Advertisement

A solving algorithm of navigational collision risk through data analysis of fishing vessel activities

  • Yancai Hu
  • Gyei-Kark ParkEmail author
  • Thi Quynh Mai Pham
Original Article
  • 72 Downloads

Abstract

There is increasing concern that many collision accidents happened in fishing areas. The probability of collision risk increases due to the vulnerability of fishing areas where the collision accidents involving large ships may cause serious consequence. To improve collision risk assessment, this paper focuses on solving navigational collision risk based on the analysis of fishing vessel activities using Automatic Identification System (AIS) data. AIS provides a high level of solution to assess fishing activities to reduce the collision influence of fishing vessels. The identification of fishing activities should be handled among the large number of fishing vessels. Fuzzy C-Means clustering method is applied to accomplish the clusters considering the position and speed of fishing vessels. The distance to the cluster of fishing area and its size are chosen as main factors to infer the vulnerability of fishing area using fuzzy logic reasoning. Distance to the Closest Point of Approach (DCPA) and Time to the Closest Point of Approach (TCPA) are used as significant variables to calculate basic collision risk in practice. Then basic collision risk and vulnerability of fishing boats are combined to build an algorithm to solve navigational collision risk for e-Navigation. Simulation is implemented to validate its effectiveness.

Keywords

Navigational collision risk Vulnerability Fishing activities e-Navigation 

1 Introduction

International Maritime Organization (IMO) has carried out the implementation of e-Navigation since 2006. Various projects have been pursued, and many services have been developed (IMO 2018) such as the projects of European MONALISA, ACCSEAS, Japanese ENNS and SSAP. However, most of projects and services focused on large ships on international voyages. Without considering both large and small ships, the effect of e-Navigation will be limited. Fortunately, Korea has promoted the SMART-Navigation project to develop the new e-Navigation system for the international Convention for the Safety of Life at Sea (SOLAS) ships as well as fishing boats and other small ships. SMART-Navigation is promoted by seeking a route to provide both SOLAS ships and Non-SOLAS ships including fishing vessels with e-Navigation services in a harmonized way and it is an implementation of IMO’s e-Navigation. It shares and utilizes all related information to help Non-SOLAS ships in Korean waters and SOLAS ships with Korean flags to enhance safety and efficiency of maritime navigation. SMART-Navigation is developed to provide the LTE-Maritime communication network for Non-SOLAS ships in order to reduce the navigational collision risk. One of the key technologies of SMART-Navigation project is the identification of maritime traffic situations and collision risk assessment of the traffic situation involving fishing area is a way to improve the SMART-Navigation. The statement is given for this issue that Non-SOLAS vessels including fishing vessels account for more than 99% of total number of ships while SOLAS ships account for less than 1% according the survey shown in Table 1 (Ministry of Oceans and Fisheries 2016). Among the number of merchant fleet and fishing vessels, the latter is more than 85% of the total. According to the survey, in Korea, 68% of maritime accidents are related to fishing vessels, 82% of maritime accidents are caused by human errors (KMST 2016) and fishing vessels have relatively poor navigation and communication means in Korea (An 2016).
Table 1

Statistics of merchant fleet and fishing vessels in Korea

Korea (Statistical Yearbook of Ministry of Oceans and Fisheries, 2016)

SOLAS

 

0.99%

 

Non-SOLAS

99.01%

1. Merchant fleet

11,501

  

14.6%

1) Cargo ships

 

1159

   
 

(1) International

751

751

  
 

(2) Domestic

408

 

408

 

2) Passenger ships

 

1861

   
 

(1) International

29

29

  
 

(2) Domestic

1832

 

1832

 

3) Other ships (special ships)

8481

 

8481

 

2. Fishing vessels

67,226

 

67,226

85.4%

1) Marine waters

 

62,659

   

2) Inland waters

 

3101

   

3) Others

1466

   

Total

78,727

780

  
As the statistics of merchant fleet and fishing vessels in the world shown in Table 2, 98.1% of the total vessels is fishing vessels. It accounts for large proportion and leads to the causes of accidents.
Table 2

Statistics of merchant fleet and fishing vessels in the world

World (Equasis Statistics, 2015)

1. Merchant Fleet

87,233

1.9%

1) Cargo ships

 

49,948

 
 

(1) more than 500GT

43,006

 
 

(2) less than 500GT

6942

 

2) Passenger ships

 

6741

 

3) Special ships

30,544

 

2. Fishing vessels

4,606,000

98.1%

Total

4,693,233

 

Until now, many of fishing vessels don’t have communication or/and navigation means when they encounter SOLAS ships. Considering the different behaviors of fishing vessels in operation, there is still no means to harmonize SOLAS ships and fishing vessels. Therefore, it has a big impact on the collision risk in fishing area.

