CN111709571A - A method, device, equipment and storable medium for determining a ship collision avoidance route - Google Patents

Abstract

The invention is applicable to the technical field of ship navigation, and provides a method, device, equipment and storable medium for determining a collision avoidance route for ships, including: when the collision risk between the first ship and the second ship is greater than a preset threshold, Obtain the collision avoidance parameter array set of the first ship; determine the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and the preset dual-standard multi-objective collision avoidance algorithm; determine the optimal collision avoidance parameter array set according to the optimal collision avoidance parameter array set The collision avoidance route of the first ship. The invention adopts the double standard multi-objective genetic algorithm combining Pareto evolution and non-Pareto evolution to solve the collision avoidance route of ships, and by improving the updating method of non-Pareto evolution population, the convergence speed of the population is accelerated and the diversity of the population is increased. , so that the uniformity of the frontier solution set is ensured while speeding up the calculation efficiency of the algorithm; at the same time, through the analysis of the simulation results, it is proved that the present invention can solve a safe and economical path in each meeting scenario.

Description

A method, device, equipment and storable medium for determining a ship collision avoidance route

technical field

The invention belongs to the technical field of ship navigation, and in particular relates to a method, device, equipment and storage medium for determining a ship collision avoidance route.

Background technique

Ship collision avoidance has always been one of the main concerns of ship personnel since the beginning of marine transportation. In the last century, countries and the International Maritime Organization jointly formulated the International Regulations for Preventing Collisions at Sea (COLREGS). The COLREGS rules stipulate the responsibilities of ships to avoid collisions in various encounter situations, the driving and navigation rules of ships under different visibility, and the information on board lights and signs. However, ship collision avoidance is a complicated process, and it is impossible to identify the danger of collision only through COLREGS rules, and at the same time, it is impossible to obtain a safe collision avoidance route.

At present, a large number of algorithms for ship collision avoidance route planning have been proposed. Among them, evolutionary computing is an optimization tool, which is a set of stochastic optimization algorithms, based on Charles Darwin's theory of biological evolution, which means that its basic principle is survival of the fittest. In order to reduce the influence of human factors, Kang et al. used particle swarm algorithm (PSO) to plan the ship path. Xu et al. employed a hazard-immune algorithm to speed up finding the optimal path. Lyu et al. used the good real-time performance of the artificial potential field method to solve the collision avoidance route planning in the case of encountering multiple ships. Most of the collision avoidance studies above assume that the own ship implements a collision avoidance strategy while other target ships keep their course unchanged. In response to this problem, Kim et al. applied the distributed algorithm to collision avoidance research and improved the traditional method. The algorithm uses the communication between ships to obtain the sailing intention of other ships, and determines the navigation of each ship in the current period by comparison. route. However, none of the above methods of ship collision avoidance route planning fully consider the unpredictable behavior of the ship itself, and there are still problems such as the long time for collision avoidance route planning and the poor coordination between the avoidance behaviors of various ships.

SUMMARY OF THE INVENTION

The purpose of the embodiments of the present invention is to provide a method for determining a collision avoidance route for ships, which aims to solve the problem that the existing methods for planning collision avoidance routes for ships do not fully consider the unpredictable characteristics of the behavior of the ship itself, and there is still a long time for collision avoidance route planning. And the problem of poor coordination of avoidance behavior among ships.

The embodiment of the present invention is implemented in this way, a method for determining a ship collision avoidance route, comprising:

When the collision risk between the first ship and the second ship is greater than a preset threshold, acquiring the collision avoidance parameter array set of the first ship;

Determine the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and the preset dual-standard multi-objective collision avoidance algorithm; the preset dual-standard multi-objective collision avoidance algorithm is based on a dual-standard framework combined with fast non-dominant The sorting genetic algorithm and the decomposition-based multi-objective evolutionary algorithm are obtained;

The collision avoidance route of the first ship is determined according to the optimal collision avoidance parameter array set.

Another object of the embodiments of the present invention is a device for determining a ship collision avoidance route, comprising:

a collision avoidance parameter array set acquisition unit, configured to acquire the collision avoidance parameter array set of the first ship when the collision risk between the first ship and the second ship is greater than a preset threshold;

an optimal collision avoidance parameter array set determination unit, configured to determine an optimal collision avoidance parameter array set according to the collision avoidance parameter array set and a preset dual-standard multi-objective collision avoidance algorithm; the preset dual-standard multi-objective collision avoidance algorithm The collision avoidance algorithm is based on a dual-criteria framework combined with a fast non-dominated sorting genetic algorithm and a decomposition-based multi-objective evolutionary algorithm; and

A collision avoidance route determination unit, configured to determine the collision avoidance route of the first ship according to the optimal collision avoidance parameter array set.

Another object of the embodiments of the present invention is a computer device, including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor causes the processor to execute the ship The steps of the collision avoidance route determination method.

Another object of the embodiments of the present invention is a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the processor causes the processor to execute the ship avoidance method. Touch the steps of the route determination method.

In the method for determining a ship collision avoidance route provided by the embodiment of the present invention, the collision avoidance route is quantified by a set of collision avoidance parameter arrays. The collision avoidance algorithm determines the optimal collision avoidance parameter array set, so as to determine the collision avoidance route according to the optimal collision avoidance parameter array set; the present invention takes into account the complexity of the multi-target encounter environment and may cause the final Pareto frontier It presents an irregular shape, and uses a double-standard multi-objective genetic algorithm combining Pareto evolution and non-Pareto evolution to solve the collision avoidance route of ships. By improving the update method of non-Pareto evolution population, the convergence speed of the population is accelerated and the population is increased. The diversity of the algorithm can speed up the calculation efficiency of the algorithm while ensuring the uniformity of the frontier solution set; at the same time, through the analysis of the simulation results, it is proved that the present invention can solve the safe and economical path in each meeting scenario .

Description of drawings

Fig. 1 is the realization flow chart of a kind of ship collision avoidance route determination method provided by the embodiment of the present invention;

Fig. 2 is the realization flow chart of the collision risk determination method provided by the embodiment of the present invention;

3 is a schematic diagram of a four-point ship field provided by an embodiment of the present invention;

4 is a schematic diagram of a speed obstacle method combined with the field of ships provided by an embodiment of the present invention;

5 is a schematic diagram of a congestion degree provided by an embodiment of the present invention;

6 is a schematic diagram of Chebyshev decomposition provided by an embodiment of the present invention;

FIG. 7 is a schematic diagram of a dual standard (BCE) provided by an embodiment of the present invention;

8 is a flowchart of steps for determining an optimal collision avoidance parameter array set provided by an embodiment of the present invention;

9 is a schematic diagram of a collision avoidance route provided by an embodiment of the present invention;

10 is a schematic diagram of the principle of the ant-lion algorithm provided by an embodiment of the present invention;

Fig. 11 is a ZDT1 and ZDT2 operation result diagram provided by an embodiment of the present invention;

FIG. 12 is a flow chart of implementation of another method for determining a ship collision avoidance route provided by an embodiment of the present invention;

13 is a schematic diagram of a step-by-step collaborative collision avoidance strategy provided by an embodiment of the present invention;

FIG. 14 is a flowchart for realizing another method for determining a ship collision avoidance route provided by an embodiment of the present invention;

15 is a GUI interface of a collision avoidance planning simulation design platform provided by an embodiment of the present invention;

16 is a schematic diagram of a collision avoidance route in a situation where two ships cross and meet according to an embodiment of the present invention;

17 is a schematic diagram of a collision avoidance route in a situation where two ships meet and meet according to an embodiment of the present invention;

