Research ArticleMultiobjective Optimization Based Vessel Collision AvoidanceStrategy Optimization
Qingyang Xu,1 Chuang Zhang,2 and Ning Wang2
1 Shandong University, Weihai, China2Dalian Maritime University, Dalian, China
Correspondence should be addressed to Qingyang Xu; [email protected]
Received 9 November 2013; Revised 23 December 2013; Accepted 25 December 2013; Published 6 February 2014
Academic Editor: Baozhen Yao
Copyright Β© 2014 Qingyang Xu et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The vessel collision accidents cause a great loss of lives and property. In order to reduce the human fault and greatly improvethe safety of marine traffic, collision avoidance strategy optimization is proposed to achieve this. In the paper, a multiobjectiveoptimization algorithm NSGA-II is adopted to search for the optimal collision avoidance strategy considering the safety as well aseconomy elements of collision avoidance. Ship domain and Arena are used to evaluate the collision risk in the simulation. Basedon the optimization, an optimal rudder angle is recommended to navigator for collision avoidance. In the simulation example,a crossing encounter situation is simulated, and the NSGA-II searches for the optimal collision avoidance operation under theConvention on the International Regulations for Preventing Collisions at Sea (COLREGS). The simulation studies exhibit thevalidity of the method.
1. Introduction
With the appearance of larger, specialized, and faster ves-sel, the environment of marine traffic becomes more andmore sophisticated. Therefore, the collision accidents occurfrequently, even though some advanced auxiliary vesselcollision avoidance equipment is widely used aboard. Thevessel collision accidents cause a great loss of lives andproperty. Navigational collision is a major safety concern atsea. According to the investigation of Li et al. [1], 80% ofcollision accidents are caused by human factor, includingwrong decision or careless. Therefore, the subject of how toprovide reasonable navigational information for navigatorshas been studied.
In early navigation, vessel collision avoidance dependedon the experience of navigators due to lack of advancedequipment. Recently, Automatic Radar Plotting Aid (ARPA)appears and is widely installed on most merchant vessels[2]. The data handling and graphic display of equipmentenhance the collision avoidance efficiency, and the decisionis more and more objective [3]. ARPA provides an inter-face for navigators to evaluate the validity of a collision
avoidance strategy by the predicted values of two importantparameters of target vessels-Distance at Closest Point ofApproach (DCPA) and Time to Closest Point of Approach(TCPA) [4]. The navigator can make a decision accordingto result of ARPA. However, the ARPA only can assess themovement of vessels according to a certain speed and course,and it cannot evaluate the economic characteristic of thedifferent operations. Therefore, the navigatorβs decision isalways suboptimal, and sometimes the wrong judgment isrendered.
With the appearance and development of new navigationequipment like AIS [5, 6], advanced computer technology,and so forth, the application of intelligent optimizationalgorithm has been used for collision avoidance strategysearching. Genetic algorithm (GA) is a popular heuristicalgorithm, which has been used for many subjects, such assystem identification [7, 8], supply chain [9], and schedulingproblem [10]. Smierzchalski andMichalewicz [11], Szlapczyn-ski [12], and Szlapczynski and Szlapczynska [13] first madeuse of genetic algorithm to plan the route of vessel in staticor dynamic environment in order to avoid obstacles. Similarheuristic optimization algorithms have been used by other
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014, Article ID 914689, 9 pageshttp://dx.doi.org/10.1155/2014/914689
2 Mathematical Problems in Engineering
researchers: GA is used to find the optimal path andmanoeu-vres in collision avoidance [14β16]. Cheng and Liu madeuse of genetic algorithm to optimize the collision avoidanceroute of urban river [17, 18]. There are also other relatedapproaches used to vessel collision avoidance and routeplanning, such as ant colony algorithm [19β21] and expertsystem [22]. The collision avoidance optimization problemis to find a reasonable way to make the vessel avoid theobstacles with minimal wastage and maximum safety. It is amultiobjective optimization problem. The above-mentionedcollision avoidance problems are generally considered tohave been solved form the scientific point of view, evenif some solutions have not been applied yet. Moreover, inmost of above-mentioned researches, the problems are alwaysdefined as a single-objective problem or transformed to asingle-objective problem by the weights allocation. However,the parameters of weight are always experiential, which aredifficult to set.
