Metal magnetic memory quantitative inversion model based on IPSO-GRU algorithm for detecting submarine pipeline defect
-
摘要: 针对海底管道缺陷磁记忆定量反演的难题，提出一种基于改进粒子群优化的门控循环神经网络模型，即IPSO-GRU模型。以两端焊有盲板的X52管道作为实验材料，其上预制有不同直径、深度的缺陷，采用TSC-5M-32磁记忆检测仪，外接11-6W非接触探头，进行水下磁记忆检测试验，提取不同缺陷尺寸的磁记忆信号特征值。考虑到磁记忆信号特征值随缺陷尺寸呈复杂的非线性变化,引入门控循环神经网络，利用其双门结构能够记忆缺陷处的信号特征，非线性回归拟合能力强的特点，构建海底管道缺陷定量反演模型，进一步考虑到模型超参数选择的随机性，采用改进粒子群算法进行超参数寻优。验证结果表明：该模型对缺陷深度反演平均精度达96%；对缺陷直径反演平均精度达93%，为海底管道缺陷的磁记忆定量化识别与反演提供了新的思路和方法。Abstract: As submarine oil and gas are exploited further, the safety of submarine pipelines is receiving increasing attention. Due to the complex operating environment and harsh working environment, submarine pipelines are vulnerable to damage; this leads to accidents. Once an accident occurs in the submarine pipeline, it not only causes massive economic losses but also adversely affects marine ecology. The metal magnetic memory (MMM) technology was proposed in the 20th century to detect macro defects and hidden defects early. To overcome the difficulties of the MMM quantitative inversion of submarine pipeline defects, this study proposed a gated recurrent unit (GRU) neural network model based on improved particle swarm optimization (IPSO). The X52 pipe specimens with blind plates that were welded at both ends were used, pipes had prefabricated defects of different diameters and depths. An 11-6W noncontact probe was used for underwater testing; the host was the TSC-5M-32 MMM Instrument. After conducting simulated submarine tests to obtain the MMM signals of pipe defects, the characteristic parameters of MMM signals with different defect sizes were extracted. It is found that the MMM characteristic parameters exhibit a complex nonlinear variation for different defect dimensions. Exploiting the GRU’s dual-gate structure that can remember the signal characteristics of defects and its superior nonlinear regression fitting ability, a quantitative MMM GRU inversion model was established for detecting submarine pipeline defects. Furthermore, considering the randomness of the hyper-parameter selection in the model, the IPSO algorithm was used to optimize the hyper-parameters. Validation results show that the model has an average accuracy of up to 96% and 93% for defect depth inversion and defect diameter inversion, respectively. Using the MMM method, this study provides a new idea and method for the quantitative identification and defect inversion of submarine pipeline defects.
-
图 1 GRU结构图。（a）GRU神经网络结构；（b）GRU单元结构
Figure 1. GRU structure: (a) illustration of GRU neural network structure; (b) illustration of GRU cell
图 3 实验设备。（a）试验管道；（b）实验流程图
Figure 3. Experimental equipment: (a) pipe specimen; (b) experimental flowchart
图 4 ΔHp与不同缺陷尺寸的关系。（a）缺陷直径；（b）缺陷深度
Figure 4. Relationship between ΔHp and defect dimensions: (a) ΔHp vs defect diameter; (b) ΔHp vs defect depth
图 5 ΔHp/Δx与不同缺陷尺寸的关系。（a）缺陷直径；（b）缺陷深度
Figure 5. Relationship between ΔHp/Δx and defect dimensions: (a) ΔHp/Δx vs defect dimensions diameter; (b) ΔHp/Δx vs defect dimensions depth
图 6 Ka与缺陷尺寸的关系。（a）Ka与缺陷直径的关系；（b）Ka与缺陷深度的关系
Figure 6. Relationship between Ka and defect dimensions: (a) Ka vs defect diameter；(b) Ka vs defect depth
图 7 IPSO寻优图。（a）PSO与IPSO寻优能力对比图；（b）IPSO适应度迭代曲线；（c）第一隐含层节点数；（d）第二隐含层节点；（e）学习率