18 is a schematic diagram of a collision avoidance route in a situation where two ships are chasing and overtaking according to an embodiment of the present invention;

19 is a multi-ship collision avoidance scenario 1 (left) and a collision avoidance route map (right) provided by an embodiment of the present invention;

FIG. 20 is a diagram of the collision avoidance status 1 of a ship at each moment provided by an embodiment of the present invention;

21 is a multi-ship collision avoidance scenario 2 (left) and a collision avoidance route map (right) provided by an embodiment of the present invention;

22 is a diagram of a collision avoidance situation of a ship at each moment provided by an embodiment of the present invention;

23 is a multi-ship collision avoidance scenario (left) and a collision avoidance route map (right) under static obstacles provided by an embodiment of the present invention;

FIG. 24 is another view of the collision avoidance situation of ships at each moment provided by an embodiment of the present invention;

FIG. 25 is a structural block diagram of a device for determining a ship collision avoidance route provided by an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

In the method for determining a ship collision avoidance route provided by the embodiment of the present invention, the collision avoidance route is quantified by a set of collision avoidance parameter arrays. The collision avoidance algorithm determines the optimal collision avoidance parameter array set, so as to determine the collision avoidance route according to the optimal collision avoidance parameter array set; the present invention takes into account the complexity of the multi-target encounter environment and may cause the final Pareto frontier It presents an irregular shape, and uses a double-standard multi-objective genetic algorithm combining Pareto evolution and non-Pareto evolution to solve the collision avoidance route of ships. By improving the update method of non-Pareto evolution population, the convergence speed of the population is accelerated and the population is increased. The diversity of the algorithm can speed up the calculation efficiency of the algorithm while ensuring the uniformity of the frontier solution set; at the same time, through the analysis of the simulation results, it is proved that the present invention can solve the safe and economical path in each meeting scenario .

FIG. 1 is an implementation flowchart of a method for determining a ship collision avoidance route provided by an embodiment of the present invention, which is described in detail as follows.

Step S101, when the collision risk between the first ship and the second ship is greater than a preset threshold, acquire a collision avoidance parameter array set of the first ship.

In this embodiment of the present invention, the collision avoidance parameter array set is generated from a range of values corresponding to a plurality of collision avoidance parameters, the collision avoidance parameter array set includes a plurality of collision avoidance parameter arrays, and the collision avoidance parameter array includes Each collision avoidance parameter is a value randomly determined from the value range corresponding to the collision avoidance parameter; the collision avoidance parameter includes the straight flight time, the avoidance range at the avoidance point, the avoidance time and the return range.

In this embodiment of the present invention, the collision risk between the first ship and the second ship may be determined by the minimum meeting distance DCPA between the first ship (own ship) and the second ship (target ship) and the closest meeting distance The time TCPA of the distance is expressed, when the target ship will pass the own ship's bow, the DCPA is a positive value; otherwise, it is a negative value. When the TCPA is positive, there is still a risk of collision between the own ship and the target ship; when the TCPA is negative, the own ship passes the target ship, and there is no collision risk between the own ship and the target ship.

In a preferred embodiment of the present invention, the collision risk between the first ship and the second ship may also be calculated from the ship domain, the related motion parameters of the first ship and the second ship based on the risk model, as shown in FIG. 2 As shown, before the step S101, it includes:

In step S201, parameters related to the obstacle area of the first ship are acquired.

In the embodiment of the present invention, the parameters related to the obstacle area are obtained to determine the ship domain, which includes the position of the ship, the speed of the ship, the length of the ship, the heading of the ship, the advance distance of the ship's maneuverability, and the initial radius of the turn.

Step S202: Determine the speed obstacle area of the first vessel according to the parameters related to the obstacle area of the first vessel.

In the embodiment of the present invention, the speed obstacle area is the field of ships, and the field of ships is the field of quaternion ships. As shown in FIG. 3 , the position of the ship is (x, y), the speed of the ship is v, and the speed of the ship is v. With length L, this equation of the ship field can be expressed as:

where R _fore , R _aft , R _starb and R _port represent the radius length of the ship field. θ is the heading of the ship.

Among them, A _D and D _T are the advance distance and initial turning radius, which represent the ship's manoeuvrability. Under normal circumstances, the ship will indicate the value of its own advance distance and the initial radius of the turn, but for the target ship it encounters, it may be difficult for the ship to obtain the relevant parameters of its turn test. Therefore, according to the parameters of other ships, the empirical formula is used to calculate the value of the ship's advance distance and the initial radius of the turn:

Step S203, acquiring the speed vector of the first ship and the speed vector of the second ship.

Step S204, according to the speed vector of the first ship, the speed vector of the second ship and the speed obstacle area of the first ship, determine the collision risk between the first ship and the second ship .

In the embodiment of the present invention, by combining the speed obstacle method and the four-point ship field, the speed obstacle method itself not only complies with the COLREGS rules, but also can change the shape according to different ships, as shown in FIG. 4 . The navigation danger of the own ship is obtained by judging whether the speed vector of the own ship is in the speed obstacle area, and the speed vector of the own ship is in the speed obstacle area, which means that the own ship sailing at this speed will intrude into the target ship at some point in the future. The degree of intrusion is related to the degree of the ship's speed within the narrowed ship area of the target ship within the speed barrier. Therefore, the risk degree between ships can be calculated as:

Among them, L ₀ , L ₁ and L ₂ are the distances from the own ship's speed vector to the intersection of the cutting line and the center line passing through the center of the target ship and the intersection of the cutting line and the boundary of the speed barrier. (x ₀ , y ₀ ) is the own ship's position, and (v _x , v _y ) is the own ship's velocity vector. x ₀ ·v _y -y ₀ ·v _x is to judge whether the speed vector of own ship is above or below the center line. Since the centerline represents the target ship's course in geometric space, it means that the closer to the centerline, the greater the danger, and the farther away from the centerline, the lesser the danger. The risk calculated by formula (5) has a value of 1 at the center line and 0.5 at the boundary.

Step S102: Determine an optimal collision avoidance parameter array set according to the collision avoidance parameter array set and a preset dual-standard multi-objective collision avoidance algorithm.

In the embodiment of the present invention, the preset dual-standard multi-objective collision avoidance algorithm is based on a dual-standard framework to perform a fast non-dominated sorting genetic algorithm (NSGA-II) and a decomposition-based multi-objective evolutionary algorithm (MOEA/D). get combined. Among them, an improvement point of the NSGA-II algorithm is that it adopts the method of fast non-dominated sorting to classify the solution set, which speeds up the running speed of the algorithm. The algorithm sets two parameters as non-dominated sorting, namely, the domination count n _p is used to calculate the number of solutions that dominate the solution _p , and the domination set Sp is used to store the solutions dominated by the solution p. And the criterion for judging that one solution dominates the other is that if all the objective values of one solution are better than the objective value of the other solution, then the solution dominates the other solution, and in other cases, the two solutions are considered to be non-dominant of each other or dominated state. According to this criterion, the solutions dominated by the solution p are judged, and these solutions are stored in the S _p set, and the domination count n corresponding to these solutions is increased by one. The dominating set and dominating count value of all solutions are thereby updated, and the solution in which the dominating count value is zero is regarded as the first non-dominant frontier solution, and all the first non-dominant frontier solutions are stored in the set F ₁ . At the same time, the domination counts of the solutions in the dominating set corresponding to the solutions in the F ₁ set are reduced by one, and then the solution with the domination count of zero is determined as the second non-dominated frontier solution and stored in the set F ₂ . Repeating the above content continuously, the process of fast non-dominated sorting is completed. The frontier solution set obtained by non-dominated sorting is the so-called Pareto frontier solution set, and the first Pareto frontier solution set is considered to be the optimal solution set under the current iteration. Another improvement point of the NSGA-II algorithm is the use of crowding degree to ensure the diversity among population members. In order to evaluate the crowding degree, a crowding distance parameter I _distance is defined here. I _distance represents the solution density around a solution. The value of I _distance is calculated by the length of the rectangle formed by the solutions around the solution i, as shown in Figure 5. The calculation formula is as follows:

是第i+1个解的第m个目标值。此外设在边界处的解的拥挤距离是无穷大的。为了克服不同目标函数之间的量级不同，这里对拥挤距离进行了归一化。在种群的迭代过程中，利用拥挤距离以及排序级别对解进行淘汰更新。也就是说，对于不同非支配解集中的解，更优的解是非支配级别低的解，而对于同一非支配级别的解，更优的解是拥挤度大的解。因此，NSGA-II算法能够得到分布均匀且全面的帕累托最优前沿解集。而MOEA/D不需要对种群进行分级，它的总体种群是由从算法开始到现在每个子问题找到的所有最优解组成。该算法有多种将多目标问题分解为多个标量问题的方法，例如，加权求和法，切比雪夫法和边界交叉法。在本发明实施例中主要使用的是切比雪夫法，在图6中展示了该方法的分解原理。切比雪夫分解公式如下：where I[i] _distance is the crowding distance of the i-th solution,

is the mth target value of the i+1th solution. Furthermore, the crowding distance of the solution at the boundary is infinite. To overcome the magnitude difference between different objective functions, the crowding distance is normalized here. In the iterative process of the population, the solution is eliminated and updated by using the crowding distance and the ranking level. That is to say, for solutions in different non-dominated solution sets, the better solution is the solution with low non-dominated level, and for the solutions of the same non-dominated level, the better solution is the solution with high crowding degree. Therefore, the NSGA-II algorithm can obtain a uniform and comprehensive Pareto optimal frontier solution set. While MOEA/D does not need to rank the population, its overall population is composed of all the optimal solutions found for each sub-problem from the beginning of the algorithm to the present. The algorithm has various methods for decomposing a multi-objective problem into multiple scalar problems, for example, the weighted sum method, the Chebyshev method, and the boundary crossing method. In the embodiment of the present invention, the Chebyshev method is mainly used, and the decomposition principle of the method is shown in FIG. 6 . The Chebyshev decomposition formula is as follows:

是一个参考点，f_i(x)是目标函数值，因此对于

总存在一个权重向量λ使得上式的解是一个帕累托最优解x^*，并且该解也是原问题的最优最优解，因此通过改变权重向量λ获取所需的最优解集。这样通过比较新产生的解和原解的g^tc值，对种群进行更新，完成种群向最优帕累托前沿的移动。总的来说，MOEA/D算法中每个子问题仅利用其相邻子问题的信息进行优化，使得MOEA/D每代的计算复杂度低于NSGA-II，也就意味着MOEA/D算法的计算时间比NSGA-II算法更短。where x in g ^tc (x|λ, z ^* ) is the variable to be optimized, λ=(λ ₁ , . . . , λ _m ) is the weight vector,

is a reference point, f _i (x) is the objective function value, so for

There is always a weight vector λ such that the solution of the above formula is a Pareto optimal solution x ^* , and this solution is also the optimal optimal solution of the original problem, so the required optimal solution set is obtained by changing the weight vector λ. In this way, by comparing the g ^tc value of the newly generated solution and the original solution, the population is updated to complete the movement of the population to the optimal Pareto frontier. In general, each sub-problem in the MOEA/D algorithm only uses the information of its adjacent sub-problems for optimization, so that the computational complexity of each generation of MOEA/D is lower than that of NSGA-II, which means that the MOEA/D algorithm has a lower computational complexity than NSGA-II. The computation time is shorter than the NSGA-II algorithm.

In the embodiment of the present invention, considering that the complex environment of multi-ship collision avoidance may cause the final Pareto front to present an irregular shape, the NSGA-II algorithm and the MOEA/D algorithm are combined here through a double-standard algorithm, so that the While speeding up the computational efficiency of the algorithm, it can also ensure the uniformity of the frontier solution set. The specific flow of the algorithm is shown in Figure 7.

In this embodiment of the present invention, as shown in FIG. 8 , the step 102 includes:

Step S801, determining the collision avoidance parameter array set of the first ship as the parent collision avoidance parameter array set.

Step S802, according to the parent generation collision avoidance parameter array set and the fast non-dominated sorting genetic algorithm, generate a child non-dominated collision avoidance parameter array set.

Step S803, according to the parent generation collision avoidance parameter array set and the decomposition-based multi-objective evolutionary algorithm, generate a child generation preferred collision avoidance parameter array set.

Step S804, according to the non-Pareto selection, use the offspring non-dominated collision avoidance parameter array set to make up for the undiscovered optimal solution of the offspring preferred collision avoidance parameter array set, and generate a new generation of collision avoidance parameter array set;

Step S805, it is judged whether the new generation of collision avoidance parameter array set satisfies the preset termination condition; if yes, then go to step S806; if not, go back to step S802.

Step S806, determining the new generation of collision avoidance parameter array set as the optimal collision avoidance parameter array set.

In the embodiment of the present invention, the MOEA/D algorithm is used in the NPC evolution part in FIG. 7 , and the NSGA-II algorithm is used for reference in the PC evolution part. It can be seen that the two parts are basically performed in parallel. Starting from the initialization of the population, the initial population enters into NPC evolution and PC evolution at the same time, and uses the complementary advantages of the two parts to update and iterate the population. The PC selection part realizes the selection of Pareto non-dominated individuals in the mixed set of the population generated by PC evolution and the new individuals generated by NPC evolution. The part of NPC selection is to compare the population evolved by PC with the NPC population to compensate for individuals in the unsearched area of the NPC population and to find individuals with better performance. In this paper, the promising individuals in the PC population adopt the crowding degree criterion to replace the individuals in the NPC population, so as to keep the number of the entire NPC population unchanged. The purpose of population maintenance in PC evolution is to ensure that the number of PC populations remains unchanged. Through the non-dominant level and crowding degree, some poorly performing solutions can be eliminated, while also ensuring the diversity of the entire population. The individual exploration part is the key part of the information exchange between NPC evolution and PC evolution in the whole algorithm. Because the algorithm based on NPC evolution has higher selection pressure than PC evolution, it may sometimes concentrate on part of the Pareto frontier area, resulting in the entire algorithm constantly searching for a specific area repeatedly. The individual exploration part is to try to explore promising individuals in the PC population, these individuals may have been eliminated in the NPC evolution, or have not been explored. Therefore, there is no NPC individual within a certain distance r around these individuals, or there is an NPC individual. The formula for distance r is as follows:

r=(N′/N)·r ₀ (8)

where N is the number of populations before the PC population is maintained, N is the initial size of the population, and _r0 is the niche radius. Since the PC selects non-Pareto individuals, the value of N will be less than the value of N in the early stage of the algorithm, and the value of N will be greater than the value of N in the later stage of the algorithm, but will not exceed twice. This means that in the initial stage of the algorithm, the r value is relatively small, so the PC population will be explored more, and in the later stage of the algorithm, since the NPC population is relatively complete and the dependence on the PC population is small, the r value will be relatively large, reducing the PC population. explore. The value of r ₀ in the present invention is set as the length from the third closest individual to the individual.