In order to solve the issues of fishing vessels, there are some solutions to avoid collision accidents for the merchant ships. The routes should be planned according to the recommended or customary route of navigation. The route should not be too close to the shore to avoid accidents or troubles with fishing vessels. In the voyage, the long-range of the radar is suggested to use to observe whether there is a dense fishing area in front of the route. If it is found, it should be actively circumvented and the preparation for the intensive fishing vessels group should be paid attention in advance. Active and decisive actions are of importance to guarantee avoiding collisions when being caught in an urgent situation forced by fishing vessels. In the fishing area, it is also suggested to pay attention to the dynamics of fishing vessels and take full-scale considerations of surrounding merchant ships which easily cause serious accidents. However, all this work will bring much work for the officers and it is more difficult for the officer without much experience. In order to reduce the burden of emergent situation awareness for watching fishing vessels, improvement should also be put forward.

In Fig. 1, a structure of navigation risk assessment considered the vulnerability (Kim 2017) which is defined as the probability of a marine accident or the degree of a marine accident happens. The units of environmental risk assessment and navigational risk assessments of fishing vessels are listed to enhance the awareness of fishing areas to improve SMART-Navigation according to the structure of Kim. Thus, this paper is to carry out an algorithm for solving the vulnerability of fishing operating area. Vulnerability is combined with basic collision risk to build navigational collision risk system for merchant ships and fishing vessels in dense fishing area.
Fig. 1

Awareness system for vulnerability of vessel. Source: Kim 2017

2 Literature review

There are many studies analyzing collision risk assessment. Among the studies, fuzzy logic is considered as a suitable and effective method to deal with collision risk. Hasegawa (1987) designed the fuzzy inference system using DCPA and TCPA as inputs, large quantities of papers have been proposed for the improvement and application of this method. Lee and Rhee (2001) developed a fuzzy collision avoidance system using the expert system. Moreover, an autonomous fuzzy navigation algorithm was conceived in paper of Lee et al. (2004). The modified virtual force field method is presented for eight track-keeping or collision avoidance modes based on the fuzzy rules. Then, Park and Benedictos (2006) proposed a ship collision avoidance support system using fuzzy case-based reasoning to carry out decision-making by retrieving past similar cases.

Additionally, Bukhari et al. (2013) utilized DCPA, TCPA, bearing and variance of compass degree among all vessels from the view of vessel traffic service (VTS) centre. The information extracted from conventional marine equipment was exploited to calculate and display the degree of collision risk. The radar filtration algorithm helped the VTS officer to measure collision risk around particular ships. Based on the system of Bukhari, Ali et al. (2015) presented type-2 fuzzy ontology-based semantic knowledge and a simulator to reduce experimental time and the cost of marine robots. DCPA, TCPA and variation of compass degree were still used to calculate the degree of collision risk. The study (Rao and Balakrishnan 2017) also developed an early warning system for fishing boats to avoid collision with ships by using DCPA and TCPA.

However, all these studies focused on the collision risk without considering fishing area. The majority of collision and contact incidents involved a fishing vessel and a merchant vessel (Natale et al. 2005). While fishing vessels are involved in the fishing operation, skippers cannot concentrate on plotting the position and movement of other vessels approaching them. Fishing vessels also may cause urgencies and collisions between merchant ships. Kim et al. (2013) researched a control factor for the marine casualty of fishing vessel using the risk quantitative method of marine casualty and sequentially timed event analysis for the reason finding. It found that the high-risk collision, sinking and capsizing need to be strongly controlled. The studies (Lee et al. 2013) were conducted to inquire into the cause of collision between fishing vessels and non-fishing vessels. Further, Jin (2014) examined the determinants of fishing vessel accident severity using vessel accident data.

Studies of Fukuda and Shoji (2017) and Li et al. (2018) pursued to identify the high navigational risk area based on AIS data. Both could estimate the high possibility of collision by analysing the traffic conditions. The model results of Weng et al. (2018) showed that the involvement of big ships and fishing vessels has the largest impact in increasing the probability of a serious or very serious accident. Thus, there would be a bigger probability of a serious accident in fishing area if both ships involved in the collision are merchant ships.

Vulnerability was described in the study of Berle et al. (2011). It presented a structured formal vulnerability assessment methodology seeking to transfer the safety-oriented formal safety assessment framework into the domain of maritime supply chain vulnerability. Vulnerability also can be used to describe the possibility of accidents caused by fishing vessels.

Maritime situational awareness was suggested by Mazzarella et al. (2014) to combine greatly improved the automatic identification and classification of vessel activities for the capability of understanding events, circumstances and fishing activities within and impacting the maritime environment. AIS provides the possibility to integrate and enrich the available services and information in the maritime domain. Natale et al. (2005) also assessed the data of AIS instead of vessel monitoring system and the feasibility of producing a map of fishing effort with high spatial and temporal resolution at European scale. These studies helped to improve the ability of fishing activities identification and highlight the fishing vessels clusters.

Particularly, this paper aims to solve navigational collision risk considering the vulnerability of fishing area by analysing fishing vessels activities using clustering analysis of AIS data for the collision risk assessment.