Figure 7. IPSO optimization graph: (a) optimization capability comparison of PSO vs IPSO; (b) IPSO fitness value iteration curve; (c) first hidden nodes value; (d) second hidden nodes value; (e) learning rate
图 8 模型反演相对误差。（a）缺陷深度相对误差；（b）缺陷直径相对误差
Figure 8. Relative error graph of the model inversion: (a) defect depth; (b) defect diameter
表 1 部分原始数据
Table 1. Partial raw data
Data number Defect diameter/mm Defect depth/mm ΔHp/(A∙m−1) （ΔHp/Δx）/(A∙m−1∙mm−1) Ka/(A∙m−1∙mm−1) Pipe pressure /MPa x1 defect free defect free 5.89 4.25 0.87 0 x2 defect free defect free 6.36 3.87 1.62 8 x3 10 2 16.21 4.13 24.16 0 x4 10 2 36.33 6.55 48.42 8 x5 10 3 18.45 6.24 33.17 0 x6 10 3 32.16 12.67 52.94 8 x7 10 4 26.06 6.55 39.62 0 x8 10 4 47.81 8.63 56.56 8 x9 5 4 13.77 5.29 21.91 0 x10 5 4 23.15 10.93 38.76 8 x11 15 4 35.85 8.72 62.5 0 x12 15 4 78.63 17.66 100.31 8 
下载: 导出CSV
-
参考文献
[1] Wang H H, Liu G H. Statistics and analysis of subsea pipeline accidents of CNOOC. China Offshore Oil Gas, 2017, 29(5): 157王红红, 刘国恒. 中国海油海底管道事故统计及分析. 中国海上油气, 2017, 29(5):157 [2] Dubov A A. A study of metal properties using the method of magnetic memory. Met Sci Heat Treat, 1997, 39(9): 401 doi: 10.1007/BF02469065 [3] Su S Q, Liu X W, Wang W, et al. Progress and key problems in the research on metal magnetic memory testing technology. Chin J Eng, 2020, 42(12): 1557苏三庆, 刘馨为, 王威, 等. 金属磁记忆检测技术研究新进展与关键问题. 工程科学学报, 2020, 42(12):1557 [4] Min X H, Yang L J, Wang G Q, et al. Weak magnetism stress internal testing technology of the long distance oil-gas pipeline. J Mech Eng, 2017, 53(12): 19 doi: 10.3901/JME.2017.12.019闵希华, 杨理践, 王国庆, 等. 长输油气管道弱磁应力内检测技术. 机械工程学报, 2017, 53(12):19 doi: 10.3901/JME.2017.12.019 [5] Liu B, Cao Y, Wang D, et al. Quantitative analysis of the magnetic memory yielding signal characteristics based on the LMTO algorithm. Chin J Sci Instrum, 2017, 38(6): 1413 doi: 10.3969/j.issn.0254-3087.2017.06.012刘斌, 曹阳, 王缔, 等. 基于LMTO算法磁记忆屈服信号的定量化分析. 仪器仪表学报, 2017, 38(6):1413 doi: 10.3969/j.issn.0254-3087.2017.06.012 [6] Venkatachalapathi N, basha S M J, Raju G J, et al. Characterization of fatigued steel states with metal magnetic memory method. Mater Today:Proc, 2018, 5(2): 8645 doi: 10.1016/j.matpr.2018.04.002 [7] Jiao J N. Damage Testing of the Metal Pipeline Based on Metal Magnetic Memory and Research on Evaluation Methed [Dissertation]. Harbin: Harbin Institute of Technology, 2015焦江娜. 基于磁记忆的金属管道损伤检测及评估方法研究[学位论文]. 哈尔滨: 哈尔滨工业大学, 2015 [8] Shi P P, Hao S. Analytical solution of magneto-mechanical magnetic dipole model for metal magnetic memory method. Acta Phys Sin, 2021, 70(3): 105时朋朋, 郝帅. 磁记忆检测的力磁耦合型磁偶极子理论及解析解. 物理学报, 2021, 70(3):105 [9] Bao S, Zhao Z Y, Jin P F, et al. Study on the characteristics of magnetic memory signals at ferromagnetic material defects. Eng Mech, 2020, 37(Sup 1): 371包胜, 赵政烨, 金鹏飞, 等. 铁磁性材料缺陷处的磁记忆信号特征分析. 