CN111338350A

CN111338350A - Unmanned ship path planning method and system based on greedy mechanism particle swarm algorithm

Info

Publication number: CN111338350A
Application number: CN202010163351.XA
Authority: CN
Inventors: 辛峻峰; 李世鑫; 李鹏昊; 李书悦; 杨奉儒
Original assignee: Qingdao Lanhai Weilai Marine Technology Co ltd
Current assignee: Qingdao Lanhai Weilai Marine Technology Co ltd
Priority date: 2020-03-10
Filing date: 2020-03-10
Publication date: 2020-06-26
Also published as: LU101833B1

Abstract

The invention belongs to the field of path planning, and provides an unmanned ship path planning method and system based on a greedy mechanism particle swarm algorithm. The unmanned ship path planning method solves the problem of low unmanned ship path planning efficiency, and achieves the effect of high unmanned ship path planning speed. The unmanned ship path planning method comprises the steps of obtaining the current position, course data and the target position of an unmanned ship; calculating an optimal position point of the unmanned ship from the current position to the target position based on a greedy mechanism particle swarm algorithm, correcting the flight path deflection of the unmanned ship course data according to a preset constraint factor, and finally obtaining the optimal path of the unmanned ship.

Description

Unmanned ship path planning method and system based on greedy mechanism particle swarm algorithm

Technical Field

The invention belongs to the field of path planning, and particularly relates to an unmanned ship path planning method and system based on a greedy mechanism particle swarm algorithm.

Background

The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.

Unmanned Ships (USVs), also known as autonomous surface vehicles, have recently attracted considerable worldwide attention in the commercial, scientific and military fields. USVs have the advantages of low operating and maintenance costs, reduced risk of casualties, and good maneuverability and reliability under different operating conditions. With the aid of efficient and reliable navigation devices, such as Global Positioning Systems (GPS), wireless communication units and various types of sensors, USVs can be economically and efficiently used for various applications, including underwater surveying, pollutant tracking, acoustic navigation, rescue at sea, obstacle detection. Despite the significant advances made in the relevant art of USV systems in recent years, it remains a challenge to improve USV autonomy when faced with complex or hazardous environments. Such as ship hull hydrodynamics, communication technology and navigation, guidance and control (NGC) strategies, which require further development. The path planning is an important component of the USV navigation system, and has important significance for designing feasible and optimal tracks of the navigation-based control system and updating information, task requirements and environmental conditions. Its effectiveness not only determines the autonomy of the unmanned ship, but also affects the reliability and efficiency of task execution. The USV path planning problem, often described as a traveler problem (TSP), is a typical combinatorial optimization problem. A specific description is to find a shortest closed loop that does not pass repeatedly through all target cities.

The inventor finds that when the USV executes multi-target tasks in a complex marine environment, the number of possible paths grows exponentially with the increase of the number of target points, so that so-called exponential explosion is caused, and in such a case, the traditional algorithms such as an exhaustion method and a branch definition method cannot find an optimal solution within a reasonable time cost, so that the unmanned ship path planning speed is low and the efficiency is low.

Disclosure of Invention

In order to solve the above problems, a first aspect of the present invention provides an unmanned ship path planning method based on a greedy mechanism particle swarm algorithm, which guarantees that particles continuously move to a higher fitness direction based on a greedy black box, keeps population diversity to a certain extent, and simultaneously effectively retains excellent path segments from old solutions and eliminates path intersection by using 2-opt operation, thereby greatly reducing unmanned ship path planning time and rapidly planning optimal path planning of an unmanned ship.

In order to achieve the purpose, the invention adopts the following technical scheme:

an unmanned ship path planning method based on a greedy mechanism particle swarm algorithm comprises the following steps:

acquiring the current position, course data and a target position of the unmanned ship;

calculating an optimal position point of the unmanned ship from the current position to a target position based on a greedy mechanism particle swarm algorithm, and correcting the flight path deflection of the unmanned ship course data according to a preset constraint factor to finally obtain an optimal path of the unmanned ship;

the process of calculating the optimal position point of the unmanned ship from the current position to the target position based on the greedy mechanism particle swarm optimization is as follows:

establishing a greedy black box, initializing particle swarm parameters based on the position of the unmanned ship and generating an initial particle swarm; the initial particle swarm is composed of a plurality of initial paths of the unmanned ship, and the particles are position points in the paths;

entering an iteration loop, and dividing the particles in the initial particle swarm of each iteration into particle groups with preset quantity;

for each particle group, screening out two particles O (1,2) with the lowest fitness in each particle group, and generating two new particles O' (1,2) by using a greedy black box; the fitness is the reciprocal of the path length;

comparing the fitness of O (1,2) and O' (1,2), reserving two particles with high fitness and updating the particle group;

and performing 2-opt operation on all the updated particle groups, finding the shortest path in each updated particle group, updating the optimal positions of the particles and the optimal positions of the groups, and outputting the optimal positions of the current groups until an iteration loop stop condition is reached.

