基于FA-BAS-ELM的海洋油气管道外腐蚀速率预测

doi:10.16265/j.cnki.issn1003-3033.2022.02.014

摘要/Abstract

摘要：

为提高海洋油气管道外腐蚀速率预测的精度和效率,建立基于因子分析(FA)和天牛须搜索算法(BAS)的极限学习机(ELM)腐蚀速率预测模型。利用FA对影响因素数据集进行降维处理,确定预测模型的输入变量;建立ELM预测模型,并采用BAS对ELM模型的参数进行优化,避免参数取值随机性对模型预测性能的影响;以实海挂片试验为例,通过建模仿真评价模型的预测性能,并与其他模型进行对比分析。结果表明:FA-BAS-ELM预测模型的平均绝对误差(MAPE)仅为1.92%,决定系数R²高达0.994 9,相比于其他模型,该模型具有更优的预测性能。

关键词: 海洋油气管道, 外腐蚀速率, 因子分析(FA), 天牛须搜索算法(BAS), 极限学习机(ELM)

Abstract:

In order to improve prediction accuracy and efficiency of external corrosion rate of submarine oil and gas pipelines, a ELM prediction model based on FA and BAS was established. FA method was used to reduce dimension of influencing factors' data set and determine input variables of ELM network. Then, an ELM prediction model was established, and its parameters were optimized by BAS to avoid influence of randomness of parameter values on its predictive performance. Finally, with the real sea hanging film experiment as an example, predictive performance of the model was evaluated through modeling and simulation, and it was compared with other models. The results show that the mean absolute percentage error (MAPE) of FA-BAS-ELM prediction model is only 1.92, and its determination coefficient R² reaches as high as 0.994 9, indicating that the model has better prediction accuracy and performance.

Key words: submarine oil and gas pipeline, external corrosion rate, factor analysis (FA), beetle antennae search algorithm (BAS), extreme learning machine (ELM)

张新生, 常潆戈. 基于FA-BAS-ELM的海洋油气管道外腐蚀速率预测[J]. 中国安全科学学报, 2022, 32(2): 99-106.

ZHANG Xinsheng, CHANG Yingge. Prediction of external corrosion rate of offshore oil and gas pipelines based on FA-BAS-ELM[J]. China Safety Science Journal, 2022, 32(2): 99-106.