3 Framework of navigational risk solving system

The framework of navigational collision risk solving system based on vulnerability contains three modules as described in Fig. 2.
Fig. 2

Structure of the navigational collision risk solving system. Source: Authors

Firstly, DCPA and TCPA evaluating collision risk and supporting decision making, are used to calculate the collision risk. This basic collision risk can be obtained by designing the membership functions and rules of DCPA and TCPA. Secondly, in fuzzy reasoning engine 2, the distance to fishing area and the size of fishing area are used to calculate vulnerability. Finally, basic collision risk and vulnerability of fishing area are integrated to infer navigational collision risk.

3.1 Basic collision risk

DCPA and TCPA are simultaneously considered for solving basic collision risk to offer a reasonable and applicable collision risk assessment. DCPA is a significant input variable that can determine whether the encountering ships will collide with own ship if the right alteration of heading is not executed. TCPA can be used to determine the remaining time for taking collision-avoidance action. Note that to evade a possible collision, DCPA and TCPA will be considered at the same time by employment of the following equations (Jo et al. 2018).

$$ DCPA=\frac{D\left({V}_o\sin \alpha -{V}_t\sin \beta \right)}{\sqrt{V_o^2+{V}_t^2+2{V}_o{V}_t\cos \left(\alpha +\beta \right)}} $$
(1)
$$ TCPA=\frac{D\left({V}_o\cos \alpha -{V}_t\cos \beta \right)}{V_o^2+{V}_t^2+2{V}_o{V}_t\cos \left(\alpha +\beta \right)} $$
(2)
where VO is Own Ship (OS) velocity, Vt is Target Ship (TS) speed, Distance (D) is the distance from OS to TS, α is the relative bearing of TS based on OS and β is relative bearing of OS based on TS.
The calculation of collision risk has two inputs and one output, which are determined by the reasoning rules in Table 3. In this collision risk solving system, fuzzy logic will be utilized to describe such ambiguous linguistic values. The triangular membership functions are designed for the inputs and output. For the conclusion, the variables of the fuzzy rules to solve basic CR are shown as
Table 3

Reasoning rules of DCPA, TCPA and collision risk

TCPA

DCPA

VS

S

M

B

VB

VS

VB

VB

B

B

M

S

VB

B

B

M

M

M

B

B

M

M

S

B

B

M

M

S

S

VB

M

M

S

S

VS

Source: Authors

$$ \left(\mathrm{DCPA},\mathrm{TCPA}\right)\to \mathrm{Basic}\ \mathrm{collision}\ \mathrm{risk} $$
(3)

A succinct fuzzy reasoning model is used as a popular method and the membership functions for DCPA, TCPA and basic collision risk are classified as five linguistic values: Very Small (VS), Small (S), Medium (M), Big (B), Very Big (VB). The calculation of collision risk is determined by the reasoning rules. It is expressed as Multiple Inputs Single Output (MISO) system which has the rules in Table 3 (Jo et al. 2018).

The degree of basic collision risk is determined by the relation of two inputs DCPA, TCPA and single output according to the fuzzy logic reasoning rules.

In Fig. 3, it shows the fuzzy membership functions of DCPA, TCPA and basic collision risk. The centroid method is used for defuzzification (Negnevitsky 2005). If DCPA and TCPA are approaching zero, then the value of collision risk becomes bigger. Namely, when the value of output is close to one, it is riskier for the ship’s collision.
Fig. 3

Membership functions for DCPA, TCPA and basic CR. Source: Authors

3.2 Analysis of fishing vessel activities using AIS data

The Fuzzy C-Means clustering method will be used in this part to analysis the fishing operation areas. Fuzzy C-Means clustering method was developed by Dunn in 1973 and improved by Bezdek in 1981 to deal with the problem of overlapping clusters which cannot be solved in the classical models. Fuzzy C-Means clustering method can use Fuzzy theory to assign data to a plurality of clusters using the membership degree between 0 and 1 without belonging to a specific cluster.

Let X = {x1,x2,…xN} be the set of data and V = {v1, v2,…vC} be the set of clusters’ centers in a p dimensional space where p is the number of data properties, N is the number of data and C is the number of clusters. Centroids are used as centers in describing the clusters.

Condition for a fuzzy partition matrix are given by:
$$ {\upmu}_{\mathrm{i}}\in \left[0,1\right],1\le \mathrm{i}\le \mathrm{C}, $$
(4)
$$ \sum \limits_{\mathrm{i}=1}^{\mathrm{c}}{\upmu}_{\mathrm{i}\mathrm{j}}=1,1\le \mathrm{j}\le \mathrm{N}, $$
(5)
$$ 0\le \sum \limits_{\mathrm{j}=1}^{\mathrm{N}}{\upmu}_{\mathrm{ij}}<\mathrm{N} $$
(6)

FCM algorithm minimizes the objective function as follows:

$$ {\mathrm{J}}_{\mathrm{m}}\left(\mathrm{X},\mathrm{U},\mathrm{V}\right)=\sum \limits_{\mathrm{j}=1}^{\mathrm{N}}\sum \limits_{\mathrm{i}=1}^{\mathrm{C}}{\left({\upmu}_{\mathrm{i}\mathrm{j}}\right)}^{\mathrm{m}}\bullet {\left\Vert {\mathrm{x}}_{\mathrm{j}}-{\mathrm{v}}_{\mathrm{i}}\right\Vert}^2 $$
(7)
where m is weighted index number.