In order to solve the above problems, a second aspect of the present invention provides an unmanned ship path planning system based on a greedy mechanism particle swarm algorithm, which guarantees that particles move to a higher fitness direction continuously based on a greedy black box, keeps population diversity to a certain extent, and simultaneously effectively retains excellent path segments from old solutions and eliminates path intersection by using 2-opt operation, thereby greatly reducing unmanned ship path planning time and rapidly planning optimal path planning of an unmanned ship.

An unmanned ship path planning system based on a greedy mechanism particle swarm algorithm comprises the following steps:

the data acquisition module is used for acquiring the current position, the course data and the target position of the unmanned ship;

the path planning module is used for calculating an optimal position point of the unmanned ship from the current position to the target position based on a greedy mechanism particle swarm algorithm, correcting the flight path deflection of the unmanned ship course data according to a preset constraint factor, and finally obtaining an optimal path of the unmanned ship;

A third aspect of the invention provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for unmanned ship path planning based on greedy-based particle swarm optimization as described above.

A fourth aspect of the present invention provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for unmanned ship path planning based on greedy-based particle swarm optimization as described above when executing the program.

The invention has the beneficial effects that:

(1) according to the method, the greedy black box is used for initializing the particles, the randomness of the traditional method is overcome, infeasible solutions are eliminated at the beginning of optimization, and the optimization efficiency of the unmanned ship path is improved.

(2) The method utilizes the greedy black box to ensure that the particles continuously move to a higher fitness direction, keeps the diversity of the population to a certain extent, simultaneously can effectively retain excellent coding character string segments from the old solution by 2-opt operation, eliminates path intersection, greatly reduces the unmanned ship path planning time, greatly improves the robustness of the algorithm on the premise of ensuring the solution quality, greatly shortens the optimal path length, and realizes the purpose of rapidly planning the optimal path of the unmanned ship.

(3) The invention also corrects the flight path deflection of the current course data according to the constraint factor, avoids the problem of overlarge deviation between the planned path and the actual path caused by flight path deflection errors, and improves the accuracy of the unmanned ship planned path.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description, serve to explain the invention and not to limit the invention.

FIG. 1 is a flow chart of a method for planning a path of an unmanned ship based on a greedy mechanism particle swarm algorithm according to an embodiment of the invention;

FIG. 2 is an initial travel route of an embodiment of the present invention;

FIG. 3 is a 2-opt operating mechanism of an embodiment of the present invention;

FIG. 4 is a comparison of a Conventional Genetic Algorithm (CGA), an ant colony optimization Algorithm (ACO), a conventional PSO algorithm (CPSO), an improved PSO algorithm (IPSO) according to embodiments of the present invention;

5(a) -5 (h) are graphs of the results of comparing the maximum number of iterations for the Conventional Genetic Algorithm (CGA), the ant colony optimization Algorithm (ACO), the conventional PSO algorithm (CPSO), and the greedy-based particle swarm optimization (IPSO) algorithm of 300, 400, 500, 600, 700, 800, 900, and 1000, respectively, according to an embodiment of the present invention;

FIGS. 6(a) -6 (h) are iteration histories of the optimal path distance (D) and iteration (m) corresponding to FIGS. 5(a) -5 (h), respectively;

fig. 7(a) -7 (h) are the optimal trajectories generated by the ant colony optimization Algorithm (ACO) corresponding to 8 TSPLIB instances according to the embodiment of the present invention;

8(a) -8 (h) are the best traces generated by the greedy-based particle swarm optimization (IPSO) corresponding to 8 TSPLIB instances according to the embodiment of the present invention;

FIG. 9 is a schematic structural diagram of an unmanned ship path planning system based on a greedy mechanism particle swarm optimization according to an embodiment of the invention.

Detailed Description

The invention is further described with reference to the following figures and examples.

It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular is intended to include the plural unless the context clearly dictates otherwise, and it should be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of features, steps, operations, devices, components, and/or combinations thereof.

Example one

The embodiment provides an unmanned ship path planning method based on a greedy mechanism particle swarm algorithm, which comprises the following steps:

the method comprises the following steps: acquiring the current position, course data and a target position of the unmanned ship;

step two: calculating an optimal position point of the unmanned ship from the current position to the target position path based on a greedy mechanism particle swarm algorithm, correcting the flight path deflection of the unmanned ship course data according to a preset constraint factor, and finally obtaining the optimal path of the unmanned ship.

In specific implementation, the process of calculating the optimal position point of the unmanned ship from the current position to the target position based on the greedy mechanism particle swarm algorithm is as follows:

step 1: establishing a greedy black box, initializing particle swarm parameters and generating an initial particle swarm based on the position data of the unmanned ship; the initial particle swarm is composed of a plurality of initial paths of the unmanned ship, and the particles are position points in the paths;

the initialized particle swarm parameters comprise the initialized population quantity, the maximum iteration times, the calculation function of the fitness and the velocity correlation coefficient of the particles. The fitness calculation function is the inverse of the path length.

In this embodiment, the technical difficulty of the greedy mechanism combined with the particle swarm optimization is as follows: when the greedy mechanism generates particles through the particle black box, a first traversal city of the particles to be replaced needs to be determined at first, in the iterative process, the encoding of the particles can generate certain changes when passing through the optimization stage of the particle swarm, then the first traversal city can also change, so that the first traversal city number of the particles to be planned is determined in real time when the greedy mechanism is performed, and new particles are generated through the particle black box.