In the embodiment of the present invention, considering that most collision avoidance behaviors are only one avoidance operation, four parameters are used to represent the entire collision avoidance route in the present invention, as shown in FIG. 9 . These four parameters represent the timing of collision avoidance and the timing of resumption respectively, and describe the complete collision avoidance process. Among them, T _o represents the direct flight time, that is, the time from the starting point to the avoidance point _{; Ca} represents the avoidance range at the avoidance point; Ta represents the avoidance time, that is, the time from the avoidance point to the return point; C _r represents the return point. range. Considering the speed of the ship, the length of the collision avoidance route and the maneuverability of the ship itself, the present invention sets the range of these four parameters as [0, 60]. Since the present invention adopts the genetic algorithm, the parameters need to be encoded. In general, there are two encoding types, one is binary encoding and the other is real encoding. With reference to the range of parameters, if binary encoding is used, a chromosome may require 24 bits, which makes the algorithm run under a lot of pressure. However, when using real number coding, only 4 parameters are required, and no decoding operation is required at the same time. And in the process of comparing the pros and cons of collision avoidance routes, the computational complexity is reduced. To sum up, the coding mode adopted in the present invention is real number coding.

In the embodiment of the present invention, since the algorithm used is a double-standard multi-objective genetic algorithm, some basic parameters need to be determined first, and these parameters are as follows:

(1) The size of the population N: used to specify the number of individuals in each generation of the population, as the standard for the preservation of the PC evolution part of the population. The larger the population size, the more likely the desired solution will be obtained, but at the same time the algorithm will take longer to run. In this paper, the value of N is set to 50.

(2) Mutation probability P _M and crossover probability P _C : conditions for judging the generation of new individuals in the population. The size of the mutation probability _PM determines the local search ability of the algorithm, and the size of the crossover probability PC determines the global search ability of the algorithm _. In this paper, _PM is set to 0.2 and _PC to be 0.7.

(3) The maximum number of iterations T: used to determine the judgment condition for stopping the algorithm. If the maximum number of iterations is too small, the final solution set may not be optimal, and if it is too large, it will affect the running efficiency of the algorithm. This article has been adjusted through experiments, and the final T is set to 60 generations.

(4) The upper and lower bounds C _u and C _d of the selection solution set: the selection area for selecting the final solution. The Pareto optimal solution set is obtained through algorithm optimization, so by determining the selection area of the final solution on the solution set, the optimal solution is selected by random method. The present invention takes the dangerous target as the criterion for selecting the solution set.

After initializing the above parameters, the population needs to be initialized. Since the present invention uses the parameters of the collision avoidance route as the genetic factor of the algorithm, it is difficult to use experience to generate the initial population, so the method of randomly generating the population within the allowable range of each parameter is adopted here. After the population is initialized, it is necessary to use the algorithm to continuously update the population. The first is the mutation operation, which is an operation that enhances the local search ability of the algorithm. However, mutation operations should generally not be performed too much, as this will lead to a decrease in the efficiency and accuracy of the genetic algorithm. The mutation operation adopted in the present invention is a non-consistent mutation operation, and the mutation method is related to the number of iterations, and the mutation range is large in the early stage of the iteration, and with the iteration of the algorithm, the mutation range gradually decreases, so that the entire population tends to converge. Set the current iteration number as t, the parent individual as A _k , the new individual as A _k1 , and the parameter range as [down, up], the mutation formula is:

The second is the crossover operation, which is also called gene recombination, which enhances the global search ability of the algorithm and prevents the algorithm from falling into a local optimum. The crossover operation in this paper adopts the analog binary operation, and the specific formula is as follows:

Among them, A _k2 and A _k3 are the newly generated individuals of the _crossover operation, and _Am and An are the parent individuals of the crossover operation. And b _q is the set parameter, the formula is as follows:

Although the population can be updated and iterated through the above operations, its convergence speed has always been a problem for multi-objective algorithms. Therefore, the present invention draws on the population update principle of the ant-lion algorithm and introduces it into the new individual generation part of the NPC evolution to speed up the convergence speed of the algorithm. The ant lion algorithm is an algorithm developed with reference to the phenomenon of ant lions preying on ants in nature, as shown in Figure 10. There are two populations in the antlion algorithm, one is the antlion population, which is used to retain the actual individuals in the current iteration, and the other is the ant population, which is used to update the antlion individuals. The ant population is generated based on the antlion population. There is a "trap" around each antlion. Ant individuals are generated by random walk in the "trap" area, and then the newly generated ant individuals are compared with the antlion individuals. Fitness, if the fitness of the individual ant is higher, the ant lion moves to the position of the individual ant, so as to continuously update the ant lion, and finally obtain the optimal solution.

The invention draws on the generation method of the ant population and supplements it into the updating population method of the double-standard multi-objective genetic algorithm. Since the "trap" range around the antlion is continuously reduced with the number of iterations, this speeds up the convergence speed of the algorithm. The random walk formula of ants is as follows:

Ant _i (t)=[0, cumsum(2r(t ₁ )-1), ..., cumsum(2r(t _n )-1)] (12)

The generation formula of ant individuals is as follows:

是随机选择的蚁狮个体所产生的蚂蚁个体位置，

是当前最优蚁狮个体所产生的蚂蚁个体位置，这种操作将全局信息考虑到了个体生成中。in

is the position of the individual ant generated by the randomly selected individual antlion,

is the position of the ant individual generated by the current optimal ant-lion individual. This operation takes the global information into account in the individual generation.

In the embodiment of the present invention, since the multi-objective algorithm adopted in the present invention cannot evaluate a currently optimal individual, the non-dominated solution set of PC evolution is taken as the current optimal antlion population based on the actual conditions of this paper. , by passing the non-dominated solution set to the new individual part of NPC evolution to generate a solution that is closer to the Pareto optimal frontier. The dotted line in Figure 7 conveys the information of the non-dominated solution set. In order to verify the actual effect of this method, the convergence speed of the MOEA/D algorithm with the ant generation method and the original MOEA/D algorithm is compared through the multi-objective test function, as shown in Figure 11. In Figure 11, both ZDT1 and ZDT2 are multi-objective test functions, and the specific formulas are shown in Table 1. In addition, MOEA/D-M refers to an improved algorithm. It can be seen from the figure that in the same 200 generations, the improved MOEA/D algorithm is closer to the expected Pareto optimal frontier than the original algorithm. Therefore, MOEA/D with ant generation rules has better convergence effect and can enhance the convergence speed of bi-standard multi-objective genetic algorithm, which is also the reason why the number of iterations set in this paper is relatively small.

Table 1 Multi-objective test functions

In a preferred embodiment of the present invention, the decomposition-based multi-objective evolutionary algorithm includes a path objective function and a risk objective function.

In the embodiment of the present invention, the objective function refers to a requirement for a ship to select a route in the process of avoiding collision at sea. In this paper, the objectives of ship collision avoidance mainly include the safety and economy of the collision avoidance route. Generally speaking, the safer the collision avoidance route, the less economical the route. The economy is generally expressed by the length of the collision avoidance route. Here, the collision avoidance route refers to the distance from the starting point to the return point, as shown in Figure 9. The specific formula is as follows:

where (x, y) is the position of the ship, and k is the serial number of the ship's position at each moment. The smaller the risk of the ship's navigation, the longer the length of the collision avoidance route. Therefore, in order to form the Pareto condition, the safety of the ship's navigation is represented by the risk of collision avoidance. The risk of the collision avoidance route is calculated by formula (5). The formula of the risk target is as follows:

F _risk = max{Risk _a , Risk _r , Risk _o } (15)

In order to better represent the risk of the collision avoidance route, this paper considers the risk degree of the ship at the avoidance point, the return point and the return point, and takes the maximum value as the risk degree objective function of this paper. In this way, the greater the risk of the ship's collision avoidance navigation, the shorter the route length of the ship's navigation, and the two objective functions are mutually opposed, which constitutes the conditions for the formation of the Pareto front. The criterion for selecting the final solution in this paper is based on the F _risk objective, and the selection range is set to [0.2, 0.5).