Recently, the number of multiobjective evolutionaryalgorithms increases drastically due to their popularity andcapability of successfully solving multiobjective optimizationproblems [23, 24]. In this paper, we adopt a multiobjec-tive optimization algorithm (NSGA-II) instead of a single-objective optimization algorithm to optimize the collisionavoidance strategy. The NSGA proposed by Srinivas andDeb [25] has been successfully applied to solving manyproblems. An improved version of NSGA, which they calledNSGA-II, overcome some disadvantages of NSGA, such ashigh computational complexity of nondominated sorting,lack of elitism, and need a sharing parameter. NSGA-II isconsidered as the state-of-the-artmultiobjective evolutionaryalgorithm [26, 27].The collision avoidance operation of vesselis selected as the optimized variables instead of the path thatis constructed of coordinates set.Thus, the navigators can getthe rudder angle for collision avoidance by the optimizationof NSGA-II.
This paper is organized as follows. The problem ofvessel collision avoidance optimization is presented firstlyin Section 2, and then the multiobjective optimization algo-rithms (NSGA-II) for collision avoidance strategy opti-mization are discussed in Section 3. Finally, a of crossoverencounter situation is simulated, and the optimization pro-cess and results are discussed in Section 4.
2. Vessel Collision AvoidanceOptimization Problems
2.1. Multiobjective Optimization Problem. A multiobjectiveoptimization problem is defined by a set of π· parameters(decision variables), a set of π objective functions, and a setofπ constraints. The objective functions and the constraintsare functions of the decision variables. The aim of theoptimization is to
minimize (andmaximize) π¦ = π (π₯)
= [π
1(π₯) , π
2(π₯) , . . . , π
π]
s.t. π₯ = [π₯
1, π₯
2, . . . , π₯
π, . . . , π₯
π·]
π₯
πminβ€π₯
πβ€π₯
πmax
Γ (π=1, 2, . . . , π·) ,
(1)
where π₯ is the vector of decision variables, f are the objectivefunctions,M is the objective number, and π₯
π min and π₯
π maxare the bound of decision variables.
Considering a minimization problem for each objective,it is said that a decision vector π
πdominates another vector
π
π(ππ> π
π) if and only if
βπ = 1, 2 . . . π, π
π(π
π) β₯ π
π(π
π) or π
π(π
π) > π
π(π
π) .
(2)
We can say that a vector of decision variables π is aPareto-optimal solution or nondominated solution. There-fore, the Pareto-optimal set is the set of all Pareto-optimalsolutions.The aimof an optimization algorithm is to find a setof Pareto-optimal solutions approximating the true Pareto-optimal front.
2.2. Collision Avoidance Strategy Optimization Problem. Thegoal of the collision avoidance strategy optimization is tofind an optimal collision avoidance operation, which has theminimal time loss or way loss spending on maneuvering,while fulfilling some COLREGS rules [28]. Therefore, theindividual evaluation is consolidated of security and eco-nomic factors [29]. The evaluation function includes threeelements. The first part is the security assessment of thestrategy, and it is a very important part of the evaluationfunction, namely, risk of collision.The safety of the strategy ismainly reflected by the collision risk between local vessel andtarget vessel. After generating the initial strategy, the Nomotomodel is used to simulate the vessel and the collision risk isevaluated according to the information of the vessels. Dif-ferent collision avoidance strategies lead to different collisionrisks. The second factor is the economic factors [12], such asthe sailing time and distance.The third one is the smoothnessfactor, namely, the rudder angle changing or course changing.
Therefore, the collision avoidance strategy optimizationbased multiobjective is composed of three objectives asfollows:
minimize π¦ = π (π₯) = [π
1, π
2, π
3] . (3)
The fitness can be calculated by the simulation data,and the primary optimized variable is the rudder angle thatnavigator or autopilot should be adopted in the collisionavoidance.
2.2.1. Safety Evaluation. In the navigation, safety is theprimary problem. In the paper, the minimum distance of thetwo vessels in the collision avoidance is adopted to evaluatethe security. The evaluation function is shown as follows:
π
π
1= safety
π= πΉ βmin (π·
π) , (4)
Mathematical Problems in Engineering 3
where safetyπis the safety evaluation of πth strategy, D is
the distance of passing close, F is a big value to make surethat the safety
πis positive, and function min() is minimum
function. In the process of navigation, the routing is not aline, but a track belt. According to Yangβs literature [30], thedistance between two vessels should bemaintainedmore than(πΏ
0+πΏ
π‘)π/18 to prevent vessel collision caused by suction. If
the distance between the two vessels is less than (πΏ
0+πΏ
π‘)π/18
in the simulation, the two vessels have a collision and thefitness is zero.