Cluster centers are computed using the formula:

$$ {v}_i=\sum \limits_{j=1}^N{u}_{ij}^m{x}_j/\sum \limits_{j=1}^N{u}_{ij}^m $$
(8)

The relative membership function of each data towards the centroid is calculated as follows:

$$ {u}_{ij}=\frac{1}{\sum \limits_{k=1}^C{\left(\raisebox{1ex}{${\left\Vert {x}_j-{v}_i\right\Vert}^2$}\!\left/ \!\raisebox{-1ex}{${\left\Vert {x}_j-{v}_k\right\Vert}^2$}\right.\right)}^{\frac{1}{m-1}}} $$
(9)

The following sections are structured in the following main steps: (1) collecting data of the ships start from area nearby Mokpo (Korea) by using AIS; (2) choosing the suitable data for Fuzzy C-Means clustering; (3) using Fuzzy C-Means clustering to cluster and identify the fishing areas; (4) giving out the solution for solving the navigational collision risk based on the results of fishing area identification.

AIS system is providing a wealth of data that can be used to automatically extract knowledge for situational prediction or anomaly detection. Such knowledge reflects the behavior of a portion of traffic identified by the reporting system requirements. AIS data contains ship information including static/voyage-related information such as Maritime Mobile Service Identify (MMSI) number, name, IMO number, call sign and dynamic information (e.g., position, speed) of the ship. In this research, the static information just is used to discard data having the same identifier (MMSI). Dynamic data is used primarily to distinguish the operating area of the fishing vessels. Figure 4 shows the interested area in where the data was taken. That is the area nearby the Mokpo port in Korea.
Fig. 4

The selected area of implementation. Source: Authors

Tables 4 and 5 show the data collected from AIS system, the information of vessels nearby Mokpo area on May 5th 2018. The real data from AIS system is big data but for testing the navigational collision risk solving system, only the three-minutes data was used in the study. The data includes the identification number, the location and speed of the ship.
Table 4

Information of vessels near Mokpo area at 10:00 am~10:03 am on May 5, 2018

No

Longitude

Latitude

Speed

Ship identification

No

Longitude

Latitude

Speed

Ship identification

1

125.410

34.688

0

31,001,739

31

125.716

34.730

9

34,200,608

2

125.932

34.719

0

31,002,575

32

125.160

34.655

8

34,200,682

3

125.686

34.472

7

31,100,492

33

125.737

34.461

4

34,400,298

4

125.969

34.577

2

32,000,029

34

125.186

34.694

0

35,100,463

5

125.201

34.708

0

32,000,061

35

125.461

34.707

8

35,200,707

6

125.934

34.720

0

32,100,291

36

125.968

34.412

3

36,001,528

7

125.437

34.684

0

32,100,303

37

125.190

34.720

1

36,001,768

8

125.445

34.692

0

32,100,488

38

125.175

34.683

8

36,001,769

9

125.444

34.688

0

32,100,489

39

125.438

34.758

0

36,001,839

10

125.933

34.720

0

32,100,551

40

125.417

34.713

0

36,001,955

11

125.197

34.682

7

32,100,593

41

125.418

34.713

2

36,002,015

12

125.932

34.719

0

32,200,180

42

125.193

34.682

0

36,002,589

13

125.948

34.733

0

32,300,237

43

125.438

34.758

0

36,008,017

14

125.934

34.721

0

32,300,281

44

125.443

34.686

0

36,009,809

15

125.442

34.687

0

32,300,334

45

125.190

34.687

0

36,011,864

16

125.550

34.568

0

32,300,419

46

125.369

34.765

0

36,013,464

17

125.847

34.734

1

32,300,900

47

125.658

34.556

3

36,013,569

18

125.906

34.538

4

32,301,007

48

125.435

34.685

0

36,015,580

19

125.972

34.494

0

32,301,008

49

125.393

34.663

0

36,015,606

20

125.853

34.589

0

32,301,011

50

125.290

34.436

12

36,015,628

21

125.949

34.494

9

32,301,014

51

125.214

34.588

8

36,015,654

22

125.845

34.555

8

33,100,193

52

125.238

34.733

8

36,015,929

23

125.934

34.721

0

33,100,288

53

125.639

34.726

12

36,016,084

24

125.929

34.718

0

33,100,361

54

125.454

34.736

0

36,016,157

25

125.440

34.688

0

33,100,556

55

125.465

34.653

0

36,104,991

26

125.287

34.518

9

33,200,654

56

125.443

34.689

0

36,300,294

27

125.393

34.689

0

33,200,798

57

125.433

34.694

2

36,300,404

28

125.444

34.687

0

33,300,224

58

125.431

34.689

0

36,300,474

29

125.433

34.688

0

33,300,306

59

125.395

34.651

0

36,308,730

30

125.394

34.689

0

34,200,222

     

