Sea ice collision risk assessment based on Bayesian network modeling

doi:10.23919/JSEE.2026.000036

Journal of Systems Engineering and Electronics ›› 2026, Vol. 37 ›› Issue (2): 579-593.doi: 10.23919/JSEE.2026.000036

• SYSTEMS ENGINEERING • Previous Articles

Sea ice collision risk assessment based on Bayesian network modeling

Xianling LI¹(), Zixin WANG¹(), Haibin ZHANG²(), Jinhui HE²(), Yanlin WANG¹(), Xiaoming HUANG¹^,*()

¹School of Chemical Engineering, Ocean and Life Sciences, Dalian University of Technology, Panjin 124221, China
²Marine Design and Research Institute of China, Shanghai 200011, China

Received:2025-08-25 Online:2026-04-18 Published:2026-04-30
Contact: Xiaoming HUANG E-mail:lixianling@mail.dlut.edu.cn;totoro_vincent@163.com;zhanghaibin@maric.com.cn;hejinhui@maric.com.cn;wangyl@dlut.edu.cn;huangxm@dlut.edu.cn
About author:
LI Xianling was born in 2002. He received his B.E. degree from the College of Water Conservancy and Hydropower Engineering, Sichuan Agricultural University in 2024. He is currently pursuing his M.E. degree from the School of Chemical Engineering, Ocean and Life sciences, Dalian University of Technology, China. His research interests are research on the risk of collision between polar drilling vessels and sea ice, and reliability analysis of polar open deck equipment operation process and storage time. E-mail: lixianling@mail.dlut.edu.cn

WANG Zixin was born in 1997. He received his B.E. degree from North University of China, Taiyuan in 2019, and M.E. degree from Dalian University of Technology in 2024. His research interests are leveraging cutting-edge technology to improve the performance and adaptability of quadruped robots. E-mail: totoro_vincent@163.com

ZHANG Haibin was born in 1976. He received his B.S. degree, M.S. degree, and Ph.D. degree from Harbin Engineering University in 1998, 2001, and 2004, respectively. Currently, he is a deputy chief engineer with the Marine Design R&D Research Institute of China. His research interests are the R&D and design of ships and marine engineering equipment. E-mail: zhanghaibin@maric.com.cn

HE Jinhui was born in 1986. He received his B.S. degree from Huazhong University of Science and Technology in 2009, M.S. degree from the Marine Design &Research Institute of China in 2012, and Ph.D. degree from Shanghai Jiao Tong University in 2021. He is a senior engineer from Marine Design & Research Institute of China. His research interest is overall ship design. E-mail: hejinhui@maric.com.cn

WANG Yanlin was born in 1981. He received his B.S. degree, M.S. degree, and Ph.D. degree from Dalian University of Technology in 2004, 2006, and 2011, respectively. Currently, he is an associate professor with the School of Chemical Engineering, Ocean and Life Sciences, Dalian University of Technology. His research interest is monitoring, analysis and evaluation techniques for offshore engineering structures in cold regions. E-mail: wangyl@dlut.edu.cn

HUANG Xiaoming was born in 1987. She received her B.S. degree, and Ph.D. degree from Dalian University of Technology in 2010, and 2015, respectively. Currently, she is an associate professor with the School of Chemical Engineering, Ocean and Life Sciences form Dalian University of Technology. Her research interest is software and hardware design for marine transport equipment. E-mail: huangxm@dlut.edu.cn
Supported by:
This work was supported by the High-tech Ship Projects of the Ministry of Industry and Information Technology of China (CBG2N21-4-1).

Abstract

Abstract:

To address the problem of sea ice collisions threatening offshore drilling operations in polar regions, this paper proposes a Bayesian network–based collision risk assessment model for drillships. The model integrates large ice floe/iceberg conditions, natural environmental factors, and geometric factors derived from the ship’s shape, size, distance, and azimuth. Using iceberg routes, scenario simulations are conducted to evaluate collision probabilities and provide time-dependent risk values. Results demonstrate that the method yields reasonable and consistent assessments of drillship–ice interactions. The proposed method enables automatic collision risk assessment and can be applied to unattended management systems to enhance the safety of polar drilling operations.

Key words: Bayesian network, risk assessment, drillship, ice floe/iceberg collision

Xianling LI, Zixin WANG, Haibin ZHANG, Jinhui HE, Yanlin WANG, Xiaoming HUANG. Sea ice collision risk assessment based on Bayesian network modeling[J]. Journal of Systems Engineering and Electronics, 2026, 37(2): 579-593.