COLREGS rules should be considered except for π
π
1,
which is an international rule at sea. Some rules are about thecollision avoidance as follows.
(1) None of the ship domains are violated, that meansevery vessel having its own zone, which cannot beviolated by other vessels.
(2) Speed alterations are not to be applied unless nec-essary (collision cannot be avoided by a configuredmaximum course alteration value). In vessel collisionavoidance, the speed is always constant except thatthe collision avoidance cannot be achieved by courseturning.
(3) COLREGS rules are not violated (especially rules 13to 17).
(a) Rule 13: βan overtaking vessel must keep wellclear of the vessel being overtakenβ in overtak-ing situations.
(b) Rule 14: βwhen two power-driven vessels aremeeting head-on, both must alter course tostarboard, so that they pass on the port side ofthe otherβ in head-on situations.
(c) Rule 15: βwhen two power-driven vessels arecrossing, the vessel, which has the other onthe starboard side, must give wayβ in crossingsituations.
(d) Rule 16: βthe give-way vessel must take early andsubstantial action to keep well clear.β
(e) Rule 17: βthe stand-on vessel may take action toavoid collision if it becomes clear that the give-way vessel is not taking appropriate action.β
According to rules 13β17, the local vessel and targetvessel have different responsibility in different encountersituation. For example, the overtaking vessel cannot affectthe navigation of overtaken vessel in overtaking situation;in head-on situation, the two vessels must adopt collisionavoidance measures separately; crossing situation is a com-plex situation, the given-way vessel must adopt collisionavoidance measures, and the stand-on vessel must stay thespeed and course. The COLREGS rules can be considered asthe constraints of the fitness evaluation. Although there aresome constraints, we have considered it at the generation ofpopulation and do not integrate into the fitness function.
2.2.2. Economy Evaluation. Energy conservation is veryimportant. The second part of evaluation is the economical
evaluation of the strategy. Economy is mainly reflected by thetime and voyage consumption. Since the speed of vessel isconstant in the process of collision avoidance, there is a linearrelation with the distance and time of vessels. Therefore, onlyvoyage consumption is considered in the evaluation function.The evaluation function is as follows:
π
π
2= economy
π=
π
β
π
β(π₯
πβ π₯
πβ1)
2
+ (π¦
πβ π¦
πβ1)
2
, (5)
where economyπis the economical evaluation of πth strategy,
π is the simulation steps, π is the steps length of collisionavoidance, and (π₯, π¦) is the coordinate of vessel.
2.2.3. Smoothness Evaluation. The third part is the smooth-ness evaluation. Excessive smoothness of the routing is notconducive to realize collision avoidance or does not meetthe actual manipulation habit. Conversely, an unduly largeturning angle will cause a longer voyagewith excessive energyand time consumption. In the paper, rudder angle changingis used to evaluate the smoothness of strategy:
π
π
3= smooth
π= ΞπΏ
π, (6)
where smoothπis the smoothness evaluation of πth strategy
and πΏ
πis the rudder angle.
3. Multiobjective Based Collision AvoidanceStrategy Optimization
3.1. The NSGA-II Algorithm. The NSGA-II [31] is one ofthe most famous multiobjective optimization algorithm. TheNSGA [25] first is presented in 1994; then another improvedone NSGA-II was proposed in 2002. According to Section 2,the rudder angle is the primary optimization variables. Theindividual is evaluated by the nondominated sortingmethod.The flow of the algorithm is shown as in Algorithm 1.
NSGA-II is based on Pareto solutions, measuring indi-vidual fitness according to their dominance property. Thenon-dominated individuals in the population are regarded asthe fittest, and the dominated individuals are assigned lowerfitness values, such as the steps (12)β(17) of the Algorithm 1.By this way, the number of dominated individuals will becounted as the fitness values instead of the value of objectivefunction. To maintain the diversity in the Pareto solutions,NSGA-II introduced ameasure of individualβs density respectto other individuals in the objective space, such as the steps(34)β(38) of the Algorithm 1 and had an elitism mechanismand crowed comparison operator to preserve the diversityof population. In step (39) of the Algorithm 1, an arithmeticcrossover and Gaussian mutation operation will be adopted.In the simulation of this paper, the crossover ratio is 1.2, andthe scale and shrink of Gaussian mutation are 0.1 and 0.5.