Source: Korean AIS

Table 5

Information of vessels near Mokpo area at 15:00 pm~15:03 pm on May 5, 2018

No

Longitude

Latitude

Speed

Ship identification

No

Longitude

Latitude

Speed

Ship identification

1

125.410

34.688

0

31,001,739

31

125.716

34.730

3

34,200,608

2

125.932

34.719

0

31,002,575

32

125.161

34.653

0

34,200,682

3

125.686

34.472

7

31,100,492

33

125.737

34.461

4

34,400,298

4

125.969

34.577

2

32,000,029

34

125.186

34.694

0

35,100,463

5

125.201

34.708

0

32,000,061

35

125.461

34.707

8

35,200,707

6

125.934

34.720

0

32,100,291

36

125.968

34.412

3

36,001,528

7

125.437

34.684

0

32,100,303

37

125.190

34.720

1

36,001,768

8

125.445

34.692

0

32,100,488

38

125.175

34.683

8

36,001,769

9

125.444

34.688

0

32,100,489

39

125.438

34.758

0

36,001,839

10

125.933

34.720

0

32,100,551

40

125.417

34.714

1

36,001,955

11

125.198

34.681

6

32,100,593

41

125.418

34.713

2

36,002,015

12

125.932

34.719

0

32,200,180

42

125.193

34.682

0

36,002,589

13

125.948

34.733

0

32,300,237

43

125.438

34.758

0

36,008,017

14

125.934

34.721

0

32,300,281

44

125.443

34.686

0

36,009,809

15

125.442

34.687

0

32,300,334

45

125.190

34.687

0

36,011,864

16

125.550

34.568

0

32,300,419

46

125.369

34.765

0

36,013,464

17

125.847

34.734

1

32,300,900

47

125.658

34.556

3

36,013,569

18

125.906

34.538

4

32,301,007

48

125.435

34.685

0

36,015,580

19

125.972

34.494

0

32,301,008

49

125.393

34.663

0

36,015,606

20

125.853

34.589

0

32,301,011

50

125.290

34.436

12

36,015,628

21

125.949

34.494

9

32,301,014

51

125.214

34.588

8

36,015,654

22

125.845

34.555

8

33,100,193

52

125.238

34.733

8

36,015,929

23

125.934

34.720

1

33,100,288

53

125.639

34.726

12

36,016,084

24

125.929

34.718

0

33,100,361

54

125.454

34.736

0

36,016,157

25

125.440

34.688

0

33,100,556

55

125.465

34.653

0

36,104,991

26

125.287

34.518

9

33,200,654

56

125.443

34.689

0

36,300,294

27

125.393

34.689

0

33,200,798

57

125.431

34.693

6

36,300,404

28

125.444

34.687

0

33,300,224

58

125.431

34.689

0

36,300,474

29

125.433

34.688

0

33,300,306

59

125.395

34.651

0

36,308,730

30

125.394

34.689

0

34,200,222

     

Source: Korean AIS

Then next step, after discarding the similar data which have the same identifier (MMSI), we use the dynamic information to cluster the ships into groups. In order to identify the real fishing operation area but not a normal ship operation, we should consider the speed of the ship. Fabio’s research (2014) showed the behavior of reported trawling, which is a kind of fishing operation. It showed that the speed of a ship in the port is zero, the trawling speed is about 2–4 knot and the steaming speed is no more than 10 knots. Based on the research results of Antonio et al. (2016), the fishing activity has speed between 5 and 10 knots meanwhile the searching activities occur at a slightly higher speed (over 11 knots). Therefore, it can be assumed that vessels with speed above 10 knots are not in the scope of this research. In the next step, we will use Fuzzy C-Means clustering to classify the vessels passing through the test area. The results will be determined based on the dynamic vessel. They are vessels’ coordinates (longitude, latitude) and speed. Since the scope of the study only refers a small sea area, the speed will be the main factor to divide the groups. Tables 6 and 7 show grouping results in 2 different time periods. Based on the results, we can eliminate the group of vessels with speed higher than 10 knots.
Table 6

Clustering results based on speed (2018.05.05 10:00 am-10:03 am)

Cluster

No. of vessels

Longitude

Latitude

Speed

1

39

125.54

34.69

0–1

2

7

125.75

34.54

2–4

3

11

125.43

34.63

7–9

4

2

125.47

34.58

12

Source: Authors

Table 7

Clustering results based on speed (2018.05.05 15:00 pm-15:03 am)

Cluster

No. of vessels

Longitude

Latitude

Speed

1

40

125.53

34.69

0–1

2

7

125.77

34.57

2–4

3

3

125.36

34.66

6–7

4

7

125.44

34.62

8–9

5

2

125.47

34.58

12

Source: Authors

After eliminating the vessels that are not in the scope, we once again proceed to divide the groups by the position of the vessel (vessel’s coordinates). The results in Table 8, Table 9, Figs. 5 and 6 show that in both periods, we can divide the passing vessels into 3 groups based on their longitude and latitude.
Table 8

Clustering results based on vessels’ coordinates (2018.05.05 10:00 am-10:03 am)

Cluster

No. of vessels

Longitude

Latitude

1

12

125.196

34.6845

2

26

125.435

34.6907

3

19

125.898

34.5336

Source: Authors

Table 9

Clustering results based on vessels’ coordinates (2018.05.05 15:00 pm-15:03 am)