Step S103: Determine the collision avoidance route of the first ship according to the optimal collision avoidance parameter array set.

In the method for determining a ship collision avoidance route provided by the embodiment of the present invention, the collision avoidance route is quantified by a set of collision avoidance parameter arrays. The collision avoidance algorithm determines the optimal collision avoidance parameter array set, so as to determine the collision avoidance route according to the optimal collision avoidance parameter array set; the present invention takes into account the complexity of the multi-target encounter environment and may cause the final Pareto frontier It presents an irregular shape, and uses a double-standard multi-objective genetic algorithm combining Pareto evolution and non-Pareto evolution to solve the collision avoidance route of ships. By improving the update method of non-Pareto evolution population, the convergence speed of the population is accelerated and the population is increased. The diversity of the algorithm can speed up the calculation efficiency of the algorithm while ensuring the uniformity of the frontier solution set; at the same time, through the analysis of the simulation results, it is proved that the present invention can solve the safe and economical path in each meeting scenario .

FIG. 12 is an implementation flowchart of another method for determining a ship collision avoidance route provided by an embodiment of the present invention. For convenience of description, only the part related to the embodiment of the present invention is shown, which is similar to the above-mentioned embodiment, and the difference is that , in this embodiment of the present invention, the step 101 includes:

Step 1201, when the collision risk between the first ship, the second ship and the third ship is all greater than a preset threshold, obtain the collision avoidance parameter array set of the first ship, the second ship and the third ship respectively .

In the embodiment of the present invention, the collision risk between the first ship, the second ship and the third ship may be calculated with reference to the above-mentioned method for determining the collision risk between the first ship and the second ship, and the specific risk threshold is It can be set according to the actual situation, and details and limitations are not described here.

The step 102 includes:

Step 1202, according to the collision avoidance parameter array set of the first ship, the second ship and the third ship and the preset dual-standard multi-target collision avoidance algorithm, respectively determine the first ship, the second ship and the third ship The optimal collision avoidance parameter array collection of .

In the embodiment of the present invention, for the method for determining the optimal collision avoidance parameter array set of the first ship, the second ship and the third ship, reference may be made to the above-mentioned method for determining the optimal collision avoidance parameter array set for the first ship. Be specific.

The step 103 includes:

Step 1203: Determine the collision avoidance routes of the first ship, the second ship and the third ship respectively according to the optimal collision avoidance parameter array sets of the first ship, the second ship and the third ship.

Step 1204: Determine a unique avoidance vessel according to the collision avoidance routes of the first vessel, the second vessel and the third vessel and the step-by-step collaborative collision avoidance strategy, so that the avoidance vessel navigates according to the corresponding collision avoidance route.

In the embodiment of the present invention, although the collision avoidance route of the ship can be planned according to the preset dual-standard multi-objective collision avoidance algorithm, the possible collision avoidance behavior of other ships is not considered in such planning. Therefore, the present invention draws on the idea of a distributed local search algorithm, and proposes a step-by-step collaborative collision avoidance strategy, as shown in FIG. 13 . As can be seen in Figure 13, the step-by-step collaborative strategy can be roughly divided into four steps. The first step is the transfer of information through the communication equipment on board. This information includes the ship's speed, heading and collision avoidance intentions. The second step is to conduct collision avoidance planning for each ship based on the information obtained from other ships. In the algorithmic planning process, each ship uses a collision risk model to take into account the impact of other ships on the ship. After this step is completed, each ship has its own collision avoidance intention. In the third step, according to the safety and smoothness of the collision avoidance route, this paper proposes an improvement index ImpRS. Using this index to compare the collision avoidance routes of each ship, it can be analyzed which ship has greater benefits in avoiding collisions. . The formula of the improvement index ImpRS is as follows:

和

分别是船舶原始航行路线以及船舶避让路线与第i艘目标船之间的碰撞危险，该碰撞危险是指在避让点、复航点和返航点三处中的最大危险度，n是目标船舶的数目。如果船舶接近碰撞点，船舶需要尽快进行避让。基于这个情况，本文引入时间因素T，船舶越接近避让点，T的值就越大，从而增大ImpRS的值，提高船舶进行避让的可能性，公式如下：in

and

are the collision risk between the original navigation route of the ship and the avoidance route of the ship and the i-th target ship. The collision risk refers to the maximum risk at the avoidance point, the return point and the return point, and n is the target ship's number. If the ship is approaching the collision point, the ship needs to evade as soon as possible. Based on this situation, the time factor T is introduced in this paper. The closer the ship is to the avoidance point, the larger the value of T will be, thereby increasing the value of ImpRS and improving the possibility of the ship to avoid. The formula is as follows:

T=exp(-α(t _f -Kt)) (17)

where t _f is the time when the ship collides according to the original route, which is set as the minimum value among all the collision times in this paper, t is the time step, and K is the K-th generation cycle. α is a constant. By adjusting the value of α, it is ensured that Time does not affect the value of ImpRS prematurely. In this paper, according to the collision scenario and the maneuverability of the ship itself, α is set to 0.3, that is, for a ship with a speed of 15kn, when t _f is less than 10 minutes, T has a significant impact on ImpRS. Therefore, three factors are considered in formula (16): the danger reduction of the ship's avoidance, the urgency of the need to avoid and the smoothness of the avoidance route. Each ship calculates the value of ImpRS, compares the advantages of each other after avoiding, and selects the largest ImpRS value. The ship is used as the avoidance ship in the fourth step. Except for the avoidance ship, other ships will maintain the original sailing intention to sail in this time step. If there is still a risk of collision between other ships, proceed to the next cycle. In the next cycle, the avoidance ship does not plan the avoidance route, and only needs to provide the avoidance intention as a reference for the avoidance planning of other ships, so as to continue the cycle until all ships can sail safely. The present invention sets the time step t of one cycle to be 3 minutes.

In this embodiment of the present invention, as shown in FIG. 14 , the step S1204 includes:

Step S1401: Determine the improvement index values of the first, second and third ships according to the collision avoidance routes of the first, second and third ships and a preset improvement index function.

Step S1402, according to the improvement degree index values of the first ship, the second ship and the third ship, determine the ship with the largest improvement degree index value as the only avoidance ship.

The method for determining a ship collision avoidance route provided by the embodiment of the present invention combines two algorithms, NSGA-II and MOEA/D, through a dual-standard framework, and complements the advantages of the two types of algorithms, so that the algorithm is more suitable for ship collision avoidance in complex scenarios. Route planning; then quantify the collision avoidance route, and study the basic parameters and basic operations of the dual-standard multi-objective genetic algorithm, and set the algorithm's objective function under the full consideration of the safety and economy of the collision avoidance process; Finally, a step-by-step collaborative collision avoidance strategy is proposed. At the same time, considering the advantages of collision avoidance and the urgency of collision avoidance, an improvement index is proposed, so that the entire collision avoidance planning process not only considers COLREGS rules, but also takes into account other Possible collision avoidance behavior of the vessel.