3.2. The Flow of Collision Avoidance Strategy Optimization.The vessel collision avoidance strategy optimization is acomplex system, which has many procedures. Figure 1 is theflow of collision avoidance strategy optimization. In order toevaluate the fitness of strategies, we adopt a mathematical
4 Mathematical Problems in Engineering
(1) Pop = InitPop(π)(2) InitRank()(3) Pop1 = select(Pop)(4) Popt = crossover(Pop)(5) Pop2 = mutation(Popt)(6) while terminal condition(7) Pop(π) = Pop1(π) βͺ Pop2(π)(8) for π in Pop(π):(9) πππππ = [ ]
(10) np = 0(11) for π in Pop(π)(12) if π dominate π(13) pdomq.add(π)(14) else if π is dominated by π(15) np = np + 1(16) if np is the first rank of Pareto(17) p rank = 1(18) F1.add(π)(19) F.add(F1)(20) while πΉ[π](21) for π in πΉ[π]
(22) for π in pdomq(23) if nq is dominated by other individual(24) q rank = π + 1
(25) Q.add(π)(26) F.add(π)(27) π = π + 1
(28) Pop(π + 1) = []
(29) while len(Pop(π + 1) + len(πΉ[π]) < π
(30) nLen = len(πΉ[π])(31) for π in πΉ[π]
(32) init π.distance = 0(33) for objFun in M objective function(34) πΉ[π] = sort(πΉ[π], objFun)(35) for π in xrange(1, πππ(πΌ) β 2):
(37) πΉ[π][π].distance = πΉ[π][π].distance +(ππππΉπ’π(πΉ[π] [π + 1] β ππππΉπ’π(πΉ[π][π β 1])))
(πππ₯(ππππΉπ’π()) β πππ(ππππΉπ’π()))
(38) Pop1(π + 1) = [Pop(π + 1); πΉ[π]](39) Pop2(π + 1) = generate new pop(40) π+ = π + 1
π = π + 1
Algorithm 1
model of vessel of simulate the vessels. The vessel must assessthe encounter situation all the time and collision avoidancewill be carried out according to the encounter situation.The safe and economic collision avoidance strategy comesfrom numerous collision avoidance strategies that followsthe requirement of International Regulations for PreventingCollisions at Sea (COLREGS) with highest fitness.
Collision risk evaluation is a very important part. Thereare many ways to evaluate the collision risk of vessels,including collision risk models [32] and ship domains [33].In this paper, ship domain and ship Arena which are basedon human praxiology and psychology are selected as thecollision risk evaluation way. Fujii and Yamanouchi [34]proposed the concept of ship domain firstly. The domainis an ellipse, of which the geometrical centre is identical tothe position of ship center, the major semi-axis is along the
fore and aft of ship, and the minor semi-axis isalong thebear abeam of ship. Then, it is introduced to England in1971. Goodwin [35] confirmed the existence of ship domainand established the model of ship domain according to thetraffic investigation of south of the North Sea in open sea.It was derived from statistic methods from large number ofrecord and simulator data. The definition of domain madeby Goodwin [35] is βthe surrounding effective waters thatthe navigator of a ship wants to keep clear of other ships orfixed objects.β The domain is divided into three sectors. Thedomain is shown in Figure 2. Goodwinβs model has shownthat the navigatorβs actions is influenced by the COLREGS.The starboard side is larger than port side, and astern side isthe smallest part. Different sectors of Goodwinβs ship domainis not continual or convenient to carry out traffic simulationon computer. So Davis et al. [36] smoothed Goodwinβs
Mathematical Problems in Engineering 5
No need foravoidance
ARPADCPA, TCPA
calculationAIS
Encounter?YesNo
Course, speedof local ship
Collision risk evaluation
NoDanger?
Yes
Collision avoidanceStrategy optimization
Navigator operation?Decision support system
Automatic navigation
Collision avoidance
Yes
No
Figure 1: The flow of collision avoidance strategy optimization.