Cluster

No. of vessels

Longitude

Latitude

1

13

125.194

34.6851

2

24

125.432

34.6944

3

20

125.924

34.5430

Source: Authors

Fig. 5

New clustering results (10:00 am-10:03 am). Source: Authors

Fig. 6

New clustering results (15:00 pm-15:03 pm). Source: Authors

In fishing vessels operating area, fishery, fishing line, trawl net and other fishing gears limit traffic performance and cause collision accidents between merchant ships and fishing vessels, and also other accidents such as obstructing the route of other ships or netting the propeller. Catastrophic results in terms of safety were caused when the other ship involved was a fishing vessel and it highlights the need for better planning when a bulk carrier is sailing in fishing areas. Thus, the route planning of the merchant ships is usually designed to avoid traditional fishing grounds and areas with dense fishing fleets. In this case, the fishing vessel activities will be analyzed to discover the fishing areas to obtain the input variables of the reasoning engine. Based on the Fuzzy C-Means clustering results, two input variables are chosen to infer vulnerability of fishing vessels operating areas.

$$ \left(\mathrm{DFA},\mathrm{SFA}\right)\to \mathrm{Vulnerability}\ \mathrm{of}\ \mathrm{fishing}\ \mathrm{vessels}\ \mathrm{operating}\ \mathrm{area} $$
(10)
where DFA is distance from the own ship to the center of Fishing Area and SFA is size of fishing area.

The DFA is calculated by ‘haversine’ formula as below. That formula is used to calculate the great-circle distance between two points which is the shortest distance over the earth’s surface – ‘as-the-crow-flies’ distance between the points.

$$ a={\mathit{\sin}}^2\left(\frac{\Delta \varphi }{2}\right)+\mathit{\cos}{\varphi}_1.\mathit{\cos}{\varphi}_2.{\mathit{\sin}}^2\left(\frac{\Delta \lambda }{2}\right) $$
(11)
$$ \mathrm{c}=2. atan2\left(\sqrt{a},\sqrt{\left(1-a\right)}\right) $$
(12)
$$ \mathrm{DFA}=\mathrm{R}\ \mathrm{c} $$
(13)
where φ1, φ2 are latitude, λ1, λ2 are longitude, ∆φ = φ1 − φ2, ∆λ = λ1 − λ2, R is earth’s radius (mean radius = 6371 km); c is the angular distance in radians, and a is the square of half the chord length between the points.

SFA usually is almost irregular and cannot be measured exactly. Mathematically, the normal way to calculate the irregular shape is to use calculus. However, considering the discrete nature of the data and the difficulty of implementing calculus in code, the most efficient method to calculate the area of irregular shape is breaking up into small regular shape. The trapezoid method has been successfully used in paper of Zhang and Zhou (2009) to calculate the size of an area. Trapezoid method breaks an irregular shape into many trapezoids whose areas are easier to be calculated. Then adding relevant area of trapezoids and subtracting relevant area of trapezoids to obtain the total area of the irregular shape.

Fuzzy logic membership functions for fishing vessels operating area are noted in Fig. 7 for the distance to the center of fishing area and size of fishing area. Based on the clusters and the survey of experts and navigational officers, 30 miles is deemed to be big value of the size of fishing area and 20 miles is deemed to be big distance to fishing area. Distance of 10 miles and 15 miles size of fishing area are considered as medium values.
Fig. 7

Fuzzy logic membership functions for fishing beats operating area. Source: Authors

According to the membership functions of DFA and SFA, the reasoning rules are described to obtain the vulnerability of fishing area in Table 10 according to experts and navigational officers.
Table 10

Reasoning rules for vulnerability for fishing beats operating area

SFA

DFA

Small

Medium

Big

Small

M

S

VS

Medium

B

M

S

Big

VB

B

M

Source: Authors

Because it is not possible to calculate the navigational collision risk by adding the values of collision risk and vulnerability mathematically, fuzzy logic will be also used to get the navigational collision risk in the following part.

3.3 Navigational collision risk solving system

Combining with the modules one and two, the navigational risk will be calculated using two input variables of collision risk and vulnerability according to the designed fuzzy rules as shown in Table 11.
Table 11

Reasoning rules of navigational collision risk

Collision Risk

Vulnerability

VS

S

M

B

VB

VS

VS

VS

S

M

B

S

S

S

M

B

VB

M

M

M

B

VB

VB

B

B

B

VB

VB

VB

VB

VB

VB

VB

VB

VB

Source: Authors

The reasoning rules for basic CR, vulnerability and navigational CR are shown in Table 9. If vulnerability becomes bigger, the navigational collision risk will increase to a higher level approaching to 1. The membership functions of basic collision risk, vulnerability and consequence are designed in Fig. 8.
Fig. 8

Fuzzy membership functions for module three. Source: Authors

4 Application of navigational collision risk solving system

The proposed algorithm will be tested with simulation to prove its validity. In the simulation, the course and speed of OS are 10° and 14 knots respectively. Four TSs A, B, C, D in the vicinity of own ship in coastal waterway are shown in Table 12 with the information of course, speed, bearing and distance which are used to calculate DCPA and TCPA (Jo et al. 2018).
Table 12