Simulation and result analysis:

In order to verify the effectiveness of the collision avoidance algorithm, the present invention uses MATLAB software for coding and implementation, and Fig. 15 is the experimental simulation GUI interface based on the ship's intelligent collision avoidance algorithm. The purpose of building the simulation platform is to have a better visualization effect of the algorithm experiment verification results. The GUI interface will be briefly introduced below. As can be seen from Figure 15, the entire GUI interface can be roughly divided into six parts.

The first part is the graphic display part, which mainly displays the ship's navigation environment, the ship's collision avoidance route and the distribution of the solution set in the algorithm iteration process; the second part is the data display area above the graphic display part, where the algorithm is running The navigation data of each ship, such as speed, ship length, heading, and the position of the ship that changes with the iteration time, are displayed at the same time, in addition to the status of the algorithm, such as the current loop algebra, the number of iterations of the algorithm, the avoidance of each ship The collision parameters and the value of the improvement degree of each ship's collision avoidance trajectory, etc.; the third part is the Map information part, where the latitude and longitude range of the ship's navigation area is set, and then the m_map function is used to display the map in the first part; the fourth part is The Algorithm part, this part sets the parameters of the collision avoidance algorithm, including the population size, the maximum number of iterations, the crossover and mutation probability, and the selection range of the final solution; the fifth part is the Import data part, where the basic parameters of all ships are imported as The initial parameters at the beginning of the algorithm; the sixth part is the Mode selection part, which selects two-ship and multi-ship collision avoidance modes. The only difference between the two models is that there is no need for a step-by-step collaborative strategy for collision avoidance planning in the case of two ships. In addition to these six parts, there are also some buttons for displaying the results, such as the route after the ship collision avoidance planning, the distance between the ships and the dynamic collision avoidance navigation display.

Collision avoidance simulation and result analysis of two ships:

The complexity of two-ship collision avoidance is weaker than that of multi-ship collision avoidance, and the choice of collision avoidance behavior itself can be performed according to the COLREGS rules. Therefore, the collision avoidance planning of two ships does not need to consider the collision avoidance behavior of other ships, and the collision avoidance between ships only needs to be planned in accordance with the COLREGS rules. However, once other ships fail to avoid collisions, the ship will perform emergency avoidance behaviors. Here, the simulation analysis is mainly carried out for the three situations encountered by ships, the specific contents are as follows:

(1) Cross encounter situation. Set the own ship's heading as 0°, speed as 15kn, ship length as 250m, target ship's heading as 300°, speed as 7.5kn, and ship length as 250m. The collision avoidance route of the ship after the collision avoidance algorithm planning is shown in Figure 16. The area surrounded by the circle is the four-point ship area, the solid line is the collision avoidance route of the ship, and the dotted line is the navigation route of the target ship and the original navigation route of the ship. The dotted line is the collision avoidance route represented by all the Pareto frontier solutions of the ship after the algorithm planning. It can be seen that the ship complies with the COLREGS rules to avoid the target ship. The ship area of the two ships has not been invaded, and the ship passes around the target ship safely. The ship field parameters of the two ships are shown in Table 2. It can be seen that the speed of the own ship is greater than that of the target ship, so its ship field is larger than that of the target ship. At the same time, the final solution is randomly selected from the Pareto solution set according to the demand range.

Table 2 Ship collision avoidance parameter table in cross situation

(2) Encounter encounter situation. Set the own ship's heading as 0°, speed as 15kn, ship length as 250m, target ship's heading as 172°, speed as 15kn, and ship length as 250m. At this time, the ship should turn to starboard to avoid collision. The specific collision avoidance route planned by the algorithm is shown in Figure 17. In the picture, it can be seen that the ship areas of the two ships were not invaded by the other, and the ship passed the target ship safely. And the ship's collision avoidance behavior complies with the COLREGS rules. Here, the behavior of the target ship is not considered in the collision avoidance process of the own ship, because regardless of the behavior of the target ship, the own ship needs to perform the avoidance operation, so here the target ship does not change its course as the reference for the collision avoidance planning of the own ship. The parameters of ship collision avoidance are shown in Table 3. It can be seen that the size of the ship field of the two ships is the same.

Table 3 Ship collision avoidance parameter table in confrontation situation

(3) Chasing more will meet the situation. Set the own ship's heading as 330°, speed as 15kn, ship length as 250m, target ship's heading as 0°, speed as 7.5kn, and ship length as 250m. At this time, the state of the two ships is that the ship is overtaking the target ship, and the two ships are chasing at a small angle. According to the COLREGS rule, the ship needs to turn to starboard. The specific collision avoidance route planned by the algorithm is shown in Figure 18. As can be seen from the picture, there is no collision between the two ships, and the ship area of the two ships has not been invaded. In addition, the collision avoidance behavior of the ship is also in compliance with the collision avoidance rules. At the beginning, the ship is located at the right rear of the target ship. Since the speed of the ship is higher than that of the target ship, finally the ship passes directly in front of the target ship, and the sailing route meets the requirements of safety and economy. As can be seen in Table 4, the scope of the ship field is wider due to the faster speed of the ship.

Table 4 Ship collision avoidance parameter table in encounter situations

Multi-ship collision avoidance simulation and result analysis:

The present invention designs three different simulation experiments: multi-ship encounter collision avoidance simulation and multi-ship collision avoidance simulation with static obstacles, respectively verifying the validity and reliability of the collision avoidance algorithm for solving route problems in different scenarios, and evaluating the simulation results. The results are analyzed and explained.

The present invention designs two collision avoidance scenarios: a scenario where there is an intersection and a confrontation encounter, and a scenario where there is an intersection and a chasing encounter. Through these two scenarios, it is verified that the multi-ship collision avoidance process complies with the COLREGS rules and the ships Safety of collision avoidance navigation.

(1) Scenario simulation and analysis of crossover and encounter encounter situations

The specific situation of this scenario is shown in Figure 19, and the basic parameters of each ship are shown in Table 5. It can be seen from the right picture of Fig. 19 that ship S ₁ turned to starboard to avoid collision. According to the risk information in Table 5, it can be judged that ship S ₁ is based on the collision avoidance behavior of ship S ₃ , which is in line with COLREGS rule. Similarly, the behavior of other ships also complies with the provisions of the COLREGS Code.

Table 5 Basic parameter information of each ship

Figure 20 lists the sailing states of each ship at each moment, and it can be seen that each ship has its own ship domain around it. At each moment, each ship has not intruded into the ship field of other ships, and each ship is safe to avoid sailing. Table 6 shows the comparison results of the ship improvement function in each generation cycle. It can be seen that in the first cycle, the improvement function of the S ₁ ship is the largest, and its avoidance data is displayed after this row, and then in the second cycle. In the middle is the S ₃ ship, and the last is the S ₂ ship. Due to the evasive actions of the other three ships, the S ₄ ship does not have the danger of collision in the fourth cycle, so the S ₄ ship sails straight, so far all ships are Safe sailing, the entire algorithm operation stops.

Table 6 ImpRS values in each cycle

S₁ S₂ S₃ S₄ T_o C_a T_a C_r first cycle 1.98 1.21 1.58 0.91 0.6 13.4 19.9 57.8 second cycle 0 0.4794 0.6079 0.5899 4 21.5 14.3 53.2 third cycle 0 0.9746 0 0.1939 6.3 30.4 11.2 60.0 fourth cycle 0 0 0 0 - - - -

(2) Scenario simulation and analysis of crossing and overtaking encounter situations

The specific situation of this scenario is shown in Figure 21. The two ships S ₁ and S ₃ are in a state of overtaking, and the other ships are in a state of crossing and meeting. It can be seen from the right picture of Figure 21 that S ₁ and the two ships performed evasive maneuvers, while S ₃ and S ₄ kept the original course and continued to sail. And the avoidance behavior of all ships complies with the provisions of COLREGS rules. The basic parameters of each ship are shown in Table 7. It can be seen that the size of the ship field of ships with different ship speeds is different. _In addition, the most dangerous for the S3 ship is the S1 ship. Due to the avoidance operation _of the S1 ship, the _S3 ship does not have the risk _of collision, so the _S3 ship sails directly, and the same is true for the S4 ship _.