0.7 n mile
0β
0.85 n mile
112.5β
247.5β
0.45n mile
Figure 2: Ship domain at open sea.
domain boundary. Hemade use of a circle whose area is equalto the sum area of the three sectors to Goodwinβs domain.Theship shifted to bottom-left corner in the Davisβs ship domainto keep the characteristic of Goodwinβs. It is advantage forcomputer simulation as in Figure 3. The numerical model isshown as follows:
πΉ = π
2+ π
2β π
2,
π = (π₯ β π₯
πΌ) cosπ β (π¦ β π¦
πΌ) sinπ,
π = (π₯ β π₯
πΌ) sinπ β (π¦ β π¦
πΌ) cosπ,
π₯
πΌ= π₯
π + π sin (π + 19
β) ,
Imaginationship
Real ship
Figure 3: Ship domain of Davis.
π¦
πΌ= π¦
π + π cos (π + 19
β) , (7)
where πΉ is the distance to domain. If the ship is outsideof the domain, πΉ < 0. π is the course of ship. π₯
πΌ, π¦
πΌare
the coordination of imagination ship center. π₯π , π¦
π are the
coordination of real ship center. π is the radius of domain. πis the distance from real ship to imagination ship.
The ship Arena is used for navigators to determine thetime of taking collision avoidance actions [37]. If any, weneeded to keep our own ship domain unviolated. It is a biggerarea than ship domain.The parameters are shown in Figure 4.
Arena is a bigger area that navigator can adopt actionor not when the target vessel is in the Arena. If the targetvessel violates our domain in future, the navigator will adopt
6 Mathematical Problems in Engineering
2.7 n mile
Arenacenter
199β
1.7 n mile
Domaincenter
Figure 4: The comparison diagram of ship domain and Arena.
270
247.5
180
90
112.5
Crossing Crossing
Head-on
Overtaking
355 0
Figure 5: Encounter situation.
action to avoid this. In that period and in the followingyears or so, many scholars modified the ship domain andcarried out practical researches. Since then, the ship domainhas been widely used in shipsβ collision avoidance, marinetraffic simulation, calculation of encounter rates, appraisal ofcollision risk, VTS design, and so forth.
4. Simulations
The encounter statuses of vessels are divided into head-on,crossing, and overtaking situations when two vessels havean encounter with a good visibility, the encounter statusesof vessels are divided into head-on, crossing, and overtakingsituations [19]. Figure 5 is the three encounter situations. Inthe collision avoidance decision-making supporting system,we need to judge the encounter situation according tothe status of two vessels, so as to determine the collisionavoidance strategy [38].
2
3
4
5
6
7
β2 β1 0 1 2 3 4
Arena2
AB
Domain2
(0, 2.625)
(0, 3.3958)
Arena1
Domain1
Figure 6: The encounter status of two vessels.
In the simulations, the parameters of target vessel are thatthe vessel moves with a speed π
π‘= 25 kn, the course is πΆ
π‘
= 250β, the starting coordinates is (7, 5.5), and the length ofthe vessel is 110m. The parameters of local vessel are thatthe vessel moves with a speed π
0= 15 kn, the course is πΆ
0
= 0β, the starting coordinates is (0, 0), the length of vessel is250m, and the indexing of πΎπ are πΎ = 0.193, π = 34.119. Inthe simulations, the speed of vessels is constant, and there isno wind and waves. For NSGA-II, the ratio of intermediatecrossover is 1.2 and the scale and shrink of Gaussian mutatinare 0.1 and 0.5.
According to status of two vessels, there is no collisionrisk at the beginning according ship Arena and domain. Thetwo vessels will have a crossing encounter situation, and localvessel is the give-way vessel and the target vessel is stand-on vessel according to the collision liability division. Withthe navigation of the vessels, the target vessel will violate theArena of local vessel, which is shown in Figure 6. The targetvessel violates Arena1 of local vessel at point A but does notviolate the Domain1 of local vessel. Therefore, the local vesseldoes not need to adopt collision operation. However, thetarget violates theDomain2 of local when his position is pointB, and we should adopt collision avoidance operations toavoid danger. Otherwise, the two vessels will have a collisionin future, which is shown in Figure 7.
In order to reduce the evaluation amount of unnecessaryindividuals, the individuals are generated in the feasibleregion according to the COLREGS. In order to facilitatea clear description of the process of collision avoidanceoptimization, the population size of NSGA-II is set to 30. InFigure 8, it is the collision avoidance effect of population ofinfantile iteration. In Figure 9, it is the collision avoidanceeffect of population of later optimization. From Figure 9 wecan know that the individuals are convergent to a certain area.
Mathematical Problems in Engineering 7
2
3
4
5
6
7
β2 β1 0 1 2 3 4
Collision place
(0, 4.438)
Figure 7: The two vessels have a collision.