工程力学, 2020, 37(增刊1): 371 [10] Xu K S, Yang K, Liu J, et al. Study on metal magnetic memory signal of buried defect in fracture process. J Magn Magn Mater, 2020, 498: 166139 doi: 10.1016/j.jmmm.2019.166139 [11] Xu K S, Feng M G, Qiu X X. Distinguishing welding defects from the stress concentration zone using metal magnetic memory field parameters. Trans Indian Inst Met, 2019, 72(2): 343 doi: 10.1007/s12666-018-1485-7 [12] Xu K S, Liu J W, Yang K, et al. Effect of applied load and thermal treatment on the magnetic memory signal of defect-bearing Q345R steel samples. J Magn Magn Mater, 2021, 539: 168366 doi: 10.1016/j.jmmm.2021.168366 [13] Li L G, Wan Y, Wang Y, et al. Quantitative inversion of pipeline defect depth based on support vector machine and magnetic memory technology. Corros Prot, 2020, 41(1): 29 doi: 10.11973/fsyfh-202001006李立刚, 万勇, 王宇, 等. 基于支持向量机和磁记忆技术的管道缺陷深度的定量化反演研究. 腐蚀与防护, 2020, 41(1):29 doi: 10.11973/fsyfh-202001006 [14] Wan Y, Wang Y, Yang Y, et al. Multi-characteristic threshold identification method for pipeline defect types. Oil Gas Storage Transp, 2020, 39(3): 268万勇, 王宇, 杨勇, 等. 管道缺陷类型多特征量阈值识别方法. 油气储运, 2020, 39(3):268 [15] Yi F, Li Z X, Lü H Q, et al. Defect recognition by metal magnetic memory detection of pipelines based on the fuzzy kernel function SVM. Acta Petrolei Sin, 2010, 31(5): 863 doi: 10.7623/syxb201005030易方, 李著信, 吕宏庆, 等. 基于模糊核支持向量机的管道磁记忆检测缺陷识别. 石油学报, 2010, 31(5):863 doi: 10.7623/syxb201005030 [16] El-Abbasy M S, Senouci A, Zayed T, et al. A condition assessment model for oil and gas pipelines using integrated simulation and analytic network process. Struct Infrastructure Eng, 2015, 11(3): 263 doi: 10.1080/15732479.2013.873471 [17] Wang S, Huang H H, Han G, et al. Quantitative evaluation of magnetic memory signal based on PCA & GA-BP neural network. J Electron Meas Instrum, 2018, 32(10): 190王帅, 黄海鸿, 韩刚, 等. 基于PCA与GA-BP神经网络的磁记忆信号定量评价. 电子测量与仪器学报, 2018, 32(10):190 [18] Luo Z S, Zhao L X, Wang X W. Failure model for pitting fatigue damaged pipeline of subsea based on dynamic Bayesian network. Surf Technol, 2020, 49(1): 269骆正山, 赵乐新, 王小完. 基于动态贝叶斯网络的海底管道点蚀疲劳损伤失效模型研究. 表面技术, 2020, 49(1):269 [19] Luo Z S, Qin Y, Zhang X S, et al. Prediction of external corrosion rate of marine pipelines based on LASSO-WOA-LSSVM. Surface Technology, 2021(5): 245骆正山, 秦越, 张新生, 等. 基于LASSO-WOA-LSSVM的海洋管线外腐蚀速率预测. 表面技术, 2021(5):245 [20] Xing H Y, Chen S Y, Li S Q, et al. MMM accurate location model of early hidden damage in welded joints based on PSO and MLE. Chin J Eng, 2017, 39(10): 1559邢海燕, 陈思雨, 李思岐, 等. 基于粒子群最大似然估计的焊缝早期隐性损伤磁记忆精确定位模型. 工程科学学报, 2017, 39(10):1559 [21] Xing H Y, Chen Y H, Li X F, et al. Magnetic memory identification model of mental weld defect levels based on dynamic immune fuzzy clustering. Chin J Sci Instrum, 2019, 40(11): 225邢海燕, 陈玉环, 李雪峰, 等. 基于动态免疫模糊聚类的金属焊缝缺陷等级磁记忆识别模型. 仪器仪表学报, 2019, 40(11):225 [22] Zhou W, Fan J C, Liu X Y, et al. Metal magnetic memory testing of the drilling riser pipeline steel based on pulsating-impact-fatigue test. Mater Sci Forum, 2020, 975: 15 doi: 10.4028/www.scientific.net/MSF.975.15 [23] Zhao B X, Yao K, Wu L B, et al. Application of metal magnetic memory testing technology to the detection of stress corrosion defect. Appl Sci, 2020, 10(20): 7083 doi: 10.3390/app10207083 [24] Cho K, Merrirnboer B V, Gulcehre C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[J/OL]. arXiv. org online (2014-6-3) [2021-5-1]. https://arxiv.org/abs/1406.1078 [25] Kennedy J, Eberhart R. Particle swarm optimization // Proceedings of ICNN'95-IEEE International Conference on Neural Networks. Perth, 1995: 1942 -
点击查看大图
计量
- 文章访问数: 154
- HTML全文浏览量: 198
- PDF下载量: 24
- 被引次数: 0