Details of target ships in the vicinity of own ship

Ship

Course (degrees)

Speed (knots)

Bearing (degrees)

Distance (miles)

D CPA(miles)

TCPA (minutes)

A

240

30

050

5.0

0.46

7.38

B

260

10

025

7.5

4.22

9.83

C

150

25

350

6.1

1.43

9.15

D

087

37

280

6.5

0.12

11.5

Source: Authors

The information of the merchant ships and fishing vessels is listed in Table 10 and the results of vulnerability of fishing areas are shown in Table 13 where Vul. is vulnerability. 10 positions of own ship are selected from the situations of Figs. 4 and 5. For instance, the own ship is involved in an encounter with four target ships including merchant ships and fishing vessels in crossing with each other. This proposed scheme for navigational collision risk solving system will be implemented to validate its practicability.
Table 13

Results of vulnerability around fishing areas

Position

1

2

3

4

5

6

7

8

9

10

DFA

22.91

22.91

30.83

14.03

14.03

30.83

12

10

8

6

SFA

7.71

18.62

11.74

16.31

22.24

11.84

22.24

22.24

22.24

22.24

Vul.

0.19

0.32

0.22

0.42

0.52

0.22

0.56

0.62

0.65

0.67

Source: Authors

The vulnerability of fishing area and basic collision risk of four ships can be seen from Table 14. Under the condition of encountering ships in position 9 and 10, the collision risk is detected for a potential collision for ship A as the value is 0.81. So that, ship A is considered as an alert of collision, while the threshold value set as 0.80 is exceeded by the detected value. If the fishing activities are not considered, the collision risk only using DCPA and TCPA is 0.62 which fails to alert for collision risk and may lead to miss the best time to take collision avoidance.
Table 14

Results of the navigational collision risk

TS

Vulnerability

Basic CR

Nav. CR

TS

Vulnerability

Basic CR

Nav. CR

A

0.19

0.62

0.62

A

0.22

0.62

0.62

B

0.07

0.14

B

0.07

0.14

C

0.17

0.21

C

0.17

0.21

D

0.50

0.50

D

0.50

0.50

A

0.32

0.62

0.62

A

0.56

0.62

0.69

B

0.07

0.14

B

0.07

0.28

C

0.17

0.21

C

0.17

0.34

D

0.50

0.50

D

0.50

0.64

A

0.22

0.62

0.62

A

0.62

0.62

0.76

B

0.07

0.14

B

0.07

0.31

C

0.17

0.21

C

0.17

0.39

D

0.50

0.50

D

0.50

0.71

A

0.42

0.62

0.62

A

0.65

0.62

0.81

B

0.07

0.14

B

0.07

0.33

C

0.17

0.21

C

0.17

0.41

D

0.50

0.50

D

0.50

0.75

A

0.52

0.62

0.66

A

0.67

0.62

0.81

B

0.07

0.25

B

0.07

0.38

C

0.17

0.30

C

0.17

0.45

D

0.50

0.59

D

0.50

0.77

Source: Authors

When the ships go through dense fishing vessels, it is dangerous to ignore the influence caused by the fishing operations which may lead to accidents. Compared with conventional collision risk assessment, this algorithm is accurate and reasonable for collision assessment by way of integration. Collision accidents could be effectively prevented if the navigational collision risk solving system is suggested to the officers and the cadets who have insufficient sea experience and navigation competency. The proper collision avoidance actions will be taken in advance when the emergent encountering occurs to both fishing vessels in operations and merchant ships that may cause serious accidents and lead to large loss of life and property.

5 Conclusion

Under the background of SMART-Navigation project, this paper proposed a comprehensive estimation to investigate potentials for navigational collision risk solving system and apply effects to the implementation of e-Navigation solutions to fishing vessels. Based on the previous studies and surveys, the fishing abilities of fishing vessels are universally acknowledged as a significant factor to reduce the collision accidents. A fuzzy methodology for navigational collision risk based on marine accident vulnerability was carried out for encountering ships including fishing vessels. The process of this solving system of navigational risk are listed as below. Firstly, in the designed framework, basic collision risk solving system generally unites DCPA and TCPA to reduce the burden of calculation; secondly, the analysis of fishing vessel vulnerability was investigated by using Fuzzy C-Means; finally, navigational collision risk solving system integrated the vulnerability with basic collision risk to provide navigational collision risk. Simulations were completed for testing the validity of the proposed solving system for this navigational collision risk assessment.

Notes

Acknowledgement

This research is a part of the project titled “SMART-Navigation project,” funded by the Ministry of Oceans and Fisheries, KOREA.