Figures/Tables 21

Fig 1

Fig 2

Table 1

Fig 3

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

Table 8

Fig 4

Table 9

Table 10

Table 11

Fig 5

Fig 6

Fig 7

Table 12

Fig 8

Table 13

References 35

1	LIU D H. Research on multi-dimensional optimization design and evaluation method of anti-ice performance of semi-submersible drilling platform. Hangzhou: Zhejiang University, 2020. (in Chinese)
2	LU L L, GOERLANDT F, TABRI K, et al Critical aspects for collision induced oil spill response and recovery system in ice conditions: a model-based analysis. Journal of Loss Prevention in the Process Industries, 2020, 66, 104198. doi: 10.1016/j.jlp.2020.104198
3	LIU W P, YAN D W, PENG Z K, et al. Vessel safety navigation under the influence of antarctic sea ice. Journal of Marine Science and Engineering, 2025, 13(7): 1267.
4	DONG H B, BIAN X, CHAI W, et al Hierarchical model-based risk assessment of polar ship-iceberg collision. Chinese Journal of Ship Research, 2024, 19 (S2): 90- 99.
5	QU Y H, FENG G Q, XU Y, et al Research on ice load caused by collision between large floating ice and ship. Ship Engineering, 2023, 45 (4): 79- 87.
6	ANDERSSON L E, SCIBILIA F, COPLAND L, et al Comparison of statistical iceberg forecast models. Cold Regions Science and Technology, 2018, 155, 69- 89. doi: 10.1016/j.coldregions.2018.07.003
7	BAADSHAUG O. Iceberg drift-trajectory modelling and probability distributions of the predictions. tromsø: The Arctic University of Norway, 2018.
8	WESCHE C, DIERKING W Estimating iceberg paths using a wind-driven drift model. Cold Regions Science and Technology, 2016, 125, 31- 39. doi: 10.1016/j.coldregions.2016.01.008
9	THALAGALA R G, SILVA O D, OLDFORD D, et al Performance evaluation of deep learning image classification modules in the MUN-ABSAI ice risk management architecture. Sensors, 2025, 25 (2): 326. doi: 10.3390/s25020326
10	ZHANG Y Y, LI G Y, ZHU J F, et al. Machine learning-based binary classification models for low ice-class vessels navigation risk assessment. Journal of Marine Science and Engineering, 2025, 13(8): 1408.
11	ZHANG X Y, ZHOU Y, LU Y, et al A risk assessment model for ships besetting in ice along the Arctic northeast passage. Ocean Engineering, 2025, 333, 121434. doi: 10.1016/j.oceaneng.2025.121434
12	AFENYO M, KHAN F, VEITCH B, et al Arctic shipping accident scenario analysis using Bayesian network approach. Ocean Engineering, 2017, 133, 224- 230. doi: 10.1016/j.oceaneng.2017.02.002
13	LI Z, HU S P, GAO G P, et al Risk reasoning from factor correlation of maritime traffic under Arctic sea ice status association with a Bayesian belief network. Sustainability, 2020, 13 (1): 147. doi: 10.3390/su13010147
14	ZHANG C, ZHANG D, ZHANG M Y, et al An integrated risk assessment model for safe Arctic navigation. Transportation Research Part A: Policy and Practice, 2020, 142, 101- 114. doi: 10.1016/j.tra.2020.10.017
15	LI Z, ZHU X M, LI R H, et al A comprehensive method for causation analysis of ship–ice collision risk in Arctic waters. Environmental Science and Pollution Research, 2024, 31 (28): 40366- 40382.
16	HSIEH T H, XU H, WANG S, et al. Quantitative analysis of ice navigation risk: multi-scale study from the Baltic sea region. Journal of Navigation, 2025: 1−22. DOI: 10.1017/S0373463325101069.