Table 7 Basic parameter information of each ship

The ImpRS values of ships in each cycle stage are listed in Table 8. It can be seen that in the first cycle stage, due to the collision risk between ship S ₁ and ship S ₄ , ship S ₄ carried out collision avoidance route planning , and at the same time, according to the COLREGS rules, it is judged that the S ₃ ship is a direct ship, and no collision avoidance planning is required, so the ImpRS of the S ₃ ship is set to 0. In the second cycle stage, since ship S ₁ was selected as the avoidance ship in the previous cycle, only ship S ₂ performs collision avoidance planning in this cycle, and the planning results are shown in Table 8. Fig. 22 shows the sailing conditions of all ships at various times. It can be seen that at 900s, the S ₄ ships and the S ₃ ships passed each other at the edge of each other's ship field, which met the standard of no intrusion in the ship field. At 1400s, the S ₁ ship surpassed the S ₃ ship and completed the process of overtaking and avoiding collision. During the whole process, all ships sailed safely.

Table 8 ImpRS values in each cycle

S₁ S₂ S₃ S₄ T_o C_a T_a C_r first cycle 0.903 0.385 0 0.8537 3.3 19.3 14.6 45.5 second cycle 0 02985 0 0 9.6 22.2 11.2 57.2 third cycle 0 0 0 0 - - - -

Simulation and analysis of multi-ship encounter situations with static obstacles:

The specific situation of this scene is shown in Figure 23. There is a risk of collision between the three ships, and there are static obstacles such as islands around the ships. It can be seen that when the three ships do not change their course, they will not collide with the surrounding static obstacles, but since the ships need to avoid other ships, the impact of static obstacles on the collision avoidance route planning needs to be considered here. In the right picture of Figure 23, the avoidance path of the ship is shown. It can be seen that under the condition _of complying with COLREGS rules, the ship conducts the avoidance sailing, among which the S1 ship keeps sailing straight, and the _S2 and _S3 ships conduct the avoidance operation. The basic parameters of each ship are listed in Table 9. It can be seen that there is a collision risk between each ship.

Table 9 Basic parameter information of each ship

The ImpRS values for each vessel are listed in Table 10. It can be seen from the table that each ship has carried out collision avoidance planning during the first cycle, among which the S3 ship has the largest _ImpRS value and is used as the avoidance ship in this cycle. In the second cycle, due to the avoidance of the ship, the ship S ₁ does not need to perform the avoidance operation at this time according to the COLREGS rule, so the ship S ₂ is used as the avoidance ship of this cycle in this cycle. From Fig. 24, we can see the navigation status of the ships at each moment. Each ship avoids the ships in conflict with itself, and each ship sails safely.

Table 10 ImpRS values in each cycle

S₁ S₂ S₃ T_o C_a T_a C_r first cycle 0.568 0.642 0.661 0.7 17.3 16.9 58.3 second cycle 0 1.052 0 8.1 24.1 14..2 49.5 third cycle 0 0 0 - - - -

In conclusion, the present invention verifies the proposed dual-standard multi-target collision avoidance algorithm from two aspects of two-ship encounter and multi-vessel encounter, which proves the reliability and effectiveness of the algorithm. Firstly, the MATLAB simulation platform of collision avoidance algorithm is introduced, and the good graphical function of MATLAB software is used to display the effect of the algorithm; then, the simulation test is carried out for the encounter situation between two ships. Whether the planned collision avoidance route complies with the COLREGS rules, it can be seen from the final results that this algorithm can obtain a collision avoidance route that meets the optimization objectives of safety and economy; finally, for the multi-ship encounter situation, three types of collisions are designed scenarios, respectively verify the validity and reliability of the algorithms in the collision avoidance planning between multiple ships and the collision avoidance planning with static obstacles. It can be seen from the final results that the algorithm proposed by the present invention can achieve good results. The results show that the algorithm proposed by the present invention is reasonable and effective.

FIG. 25 is a schematic structural diagram of a device for determining a ship collision avoidance route provided by an embodiment of the present invention. For convenience of description, only parts related to the embodiment of the present invention are shown.

In an embodiment of the present invention, the device for determining a collision avoidance route for a ship includes:

The collision avoidance parameter array set acquisition unit 2501 is configured to acquire the collision avoidance parameter array set of the first ship when the collision risk between the first ship and the second ship is greater than a preset threshold.

In this embodiment of the present invention, the collision avoidance parameter array set is generated from a range of values corresponding to a plurality of collision avoidance parameters, the collision avoidance parameter array set includes a plurality of collision avoidance parameter arrays, and the collision avoidance parameter array includes Each collision avoidance parameter is a value randomly determined from the value range corresponding to the collision avoidance parameter; the collision avoidance parameter includes the straight flight time, the avoidance range at the avoidance point, the avoidance time and the return range.

In this embodiment of the present invention, the collision risk between the first ship and the second ship may be determined by the minimum meeting distance DCPA between the first ship (own ship) and the second ship (target ship) and the closest meeting distance The distance time TCPA is represented; it can be calculated from the ship domain, the relative motion parameters of the first ship and the second ship based on the risk model. For details, see the above description, which will not be repeated here.

The optimal collision avoidance parameter array set determination unit 2502 is configured to determine the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and the preset dual-standard multi-objective collision avoidance algorithm.

In the embodiment of the present invention, the preset dual-standard multi-objective collision avoidance algorithm is based on a dual-standard framework combined with a fast non-dominated sorting genetic algorithm (NSGA-II) and a decomposition-based multi-objective evolutionary algorithm (MOEA/D). owned.

In the embodiment of the present invention, considering that the complex environment of multi-ship collision avoidance may cause the final Pareto front to present an irregular shape, the NSGA-II algorithm and the MOEA/D algorithm are combined here through a double-standard algorithm, so that the While speeding up the calculation efficiency of the algorithm, it can also ensure the uniformity of the frontier solution set.

The collision avoidance route determination unit 2503 is configured to determine the collision avoidance route of the first ship according to the optimal collision avoidance parameter array set.

The device for determining a ship collision avoidance route provided by the embodiment of the present invention quantifies the collision avoidance route with a set of collision avoidance parameter arrays, and obtains a dual-standard multi-objective based on a dual-standard framework combined with a fast non-dominated sorting genetic algorithm and a decomposition-based multi-objective evolutionary algorithm. The collision avoidance algorithm determines the optimal collision avoidance parameter array set, so as to determine the collision avoidance route according to the optimal collision avoidance parameter array set; the present invention takes into account the complexity of the multi-target encounter environment and may cause the final Pareto frontier It presents an irregular shape, and uses a double-standard multi-objective genetic algorithm combining Pareto evolution and non-Pareto evolution to solve the collision avoidance route of ships. By improving the update method of non-Pareto evolution population, the convergence speed of the population is accelerated and the population is increased. The diversity of the algorithm can speed up the calculation efficiency of the algorithm while ensuring the uniformity of the frontier solution set; at the same time, through the analysis of the simulation results, it is proved that the present invention can solve the safe and economical path in each meeting scenario .