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
Navigation trajectory of target vessel
Navigation trajectory of local vessel
β2 β1.5 β1 β0.5 0 0.5 1 1.5 2 2.5 3
Figure 8: The collision avoidance effect of population of infantileiteration.
According to the final optimization result, an optimalcollision avoidance strategy is selected when the rudder angleis 11 degrees, according to the fitness priority. The changesof DCPA, TCPA, and RT (distance of vessels) are shown inFigure 10. From the figure, we can see that the distance of twovessels is smaller and smaller with the navigation. If there isnot collision avoidance operation, the two vessels will havecollision before Time 200.The DCPA and TCPA have a jumpfrom Time 200 when the collision avoidance operation isadopted. After collision avoidance, RT becomes bigger.
In Figure 11, it is the comparison diagram of DCPA,TCPA, and RT when rudder angle is 6, 11, and 20 degrees.
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
Navigation trajectory oftarget vessel
Navigation trajectory oflocal vessel
β2 β1.5 β1 β0.5 0 0.5 1 1.5 2 2.5 3
Figure 9: The collision avoidance effect of population of lateroptimization.
The DCPA change of rudder angle = 11 is bigger than rudderangle = 6 and smaller than rudder angle = 20. The minimalRT of rudder angle = 11 is bigger than rudder angle = 6 andsmaller than rudder angle = 20.
From Figure 11, we can see that there is a bigger DCPAfor a longer time when the rudder angle has a value of 11β.At about time 175, the collision avoidance is carried out, thenthe distance of two vessels has an inflection point at time200 and RT becomes bigger subsequently. However, differentoperations have different effects. Although a bigger rudderangle, like 20β, can obtain well safety fitness, other fitnesswill be suboptimal.Therefore, we can get an optimal collisionstrategy (11β of rudder angle) by the optimization of NSGA-II.
5. Conclusions
The environment of maritime traffic becomes more andmore hostile. Therefore, the subject of how to providereasonable collision avoidance information for navigatorsaboard has been studied. In this paper, we make use of theadvanced navigational equipment and multiobjective opti-mization algorithm to obtain an optimal collision avoidancestrategy. The optimization result is a safe and economicaloperation instruction consideringCOLREGS.The simulationresults exhibit the validity of the method. After that theoptimal collision avoidance operation is carried out, and thevessel will be out of danger. Then, the vessel can resumeto the original route. Although the results are promising,we only discussed the optimization mechanisms of thecollision avoidance system. The optimization is based on thenavigational information of local vessel and target vessel.Thecorresponding sensors are required on actual vessel, and therewill be a complex sea conditions in practical.
8 Mathematical Problems in Engineering
1.0
0.5
0.0
β0.5
β1.0
β1.5
β2.0
β2.5
0.3
0.2
0.1
0.0
β0.1
β0.2
10
8
6
4
2
0
β50 0 50 100 150 200 250 300 350 400
Time
DCP
A
TCPA
RT
DCPATCPART
Figure 10: The changes of DCPA, TCPA, and RT when rudder angle is 11 degrees.
10
8
6
4
2
0
0 50 100 150 200 250 300 350
Time
1.0
0.5
0.0
β0.5
β1.0
β1.5
β2.0
β2.5
β3.0
DCP
A
RT
DCPA of rudder angle = 6
DCPA of rudder angle = 11
DCPA of rudder angle = 20
TCAP of rudder angle = 6
TCAP of rudder angle = 11
TCAP of rudder angle = 20
RT of rudder angle = 6
RT of rudder angle = 11
RT of rudder angle = 20
0.4
0.2
0.0
β0.2
TCPA
Figure 11: The changes of DCPA, TCPA, and RT when rudder angle is 6, 11, and 20 degree.
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper.
Acknowledgments
The author thanks the reviewers for their careful reading andhelpful comments that have improved the paper. This workis supported by the National Natural Science Foundationof China (under Grant 51009017), Applied Basic ResearchFunds fromMinistry of Transport of P. R. China (underGrant2012-329-225-060), China Postdoctoral Science Foundation(under Grant 2012M520629), and Fundamental ResearchFunds for the Central Universities of China (under Grants2009QN025, 2011JC002, and 3132013025).
References
[1] L. Li, S. Yang, B. Cao, and Z. Li, βA summary of studies onthe automation of ship collision avoidance intelligence,β Journalof Jimei University (Natural Science), vol. 11, no. 2, pp. 188β192,2006.