图/表 12

图1

图2

表1

表2

表3

表4

表5

表6

表7

图3

图4

表8

参考文献 20

[1]	HASAN S, SWEET L, HULTS J, et al. Corrosion risk-based subsea pipeline design[J]. The International Journal of Pressure Vessels and Piping, 2018, 159:1-14. doi: 10.1016/j.ijpvp.2017.10.003
[2]	王如君, 王天瑜. 灰色-马尔科夫链模型在埋地油气管道腐蚀预测中的应用[J]. 中国安全生产科学技术, 2015, 11(4):102-106.
	WANG Rujun, WANG Tianyu. Application of grey Markov chain model in corrosion forecast of buried oil and gas pipelines[J]. Journal of Safety Science and Technology, 2015, 11(4):102-106.
[3]	张新生, 叶晓艳. 不同初始条件的UGM(1,1)管道腐蚀预测建模研究[J]. 中国安全科学学报, 2019, 29(3):63-69. doi: 10.16265/j.cnki.issn1003-3033.2019.03.011
	ZHANG Xinsheng, YE Xiaoyan. Study on UGM(1,1) modeling for prediction of pipes corrosion under different initial conditions[J]. China Safety Science Journal, 2019, 29(3):63-69. doi: 10.16265/j.cnki.issn1003-3033.2019.03.011
[4]	骆正山, 陈晨, 王哲. 优化的Gray Markov模型在埋地管道腐蚀速率预测中的应用[J]. 腐蚀与防护, 2019, 40(5):313-317.
	LUO Zhengshan, CHEN Chen, WANG Zhe. Application of improved Gray Markov dynamic model in predicting corrosion rates of oil and gas pipeline[J]. Corrosion & Protection, 2019, 40(5):313-317.
[5]	经建芳, 李康春, 邓富康, 等. 油气管道腐蚀的灰色线性回归组合预测模型[J]. 油气储运, 2015, 34(12):1300-1304.
	JING Jianfang, LI Kangchun, DENG Fukang, et al. Prediction model of oil and gas pipeline corrosion based on grey-linear regression combination[J]. Oil & Gas Storage and Transportation, 2015, 34(12):1300-1304.
[6]	任南南, 商永滨, 陈军斌, 等. 灰色模型在采油井管材腐蚀速率预测中的应用[J]. 热加工工艺, 2020, 49(4):34-37.
	REN Nannan, SHANG Yongbin, CHEN Junbin, et al. Application of grey model in predicting corrosion rate of oil well pipe[J]. Hot Working Technology, 2020, 49(4):34-37.
[7]	许宏良, 殷苏民. 基于改进BP神经网络优化的管道腐蚀速率预测模型研究[J]. 表面技术, 2018, 47(2):177-181.
	XU Hongliang, YIN Sumin. Prediction model of pipeline corrosion rate based on improved BP neural network[J]. Surface Technology, 2018, 47(2):177-181.
[8]	JIMÉNEZ-COME M J, MARTIN De La Luzm, MATRES V. A support vector machine-based ensemble algorithm for pitting corrosion modeling of EN 1.4404 stainless steel in sodium chloride solutions[J]. Materials and Corrosion, 2019, 70(1):19-27.
[9]	王盼锋, 王寿喜, 马钢, 等. 基于PCA-PSO-SVM模型的海底多相流管道内腐蚀速率预测[J]. 安全与环境工程, 2020, 27(2):183-189.
	WANG Panfeng, WANG Shouxi, MA Gang, et al. Prediction of internal corrosion rate of submarine multiphase flow pipelines based on PCA-PSO-SVM[J]. Safety and Environmental Engineering, 2020, 27(2):183-189.
[10]	WANG Chen, MA Gang, LI Junfei, et al. Prediction of corrosion rate of submarine oil and gas pipelines based on IA-SVM model[J]. IOP Conference Series: Earth and Environmental Science, 2019, 242(2):DOI: 10.1088/1755-1315/242/2/022023. doi: 10.1088/1755-1315/242/2/022023
[11]	OH D, RACE J, OTERKUS S, et al. Burst pressure prediction of API 5L X-Grade dented pipelines using deep neural network[J]. Journal of Marine Science and Engineering, 2020, 8(10):DOI: 10.3390/jmse8100766. doi: 10.3390/jmse8100766
[12]	周德慧. 物流上市公司融资结构影响因素实证研究[J]. 商业经济研究, 2020(22):176-178.
	ZHOU Dehui. An empirical study on the influencing factors of financing structure of listed logistics companies[J]. Journal of Commercial Economics, 2020(22):176-178.
[13]	张长鲁. 基于因子分析的煤质指标降维及其可视化应用[J]. 煤炭技术, 2017, 36(1):310-313.
	ZHANG Changlu. Dimension reduction for coal quality index and its visual application based on factor analysis[J]. Coal Technology, 2017, 36(1):310-313.
[14]	高放, 包燕平, 王敏, 等. 基于FA-ELM的转炉终点磷含量预测模型[J]. 钢铁, 2020, 55(12):24-30.
	GAO Fang, BAO Yanping, WANG Min, et al. Prediction model of end-point phosphorus content of converter based on FA-ELM[J]. Iron and Steel, 2020, 55(12):24-30.
[15]	王甜甜, 刘强. 基于BAS-BP模型的风暴潮灾害损失预测[J]. 海洋环境科学, 2018, 37(3):457-463.
	WANG Tiantian, LIU Qiang. The assessment of storm surge disaster loss based on BAS-BP model[J]. Marine Environmental Science, 2018, 37(3):457-463.
[16]	CHEN Cai, QIAN Qian, FU Yunfa. Application of BAS-Elman neural network in prediction of blasting vibration velocity[J]. Procedia Computer Science, 2020, 166:491-495. doi: 10.1016/j.procs.2020.02.059
[17]	李易安, 骆正山. 基于KPCA-ICS-ELM算法的埋地管线土壤腐蚀深度预测[J]. 中国安全科学学报, 2020, 30(12):100-105. doi: 10.16265/j.cnki.issn 1003-3033.2020.12.014
	LI Yian, LUO Zhengshan. Soil corrosion depth prediction of buried pipelines based on KPCA-ICS-ELM algorithm[J]. China Safety Science Journal, 2020, 30(12):100-105. doi: 10.16265/j.cnki.issn 1003-3033.2020.12.014
[18]	温廷新, 靳露露. 基于预处理的IFOA-ELM煤与瓦斯突出预测模型[J]. 中国安全科学学报, 2020, 30(1):35-41. doi: 10.16265/j.cnki.issn1003-3033.2020.01.006
	WEN Tingxin, JIN Lulu. IFOA-ELM prediction model of coal and gas outburst based on preprocessing[J]. China Safety Science Journal, 2020, 30(1):35-41. doi: 10.16265/j.cnki.issn1003-3033.2020.01.006
[19]	骆正山, 田珮琦. RS-PSO-ELM下腐蚀管道失效压力预测[J]. 中国安全科学学报, 2021, 31(3):28-34. doi: 10.16265/j.cnki.issn1003-3033.2021.03.004
	LUO Zhengshan, TIAN Peiqi. Prediction of failure pressure of corrosion pipelines based on RS-PSO-ELM[J]. China Safety Science Journal, 2021, 31(3):28-34. doi: 10.16265/j.cnki.issn1003-3033.2021.03.004
[20]	邓志安, 李姝仪, 李晓坤, 等. 基于模糊神经网络的海洋管线腐蚀速率预测新方法[J]. 中国腐蚀与防护学报, 2015, 35(6):571-576.
	DENG Zhi'an, LI Shuyi, LI Xiaokun, et al. A prediction method based on fuzzy neural network for corrosion rate of marine pipelines[J]. Journal of Chinese Society for Corrosion and Protection, 2015, 35(6):571-576.