17	CHENG K, LI W F, ZHOU Y H, et al A novel fuzzy comprehensive evaluation method for ship collision risk. Ocean Engineering, 2025, 332, 121462. doi: 10.1016/j.oceaneng.2025.121462
18	LEE H J, NAMGUNG H A CDC–ANFIS-based model for assessing ship collision risk in autonomous navigation. Journal of Marine Science and Engineering, 2025, 13 (8): 1492. doi: 10.3390/jmse13081492
19	YU Q, LIU K Z, YANG Z S, et al Geometrical risk evaluation of the collisions between ships and offshore installations using rule-based Bayesian reasoning. Reliability Engineering & System Safety, 2021, 210, 107474.
20	LI K Y, CHEN L, CAI X X, et al A Bayesian network approach to predicting severity status in nuclear reactor accidents with resilience to missing data. Energies, 2025, 18 (11): 2684. doi: 10.3390/en18112684
21	GOVENDER I H, SAHLIN U, O'BRIEN G C. Bayesian network applications for sustainable holistic water resources management: modeling opportunities for south Africa. Risk Analysis, 2022, 42(6): 1346−1364.
22	KHAN B, KHAN F, VEITCH B A dynamic Bayesian network model for ship-ice collision risk in the Arctic waters. Safety Science, 2020, 130, 104858. doi: 10.1016/j.ssci.2020.104858
23	IRAD B. Bayesian networks. Hoboken: Wiley Online Library, 2008.
24	KUBYSHKIN N V, BUZIN I V, SKUTIN A A, et al. Determination of the area of generation of big icebergs in the barents sea-temperature distribution analysis. Proc. of the ISOPE International Ocean and Polar Engineering Conference, 2006: ISOPE-I-06-109.
25	KUBAT I, SAYED M, SAVAGE S B, et al. An operational iceberg deterioration model. Proc. of the ISOPE International Ocean and Polar Engineering Conference, 2007: ISOPE-I-07-282.
26	LØSET S, CARSTENS T Sea ice and iceberg observations in the western Barents Sea in 1987. Cold Regions Science and Technology, 1996, 24 (4): 323- 340. doi: 10.1016/0165-232X(95)00029-B
27	XU S, KIM E, HAUGEN S, et al A Bayesian network risk model for predicting ship besetting in ice during convoy operations along the Northern Sea route. Reliability Engineering & System Safety, 2022, 223, 108475.
28	ISO/FDIS 35104. Petroleum and natural gas industries -Arctic operations -Ice management. Geneva: International Organization for Standardization, 2018.
29	WANG L Y, HE Y J, ZHANG B, et al Retrieval of Arctic Sea ice drift using HY-2 satellite scanning microwave radiometer data. Haiyang Xuebao, 2017, 39 (9): 110- 120.
30	WELLS D. Big rigs: how to move the world’s largest rigs. https://www.offshore-technology.com.
31	LARSEN P H, HANSEN M O, BUUS-HINKLER J, et al Field tracking (GPS) of ten icebergs in eastern Baffin Bay, offshore Upernavik, northwest Greenland. Journal of Glaciology, 2015, 61 (227): 421- 437. doi: 10.3189/2015JoG14J216
32	GOERLANDT F, MONTEWKA J, KUZMIN V, et al A risk-informed ship collision alert system: framework and application. Safety Science, 2015, 77, 182- 204. doi: 10.1016/j.ssci.2015.03.015
33	ISO/FDIS 19906. Petroleum and natural gas industries - Arctic offshore structures. Geneva: International Organization for Standardization, 2010.
34	European Centre for Medium-range Weather Forecasts. ERA5 monthly averaged data on single levels: monthly means from 2014 to 2023. https://cds.climate.copernicus.eu.
35	European Centre for Medium-range Weather Forecasts. Arctic reanalysis data on single levels. https://cds.climate.copernicus.eu.