In one embodiment, a computer device is proposed, the computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the computer The program implements the following steps:

When the collision risk between the first ship and the second ship is greater than a preset threshold, acquiring the collision avoidance parameter array set of the first ship;

Determine the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and the preset dual-standard multi-objective collision avoidance algorithm; the preset dual-standard multi-objective collision avoidance algorithm is based on a dual-standard framework combined with fast non-dominant It is obtained by sorting genetic algorithm and multi-objective evolutionary algorithm based on decomposition;

The collision avoidance route of the first ship is determined according to the optimal collision avoidance parameter array set.

In one embodiment, a computer-readable storage medium is provided, and a computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the processor performs the following steps:

When the collision risk between the first ship and the second ship is greater than a preset threshold, acquiring the collision avoidance parameter array set of the first ship;

Determine the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and the preset dual-standard multi-objective collision avoidance algorithm; the preset dual-standard multi-objective collision avoidance algorithm is based on a dual-standard framework combined with fast non-dominant It is obtained by sorting genetic algorithm and multi-objective evolutionary algorithm based on decomposition;

The collision avoidance route of the first ship is determined according to the optimal collision avoidance parameter array set.

It should be understood that although the steps in the flowcharts of the embodiments of the present invention are sequentially displayed in accordance with the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed and completed at the same time, but may be executed at different times. The order of execution is also not necessarily sequential, but may be performed alternately or alternately with other steps or sub-steps of other steps or at least a portion of a phase.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium , when the program is executed, it may include the flow of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other medium used in the various embodiments provided in this application may include non-volatile and/or volatile memory. Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

The above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included in the protection of the present invention. within the range.

Claims (10)

1. A ship collision avoidance route determining method is characterized by comprising the following steps:

when the collision risk degree between a first ship and a second ship is larger than a preset threshold value, acquiring a collision avoidance parameter array set of the first ship;

determining an optimal collision avoidance parameter array set according to the collision avoidance parameter array set and a preset dual-standard multi-target collision avoidance algorithm; the preset double-standard multi-target collision avoidance algorithm is obtained by combining a rapid non-dominated sorting genetic algorithm and a multi-target evolutionary algorithm based on decomposition based on a double-standard frame;

and determining the collision avoidance route of the first ship according to the optimal collision avoidance parameter array set.

2. The method for determining the ship collision avoidance line according to claim 1, wherein the step of determining the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and a preset dual-standard multi-target collision avoidance algorithm comprises:

determining the collision avoidance parameter array set of the first ship as a parent collision avoidance parameter array set;

generating a child non-domination collision avoidance parameter array set according to the parent collision avoidance parameter array set and a rapid non-domination sorting genetic algorithm;

generating a child preferred collision avoidance parameter array set according to the parent collision avoidance parameter array set and a multi-objective evolutionary algorithm based on decomposition;

according to non-pareto selection, utilizing the offspring non-domination collision avoidance parameter array set to make up the undetected optimal solution of the offspring preferred collision avoidance parameter array set, and generating a new generation collision avoidance parameter array set;

judging whether the new generation collision avoidance parameter array set meets a preset termination condition or not; if so, determining the new generation of collision avoidance parameter array set as an optimal collision avoidance parameter array set; and if not, determining the new generation of collision avoidance parameter array set as a parent collision avoidance parameter array set, and returning to the step of generating a child non-dominated collision avoidance parameter array set according to the parent collision avoidance parameter array set and the fast non-dominated sorting genetic algorithm.

3. The method for determining the ship collision avoidance line according to claim 1, wherein the step of acquiring the collision avoidance parameter array set of the first ship when the collision risk between the first ship and the second ship is greater than a preset threshold value comprises:

when collision risk degrees among a first ship, a second ship and a third ship are all larger than a preset threshold value, acquiring collision avoidance parameter array sets of the first ship, the second ship and the third ship respectively;

the step of determining the optimal collision avoidance parameter array set according to the collision avoidance parameter array set and a preset dual-standard multi-target collision avoidance algorithm comprises the following steps:

respectively determining the optimal collision avoidance parameter array sets of the first ship, the second ship and the third ship according to the collision avoidance parameter array sets of the first ship, the second ship and the third ship and a preset double-standard multi-target collision avoidance algorithm;

the step of determining the collision avoidance line of the first ship according to the optimal collision avoidance parameter array set comprises the following steps:

determining collision avoidance routes of the first ship, the second ship and the third ship respectively according to the optimal collision avoidance parameter array sets of the first ship, the second ship and the third ship;

and determining the only avoiding ship according to the collision avoidance routes of the first ship, the second ship and the third ship and the step-by-step cooperative collision avoidance strategy so that the avoiding ship navigates according to the corresponding collision avoidance routes.

4. The method for determining the ship collision avoidance line according to claim 3, wherein the step of determining the only ship to avoid according to the collision avoidance lines of the first ship, the second ship and the third ship and the step-by-step cooperative collision avoidance strategy comprises:

determining progress changing index values of the first ship, the second ship and the third ship according to collision avoidance routes of the first ship, the second ship and the third ship and a preset progress changing index function;

and determining the ship with the maximum progress change index value as the only avoiding ship according to the progress change index values of the first ship, the second ship and the third ship.

5. The ship collision avoidance route determining method according to claim 1 or 2, wherein the decomposition-based multi-objective evolutionary algorithm includes a path objective function and a risk degree objective function.

6. The ship collision avoidance route determining method according to claim 1, wherein the collision avoidance parameter array set is generated from a numerical range corresponding to a plurality of collision avoidance parameters, the collision avoidance parameter array set includes a plurality of collision avoidance parameter arrays, and the collision avoidance parameter arrays include numerical values of the collision avoidance parameters randomly determined from the numerical range corresponding to the collision avoidance parameters; the collision avoidance parameters comprise straight voyage time, avoidance amplitude at an avoidance point, avoidance time and re-voyage amplitude.

7. The method for determining a ship collision avoidance line according to claim 1, wherein before the step of obtaining the collision avoidance parameter array set of the first ship when the risk of collision between the first ship and the second ship is greater than a preset threshold, the method further comprises:

acquiring relevant parameters of an obstacle area of a first ship;

determining a speed obstacle area of the first ship according to the obstacle area related parameters of the first ship;

acquiring a speed vector of the first ship and a speed vector of a second ship;

and determining the collision risk degree between the first ship and the second ship according to the speed vector of the first ship, the speed vector of the second ship and the speed obstacle area of the first ship.

8. A collision avoidance line determining apparatus for a ship, comprising:

the collision avoidance parameter array set acquisition unit is used for acquiring a collision avoidance parameter array set of the first ship when the collision risk degree between the first ship and the second ship is greater than a preset threshold value;

the optimal collision avoidance parameter array set determining unit is used for determining an optimal collision avoidance parameter array set according to the collision avoidance parameter array set and a preset dual-standard multi-target collision avoidance algorithm; the preset double-standard multi-target collision avoidance algorithm is obtained by combining a rapid non-dominated sorting genetic algorithm and a multi-target evolutionary algorithm based on decomposition based on a double-standard frame; and

and the collision avoidance route determining unit is used for determining the collision avoidance route of the first ship according to the optimal collision avoidance parameter array set.

9. A computer arrangement, comprising a memory and a processor, the memory having stored therein a computer program which, when executed by the processor, causes the processor to carry out the steps of the ship collision avoidance route determination method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, causes the processor to carry out the steps of the ship collision avoidance route determination method according to any one of claims 1 to 7.

Images