[2] T. Statheros, G. Howells, and K. McDonald-Maier,βAutonomous ship collision avoidance navigation concepts,technologies and techniques,β Journal of Navigation, vol. 61, no.1, pp. 129β142, 2008.
[3] X. Zeng,M. Ito, and E. Shimizu, βCollision avoidance ofmovingobstacles for ship with genetic algorithm,β in Proceedings of the6th International Workshop on Advanced Motion Control, pp.513β518, Nagoya, Japan, 2000.
[4] H. C. Chin and A. K. Debnath, βModeling perceived collisionrisk in port water navigation,β Safety Science, vol. 47, no. 10, pp.1410β1416, 2009.
Mathematical Problems in Engineering 9
[5] H.-Z. Hsu, N. A. Witt, J. B. Hooper, and A. P. Mcdermott, βTheAIS-Assisted collision avoidance,β Journal of Navigation, vol. 62,no. 4, pp. 657β670, 2009.
[6] J. M. Mou, C. van der Tak, and H. Ligteringen, βStudy oncollision avoidance in busywaterways by usingAIS data,βOceanEngineering, vol. 37, no. 5-6, pp. 483β490, 2010.
[7] R. D. Al-Dabbagh, A. Kinsheel, M. S. Bin Baba, and S.Mekhilef,βAn integration of compact Genetic algorithm and local searchmethod for optimizing ARMA (1,1) model of likelihood esti-mator,β in Proceedings of the 2nd International Conference onComputer Science and Computational Mathematics, pp. 60β67,Melaka, Malaysia, 2013.
[8] R. D. Al-Dabbagh, M. S. Baba, S. Mekhilef, and A. Kinsheel,βThe compact Genetic Algorithm for likelihood estimator offirst order moving average model,β in Proceedings of the 2ndInternational Conference on Digital Information and Communi-cation Technology and itβs Applications, pp. 474β481, 2012.
[9] B. Yu, Z. Yang, and C. Cheng, βOptimizing the distribution ofshopping centers with parallel genetic algorithm,β EngineeringApplications of Artificial Intelligence, vol. 20, no. 2, pp. 215β223,2007.
[10] B. Yu, Z.-Z. Yang, and B. Yao, βAn improved ant colonyoptimization for vehicle routing problem,β European Journal ofOperational Research, vol. 196, no. 1, pp. 171β176, 2009.
[11] R. Smierzchalski and Z. Michalewicz, βModeling of ship trajec-tory in collision situations by an evolutionary algorithm,β IEEETransactions onEvolutionaryComputation, vol. 4, no. 3, pp. 227β241, 2000.
[12] R. Szlapczynski, βEvolutionary sets of safe ship trajectories: anew approach to collision avoidance,β Journal of Navigation, vol.64, no. 1, pp. 169β181, 2011.
[13] R. Szlapczynski and J. Szlapczynska, βOn evolutionary comput-ing in multi-ship trajectory planning,β Applied Intelligence, vol.37, pp. 155β174, 2012.
[14] M. Ito, F. Zhang, and N. Yoshida, βCollision avoidance controlof ship with genetic algorithm,β in Proceedings of the IEEEInternational Conference on Control Applications and IEEEInternational Symposium on Computer Aided Control SystemDesign, pp. 1791β1796, Kohala Coast, Hawaii, USA, August 1999.
[15] X. Zeng, βEvolution of the safe path for ship navigation,βAppliedArtificial Intelligence, vol. 17, no. 2, pp. 87β104, 2003.
[16] M.-C. Tsou, S.-L. Kao, and C.-M. Su, βDecision support fromgenetic algorithms for ship collision avoidance route planningand alerts,β Journal of Navigation, vol. 63, no. 1, pp. 167β182, 2010.
[17] X. Cheng and Z. Liu, βStudy on ship collision avoidance trajec-tory optimization in inland waterways with genetic algorithm,βNavigation of China, vol. 67, no. 2, pp. 38β46, 2006.
[18] X. Cheng and Z. Liu, βTrajectory optimization for ship collisionavoidance in three fork sea-route of inland waterway,β Ship &Ocean Engineering, vol. 37, no. 4, pp. 74β76, 2008.
[19] M.-C. Tsou and C.-K. Hsueh, βThe study of ship collisionavoidance route planning by ant colony algorithm,β Journal ofMarine Science and Technology, vol. 18, no. 5, pp. 746β756, 2010.