Top event	Intermediate event	Main/basic event	Description
Collision risk	Iceberg condition	Iceberg type	Icebergs are classified by mass
		Iceberg speed/(m/s)	Iceberg drift speed
		Iceberg distance/(n mile)	Closest distance from the iceberg to the drilling ship
		Iceberg azimuth/(°)	Iceberg drift direction and drilling ship azimuth
	Ice floe condition	Ice floe type	Large ice floes are classified by thickness
		Ice floe speed/(m/s)	Large ice floes drift speed
		Ice floe distance/mile	Shortest distance from the large ice floes to the drilling ship
		Ice floe azimuth/(°)	Large ice floes drift direction and drilling ship azimuth
	Natural environmen	Sea state	Sea state level (Douglas level)
		Temperature state/(°C)	Ambient temperature
		Visibility/km	Horizontal visibility

Risk factor	Risk level
Risk factor	Low	Middle	High	Very high
Large ice floes speed/(m/s)	<0.05	(0.05,0.1)	(0.1,0.2)	>0.2
Relative distance/(n mile)	>6	(2,6)	(0.5,2)	<0.5
Relative azimuth/(°)	<−90	(−90,−45)	(−45,−22.5)	(−22.5,22.5)
Relative azimuth/(°)	>90	(45,90)	(22.5,45)	(−22.5,22.5)
Thickness of large ice floes/m	<0.3	(0.3,0.5)	(0.5,1.5)	>1.5

Risk factor	Risk level
Risk factor	Low	Middle	High	Very high
Icebergs speed/(m/s)	<0.1	(0.1,0.25)	(0.25,0.5)	>0.5
Relative distance/(n mile)	>6	(2,6)	(0.5,2)	<0.5
Relative azimuth/(°)	<−90	(−90,−45)	(−45,−22.5)	(−22.5,22.5)
Relative azimuth/(°)	>90	(45,90)	(22.5,45)	(−22.5,22.5)
Icebergs mass/t	<8×10³	(8×10³,4×10⁵)	(4×10⁵,5×10⁶)	>5×10⁶

Risk factor	Risk level
Risk factor	Low	Middle	High
sea state/Douglas classification of wave	<3	3−5	>5
Temperature /(°C)	>5	2−5	<2
Visibility/km	>2	1−2	<1

Sea state level	Wave height / m	Wave height / ft
0 (Calm)	None	None
1 (Slight)	0−0.1	0.00−0.33
2 (Mild)	0.10−0.50	0.33−1.64
3 (Moderate)	0.50−1.25	1.64−4.10
4 (Larger)	1.25−2.50	4.10−8.20
5 (Huge waves)	2.50−4.00	8.20−13.1
6 (Wild waves)	4.00−6.00	13.1−19.7
7 (Storm)	6.00−9.00	19.7−29.5
8 (Berserk)	9.00−14.00	29.5−45.9
9 (Giant storm)	>14.00	>45.9

Sea ice collision risk assessment based on Bayesian network modeling

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

Share this article

Figures/Tables 21

References 35

Related Articles 15

Recommended Articles

Metrics

Comments

Risk variable	Range	Risk value after discretization
Low risk	[0,0.333)	L
Medium risk	[0.333,0.666)	M
High risk	[0.666,1]	H

Risk factor	Subjective method weight	Entropy method weight	Combined weights
Sea state	30	36.88	34.13
Temperature	15	32.82	25.69
Visibility	55	30.30	40.18

Sample size	Risk node
Sample size	Sea state	Temperature	Visibility	Natural environment
1	M	H	L	M
2	M	M	M	H
3	M	L	L	M
4	M	M	L	M
$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $
120	L	L	M	M

Scenario category	Detailed information
Drillship length/m	228
Drillship width/m	42
Drillship position	(0,5000)
Icebergs mass/t	90000
Icebergs initial position	(0,0)
North direction/(°)	0
Icebergs starting direction/(°)	67.17

Observation time/h	Speed/(m/s)	Relative distance/(n mile)	Relative azimuth/(°)		Movement direction/(°)
Observation time/h	Speed/(m/s)	Relative distance/(n mile)	α	β	Movement direction/(°)
0	0.20	2.70	28.25	27.75	22.83
2	0.15	2.43	22.27	21.45	18.77
4	0.09	2.12	14.41	13.47	7.56
6	0.05	1.90	0.99	−0.05	−7.36
8	0.12	1.60	−4.49	−5.35	−12.89
10	0.18	0.70	−49.11	−86.2	−13.66
12	0.22	1.23	−170.26	−172.91	−13.09
14	0.25	2.51	−177.54	−178.93	−15.03
16	0.29	4.01	−177.54	−178.57	−16.72
18	0.44	5.63	178.39	177.80	−18.19
20	0.50	7.37	177.47	177.01	−19.50
22	0.52	9.21	176.83	176.46	−20.65
24	0.46	11.17	175.86	175.55	−22.18

Risk node	Accuracy 1	Accuracy 2	Accuracy 3	Average accuracy
Natural environment	99.17	95.83	97.50	97.50

Risk node	Sensitivity factor	Sorting
Relative distance (icebergs)	0.062	1
Icebergs speed	0.054	2
Relative distance (large ice floes)	0.052	3
Icebergs type	0.045	4
Large ice floes speed	0.040	5
Large ice floes type	0.030	6
Relative azimuth (icebergs)	0.018	7
Visibility	0.016	8
Relative azimuth (large ice floes)	0.013	9
Sea state	0.009	10
Temperature	0.005	11