References

  1. Ali F, Kim EK, Kim YG (2015) Type-2 fuzzy ontology-based semantic knowledge for collision avoidance of autonomous underwater vehicles. Inf Sci 295:441–464CrossRefGoogle Scholar
  2. An K (2016) E-navigation services for non-SOLAS ships. International Journal of e-Navigation and Maritime Economy(e-Navi) 4:13–22CrossRefGoogle Scholar
  3. Antonio P, Jon L, Maria S, Hilario M (2016) Validation of VMS data and identification of fishing activities of the Spanish tuna purse seine fleet.  https://doi.org/10.13140/RG.2.2.36696.49922
  4. Berle Ø, Asbjørnslett BE, Rice JB (2011) Formal vulnerability assessment of a maritime transportation system. Reliab Eng Syst Safe 96(6):696–705CrossRefGoogle Scholar
  5. Bukhari AC, Tusseyeva I, Kim YG (2013) An intelligent real-time multi-vessel collision risk assessment system from VTS viewpoint based on fuzzy inference system. Expert Syst Appl 40(4):1220–1230CrossRefGoogle Scholar
  6. Fukuda G, Shoji R (2017) Development of analytical method for finding the high-risk collision areas. TransNav: International Journal on Marine Navigation and Safety of Sea Transportation. 11(3):531–536 CrossRefGoogle Scholar
  7. Hasegawa, K., Kouzuki, A. (1987). Automatic collision avoidance system for ships using fuzzy control. Journal of the Kansai Society of Naval Architects. No.205: 1–10Google Scholar
  8. IMO (2018). e-Navigation, Draft e-Navigation strategy implementation plan. http://www.imo.org/en/OurWork/Safety/Navigation/Pages/eNavigation.aspx. Access date: January 20
  9. Jin D (2014) The determinants of fishing vessel accident severity. Accident Analysis and Prevention. 66:1–7CrossRefGoogle Scholar
  10. Jo GJ, Hu YC, Park GK, Hong T (2018) Building of an algorithm to generate ship’s collision risk based on accident vulnerability under bad weather using fuzzy logic. J Korean Inst Intell Syst 28(4):369–374CrossRefGoogle Scholar
  11. Kim GN (2017) Survey on the Korean VMSs for fishing vessels and challenges for monitoring the high-risk ships. SMART Navigation Project, e-Navigation Underway 2017 Asia-Pacific Conference proceeding, Jeju Island, Korea in 18-20 June.Google Scholar
  12. Kim WS, Lee JH, Kim SJ, Kim HS, Lee YW (2013) A basic study on control factor for the marine casualties of fishing vessel in Korea. Journal of the Korean society of Fisheries Technology 49(1):40–50CrossRefGoogle Scholar
  13. Korean Maritime Safety Tribunal (2016) Investigation Report. http://kmst.go.kr/eng/report/investigationReportList.jsp
  14. Lee HJ, Rhee KP (2001) Development of collision avoidance system by using expert system and search algorithm. International Shipbuilding Progress 48(3):197–212Google Scholar
  15. Lee SM, Kwon KY, Joh J (2004) A fuzzy logic for autonomous navigation of marine vehicles satisfying COLREG guidelines. Int J Control Autom Syst 2(2):171–181Google Scholar
  16. Lee YW, Kim SJ, Park MK (2013) A study on the collision between fishing vessel and non fishing vessel using the analysis of written verdict. Journal of the Korean Society of Fisheries Technology 49(2):136–143CrossRefGoogle Scholar
  17. Li L, Lu W, Niu J, Liu J, Liu D (2018) AIS data-based decision model for navigation risk in sea areas. J Navig 71(3):664–678CrossRefGoogle Scholar
  18. Mazzarella F, Vespe M, Damalas D, Osio G (2014) Discovering vessel activities at sea using AIS data: Mapping of fishing footprints. 17th International Conference on Information Fusion (FUSION). Salamanca: 1–7Google Scholar
  19. Natale F, Gibin M, Alessandrini A, Vespe M, Paulrud A (2005) Mapping fishing effort through AIS data. PLoS One 10(6)CrossRefGoogle Scholar
  20. Negnevitsky M (2005) Artificial Intelligence: A Guide to Intelligent Systems. Second ed. Addison-WesleyGoogle Scholar
  21. Park GK, Benedictos JLRM (2006) Ships collision avoidance support system using fuzzy-CBR. Journal of Fuzzy logic and Intelligent Systems 16(5):635–641CrossRefGoogle Scholar
  22. Rao SN, Balakrishnan A (2017) Harnessing low-cost marine internet for collision avoidance of vessels at sea. In 2017 international conference on wireless communications, signal processing and networking IEEE, pp 2147–2152Google Scholar
  23. Statistical Yearbook of Mimistry of Oceans and Fisheries, 2016. The Korean version can be downloaded from the link: http://www.mof.go.kr/article/view.do?articleKey=13940&boardKey=32&menuKey=396¤tPageNo=1
  24. Weng J, Li G, Chai T, Yang D (2018) Evaluation of two-ship collision severity using ordered Probit approaches. J Navig 71(4):822–836CrossRefGoogle Scholar
  25. Zhang YF, Zhou C (2009) Mobile device applications development for measure area based on android. Computer & Modernization 1(12):143–145Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.School of NavigationShandong Jiaotong UniversityJinanChina
  2. 2.Department of Maritime TransportMokpo National Maritime UniversityMokpoSouth Korea

Personalised recommendations