序号	V₁/℃	V₂/(mg·L^-1)	V₃/(10³mg·L^-1)	V₄	V₅/mV	V₆/(μA·cm^-2)
1	25.90	6.71	30.10	5.10	378	16.40
2	29.35	6.09	29.00	6.30	400	16.90
3	27.90	6.18	31.50	7.00	363	15.57
4	24.00	7.95	30.20	8.10	324	13.65
5	28.00	5.05	31.40	9.20	240	13.24
︙	︙	︙	︙	︙	︙	︙
46	24.31	6.42	40.67	7.88	250	8.75
47	24.11	6.38	41.00	7.98	228	8.99
48	17.45	7.48	34.08	8.10	135	17.05
49	21.95	8.28	34.64	7.95	113	17.34
50	27.19	4.91	33.50	7.99	275	15.48

成分	初始特征值			旋转载荷平方和
成分	特征值	方差比/%	累积比/%	特征值	方差比/%	累计比/%
1	1.849	36.989	36.989	1.389	27.789	27.789
2	1.245	24.898	61.887	1.09	21.807	49.596
3	0.871	17.428	79.315	1.075	21.494	71.09
4	0.59	11.808	91.123	1.002	20.033	91.123
5	0.444	8.877	100	—

变量	旋转前				旋转后
变量	F₁	F₂	F₃	F₄	F₁	F₂	F₃	F₄
V₄	0.828	0.112	-0.182	-0.209	-0.894	-0.24	-0.106	-0.026
V₅	0.777	0.091	-0.03	0.606	0.73	-0.316	0.371	-0.064
V₂	-0.744	0.327	-0.248	0.382	0.044	0.961	-0.027	-0.105
V₁	-0.071	-0.821	0.475	0.18	0.235	-0.018	0.961	0.032
V₃	0.031	0.666	0.742	-0.014	-0.011	-0.097	0.024	0.992

变量	F₁	F₂	F₃	F₄
V₁	-0.287	0.125	1.068	0.006
V₂	-0.025	0.902	0.102	0.045
V₃	0.02	0.067	-0.014	1.003
V₄	-0.773	-0.18	0.306	-0.092
V₅	0.518	-0.314	0.024	-0.102

序号	F₁	F₂	F₃	F₄
1	2.933 4	0.569 0	-1.110 6	0.170 7
2	1.541 7	0.060 8	0.571 9	-0.182 3
3	0.788 4	0.033 6	0.315 9	0.182 3
4	-0.170 6	0.563 4	-0.563 9	-0.081 7
5	-2.071 7	-0.581 1	1.122 2	0.016 5
︙	︙	︙	︙	︙
46	-0.342 8	0.275 4	-0.664 1	1.694 9
47	-0.544 0	0.303 5	-0.705 3	1.759 9
48	-0.578 9	0.752 0	-2.995 2	0.735 8
49	-0.995 1	1.430 3	-1.405 6	0.896 1
50	-0.592 1	-0.498 1	0.354 3	0.457 5