[1]	Jianjun ZHU, Lin JIANG. Performance improvement method of new R&D institutions considering Bayesian network [J]. Journal of Systems Engineering and Electronics, 2026, 37(1): 257-271.
[2]	Jingwei DONG, Wei TANG, Minggang YU. A method for modeling and evaluating the interoperability of multi-agent systems based on hierarchical weighted networks [J]. Journal of Systems Engineering and Electronics, 2025, 36(3): 754-767.
[3]	Weijian PANG, Xinyi MA, Xueming LIANG, Xiaogang LIU, Erwa DONG. Role-based Bayesian decision framework for autonomous unmanned systems [J]. Journal of Systems Engineering and Electronics, 2023, 34(6): 1397-1408.
[4]	Shunqi YANG, Ying ZENG, Xiang LI, Yanfeng LI, Hongzhong HUANG. Reliability analysis for wireless communication networks via dynamic Bayesian network [J]. Journal of Systems Engineering and Electronics, 2023, 34(5): 1368-1374.
[5]	Renjie XU, Xin LIU, Donghao CUI, Jian XIE, Lin GONG. An evaluation method of contribution rate based on fuzzy Bayesian networks for equipment system-of-systems architecture [J]. Journal of Systems Engineering and Electronics, 2023, 34(3): 574-587.
[6]	Honghong ZHANG, Xusheng GAN, Shuangfeng LI, Zhiyuan CHEN. UAV safe route planning based on PSO-BAS algorithm [J]. Journal of Systems Engineering and Electronics, 2022, 33(5): 1151-1160.
[7]	Fang YE, Ying MAO, Yibing LI, Xinrui LIU. Target threat estimation based on discrete dynamic Bayesian networks with small samples [J]. Journal of Systems Engineering and Electronics, 2022, 33(5): 1135-1142.
[8]	Xiangyuan TAN, Xiaoguang GAO, Chuchao HE, Zidong WANG. Causal constraint pruning for exact learning of Bayesian network structure [J]. Journal of Systems Engineering and Electronics, 2021, 32(4): 854-872.
[9]	Yu YANG, Xiaoguang GAO, Zhigao GUO. Finding optimal Bayesian networks by a layered learning method [J]. Journal of Systems Engineering and Electronics, 2019, 30(5): 946-958.
[10]	Xiaoguang GAO, Yu YANG, Zhigao GUO. Learning Bayesian networks by constrained Bayesian estimation [J]. Journal of Systems Engineering and Electronics, 2019, 30(3): 511-524.
[11]	Chuchao HE, Xiaoguang GAO, Zhigao GUO. Structure learning on Bayesian networks by finding the optimal ordering with and without priors [J]. Journal of Systems Engineering and Electronics, 2018, 29(6): 1209-1227.
[12]	Ruohai Di, Xiaoguang Gao, and Zhigao Guo. Learning Bayesian network parameters under new monotonic constraints#br# [J]. Journal of Systems Engineering and Electronics, 2017, 28(6): 1248-1255.
[13]	Binghua Song, Zhongbao Zhou, Chaoqun Ma, Jinglun Zhou, and Shaofeng Geng. Reliability analysis of monotone coherent multi-state systems based on Bayesian networks [J]. Journal of Systems Engineering and Electronics, 2016, 27(6): 1326-1335.
[14]	Zhiqiang Cai, Shubin Si, Shudong Sun, and Hongyan Dui. Learning Bayesian network structure with immune algorithm [J]. Journal of Systems Engineering and Electronics, 2015, 26(2): 282-291.
[15]	Zhengdao Zhang, Jinlin Zhu, and Feng Pan. Fault detection and diagnosis for data incomplete industrial systems with new Bayesian network approach [J]. Journal of Systems Engineering and Electronics, 2013, 24(3): 500-.

Observation time/h	Collision risk
Observation time/h	L	M	H
0	0.29	0.37	0.34
2	0.26	0.38	0.36
4	0.28	0.36	0.36
6	0.23	0.38	0.41
8	0.19	0.39	0.42
10	0.13	0.33	0.54
12	0.30	0.37	0.33
14	0.35	0.36	0.29
16	0.41	0.34	0.25
18	0.45	0.34	0.21
20	0.51	0.33	0.16
22	0.58	0.29	0.13
24	0.62	0.28	0.10