal of Fatigue, 2024, 182: 108169. doi: 10.1016/j.ijfatigue.2024.108169 [18] Zhu H J, Gao Y, Zhao H L. Experimental investigation of slug flow-induced vibration of a flexible riser[J]. Ocean Engineering, 2019, 189: 106370. [19] 袁海燕. 提升管道系统固液两相流工程应用研究[D]. 长沙: 湖南大学, 2012. [20] 刘磊. 深海采矿水力提升固液两相流动力学特性研究[D]. 上海: 上海交通大学, 2019. [21] Guo X Q, Chen X H, Zhao L B, et al. Multi-field coupling multiple nonlinear vibration model and fatigue failure mechanism of deep-ocean mining hydraulic lifting pipe[J]. Nonlinear Dynamics, 2023, 111(18): 16777-16811. [22] 郭晓强, 姜哲夫, 杨恪伦, 等. VIV效应下深海水合物立管气-液-固三相流致振动特性[J/OL]. 西南交通大学学报, 2025-01-02. https://kns.cnki.net/KCMS/detail/detail.aspx?filename=XNJT2024123000D&dbname=CJFD&dbcode=CJFQ. [23] 赵立财. 基于BP人工神经网络的简支梁桥结构振动响应预测[J]. 沈阳工业大学学报, 2024, 46(3): 347-352.Zhao Licai. Prediction of vibration response based on BP artificial neural network for simply supported beam bridge structures[J]. Journal of Shenyang University of Technology, 2024, 46(3): 347-352. [24] 钱国超, 王山, 张家顺, 等. 基于GRNN-MC的变压器振动信号预测[J]. 电工电能新技术, 2024, 43(3): 41-48.Qian Guochao, Wang Shan, Zhang Jiashun, et al. Vibration signals prediction of power transformer based on GRNN-MC[J]. Advanced Technology of Electrical Engineering and Energy, 2024, 43(3): 41-48. [25] 苗兴园, 赵弘. 油气管道封堵机器人封堵过程的振动预测[J]. 振动与冲击, 2023, 42(13): 36-49.Miao Xingyuan, Zhao Hong. Vibration prediction of oil and gas pipeline plugging robot in plugging process[J]. Journal of Vibration and Shock, 2023, 42(13): 36-49. [26] 吴昕, 刘双白, 郝向中, 等. 供热工况下的汽轮机振动长周期预测与分析[J]. 热能动力工程, 2022, 37(3): 72-80.Wu Xin, Liu Shuangbai, Hao Xiangzhong, et al. Long period prediction and analysis of steam turbine vibration under heating condition[J]. Journal of Engineering for Thermal Energy and Power, 2022, 37(3): 72-80. [27] Ma L X, Resvanis T L, Vandiver J K. Understanding the higher harmonics of vortex-induced vibration response using a trend-constrained, machine learning approach[J]. Marine Structures, 2022, 83: 103195. [28] Xu W H, He Z Q, Zhai L B, et al. Vortex-induced vibration prediction of an inclined flexible cylinder based on machine learning methods[J]. Ocean Engineering, 2023, 282: 114956. [29] Breiman L. Random forests[J]. Machine Learning, 2001, 45(1): 5-32. [30] 姜杰, 丁国富, 邹益胜, 等. 高速列车动力学性能参数优化迭代设计[J]. 西南交通大学学报, 2025, 60(3): 679-687, 697.Jiang Jie, Ding Guofu, Zou Yisheng, et al. Iterative optimization design for dynamic performance parameters of high-speed trains[J]. Journal of Southwest Jiaotong University, 2025, 60(3): 679-687, 697. [31] 吕龙龙, 李霄洋, 廖红建, 等. 基于机器学习算法的红层软岩尺寸效应模型[J]. 西南交通大学学报, 2025, 60(6): 1390-1403.Lu Longlong, Li Xiaoyang, Liao Hongjian, et al. Size effect model for red-bed soft rock based on machine learning algorithm[J]. Journal of Southwest Jiaotong University, 2025, 60(6): 1390-1403. [32] Trojovská E, Dehghani M, Trojovský P. Zebra optimization algorithm: a new bio-inspired optimization algorithm for solving optimization algorithm[J]. IEEE Access, 2022, 10: 49445-49473. [33] 夏艺瑄, 贺兴时. 融合混沌映射和乘除算子的花授粉算法及应用[J]. 智能计算机与应用, 2024, 14(1): 76-84.Xia Yixuan, He Xingshi. Flower pollination algorithm fusing chaotic mapping and multiplication-division operators and applications[J]. Intelligent Computer and Applications, 2024, 14(1): 76-84. [34] 杨建民, 肖龙飞, 盛振邦. 海洋工程水动力学试验研究[M]. 上海: 上海交通大学出版社, 2008: 1-70. [35] 黄涛. 钻柱耦合振动的理论及试验研究[D]. 北京: 中国石油大学(北京), 2001. [36] 黄鑫, 寇鉴, 毛良杰, 等. 天然气水合物开采立管力学特性研究[J]. 西南石油大学学报(自然科学版), 2023, 45(4): 143-154.Huang Xin, Kou Jian, Mao Liangjie, et al. Mechanical characteristics of natural gas hydrate production riser[J]. Journal of Southwest Petroleum University (Science & Technology Edition), 2023, 45(4): 143-154. [37] 李畅, 阎卫东, 苑春苗. 相似理论与相似方法[M]. 沈阳: 东北大学出版社, 2022. -
下载:
点击查看大图
计量
- 文章访问数: 2
- HTML全文浏览量: 1
- PDF下载量: 0
- 被引次数: 0