[20] J. B. Escario, J. F. Jimenez, and J.M. Giron-Sierra, βOptimisationof autonomous ship manoeuvres applying Ant Colony Optimi-sation metaheuristic,β Expert Systems with Applications, vol. 39,no. 11, pp. 10120β10139, 2012.
[21] B. Yu and Z. Z. Yang, βAn ant colony optimization model: theperiod vehicle routing problem with time windows,β Trans-portation Research E, vol. 47, no. 2, pp. 166β181, 2011.
[22] T. Tran,Avoidance Navigation Totally Integrated System, Univer-stiy of Southamton, Southampton, UK, 2001.
[23] N. Belgasmi, L. B. Said, and K. Ghedira, βGreedy localimprovement of SPEA2 algorithm to solve the multiobjectivecapacitated transshipment problem,β in Learning and IntelligentOptimization, pp. 364β378, Springer, Berlin, Germany, 2011.
[24] N. Belgasmi, L. B. SaΔ±d, and K. Ghedira, βEvolutionary mul-tiobjective optimization of the multi-location transshipmentproblem,β Operational Research, vol. 8, no. 2, pp. 167β183, 2008.
[25] N. Srinivas and K. Deb, βMuiltiobjective optimization usingnondominated sorting in genetic algorithms,β EvolutionaryComputation, vol. 2, no. 3, pp. 221β248, 1994.
[26] M. R. Rani, H. Selamat, H. Zamzuri, and Z. Ibrahim, βMulti-objective optimization for PID controller tuning using theglobal ranking genetic algorithm,β International Journal ofInnovative Computing, Information and Control, vol. 8, no. 1, pp.269β284, 2012.
[27] S. Bechikh, N. Belgasmi, L. B. Said, and K. Ghedira, βPHC-NSGA-II: a novel multi-objective memetic algorithm for con-tinuous optimization,β in Proceedings of the 20th IEEE Inter-national Conference on Tools with Artificial Intelligence (ICTAIβ08), pp. 180β189, November 2008.
[28] R. Szlapczynski, βEvolutionary sets of safe ship trajectorieswithin traffic separation schemes,β Journal ofNavigation, vol. 66,no. 1, pp. 65β81, 2013.
[29] Y. Yavin, C. Frangos, T. Miloh, and G. Zilman, βCollisionavoidance by a ship with a moving obstacle: computation offeasible command strategies,β Journal of Optimization Theoryand Applications, vol. 93, no. 1, pp. 53β66, 1997.
[30] S. Yang,Decision-Making Support System for Ship Anti-CollisionBased onMulti-Agent, ShanghaiMaritimeUniversity, Shanghai,China, 2008.
[31] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, βA fastand elitist multiobjective genetic algorithm: NSGA-II,β IEEETransactions on Evolutionary Computation, vol. 6, no. 2, pp. 182β197, 2002.
[32] Q. Y. Xu, X. Y. Meng, and N. Wang, βIntelligent evaluationsystem of ship management,β International Journal on MarineNavigation and Safety of Sea Transportation, vol. 4, no. 4, pp.479β482, 2010.
[33] N. Wang, X. Meng, Q. Xu, and Z. Wang, βA unified analyticalframework for ship domains,β Journal of Navigation, vol. 62, no.4, pp. 643β655, 2009.
[34] Y. Fujii and H. Yamanouchi, βThe distribution of collisions inJapan and methods of estimating collision damage,β Journal ofNavigation, vol. 26, no. 1, pp. 108β113, 1973.
[35] E. M. Goodwin, βA statistical study of ship domains,β Journal ofNavigation, vol. 28, no. 3, pp. 328β344, 1975.
[36] P. V. Davis,M. J. Dove, andC. J. Stockel, βA computer simulationof multi-ship encounters,β Journal of Navigation, vol. 35, no. 2,pp. 347β352, 1982.
[37] P.V.Davis,M. J.Dove, andC.T. Stockel, βA computer simulationof marine traffic using domains and arenas,β The Journal ofNavigation, vol. 33, no. 1, pp. 215β222, 1980.
[38] Z. Zheng, Research on Automatic Decision-Making System ofVessel Collision Avoidance, Dalian Maritime University, Dalian,China, 2000.
Submit your manuscripts athttp://www.hindawi.com
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttp://www.hindawi.com
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com
Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Stochastic AnalysisInternational Journal of
Top Related