The principle of greedy algorithms is generally described as making a selection that appears to be locally optimal at present, and works to find a satisfactory solution in a reasonable time. The selection may refer to previous decisions but in no way depends on future selections or other selections inherent to the subproblem. In other words, the greedy algorithm is short-looking, never reconsidering its previous decisions. This is completely different from dynamic planning, which reconsiders previous decisions and ensures that the optimal solution is found after exhaustive operation. Furthermore, greedy algorithms mostly fail to achieve global optimality. However, it is still widely used as an auxiliary strategy or to give the best approximation to some of the problems of time constraints. Meanwhile, the greedy algorithm has a good effect on the optimal substructure problem of which the globally optimal solution comprises the locally optimal solution of the subproblem. When solving the TSP problem, a greedy algorithm is adopted to select the nearest unvisited city as the next target of the salesman until a completely closed path is generated. Because it is usually time consuming to solve the optimal solution of the TSP problem, greedy algorithms are integrated into various algorithms in recent years to accelerate the convergence speed of the algorithms and improve the search efficiency. Pan et al. The hybrid algorithm for deleting the crossover operators is provided, wherein the immune algorithm is adopted for global search, the greedy algorithm is adopted for population initialization, and the path crossover is eliminated. The result shows that the method improves the reliability, the global convergence speed and the searching capability of the immune algorithm. To reduce the calculation time while ensuring the quality of the solution. Basu et al. Before solving the asymmetric TSP problem by tabu search, a random greedy shrinkage method is adopted as a preprocessing step for image sparseness. Compared with other successful heuristic methods, the method has the advantage that the time cost is shortened by 1-5%. Mestria combines Greedy Random Adaptive Search Process (GRASP), iterative local search and variable neighborhood descent, and provides a mixed clustering TSP method. The hybrid heuristic is superior to the existing best methods and methods of medium and large embodiments with reasonable computation time.

Particle Swarm Optimization (PSO) was proposed by Eberhart and Kennedy in 1995, and its sensitivity was derived from bird flight models. In order to avoid the collision phenomenon of a real bird group, the bird group is abstracted into a random particle swarm without mass and volume parameters so as to obtain a global optimal solution. Each particle moves in the search space under dynamic guidance from its own flight state, own experience, and population experience. The evaluation of the solution is performed using a fitness function, and the population is expected to advance toward the most satisfactory solution. In addition, the particle swarm algorithm has the advantages of high convergence speed, simple parameter setting, easy realization and the like, and is widely applied to a large number of discrete and continuous optimization problems such as a path planning problem, a molecular docking problem, electrochemical machining, image recognition and the like, a semi-ideal facility position problem, sheet metal forming and the like. In order to improve the effectiveness of the particle swarm optimization in solving the TSP problem, a great deal of valuable research is carried out on the combination of two or three heuristic algorithms. Shuang et al. The ant colony Algorithm (ACO) is improved, the PSO is introduced to expand the search space, and the convergence is accelerated by utilizing the colony experience. A similar strategy for introducing PSO into the ant colony algorithm was also validated by Mahi et al. To optimize city selection parameters. Another improved particle swarm optimization algorithm proposed by Zhang et al adopts a priority coding method to code solution vectors, dynamically sets a speed range, eliminates the side effect of a discrete search space, and adopts a k-center method to avoid falling into local optimization. This combination algorithm has good performance in preserving group diversity. In the work of Feng et al, a large-scale search space is first partitioned into subspaces using an adaptive fuzzy C-means algorithm. Then, a particle swarm algorithm based on transformation is combined with a simulated annealing algorithm to find a local optimal solution. And finally, reconstructing a complete path by using a max-min merging algorithm. Tests combining particle swarm optimization with genetic algorithms and artificial fish swarm optimization have also yielded satisfactory results.

Particle swarm algorithm based on greedy mechanism:

on one hand, the particle swarm algorithm has inherent limitation, is easy to fall into local optimization, and has poor effect in solving the discrete problem. On the other hand, although the greedy algorithm cannot guarantee the generation of the global optimal solution, the principle is relatively simple, the implementation is easy, and the efficiency is high. Therefore, it seems to be fully feasible to combine these two algorithms to solve various optimization problems based on the idea of complementary advantages.

Conventional particle swarm optimization randomly spreads N particles within an S-dimensional search space at the beginning of the evolution process. For the ith particle, its position and velocity may beRespectively composed of two vectors X_i＝(x_i1,x_i2,..., x_iS)^TAnd V_i＝(v_i1,v_i2,...,v_iS)^TAnd (4) showing. Defining fitness function as 1/D_i(D_iRepresentative of path length), the individual optimum position (P) is updated during each iteration_is) And the population optimal position (P)_gs). At the same time, the velocity and position of each particle is updated to.

Where m denotes the current iteration number and s denotes the s-th dimension, respectively. r is₁And r₂Are iteratively updated random numbers, evenly distributed between 0 and 1. c. C₁And c₂Is an individual cognition factor and a social cognition factor, and is generally represented by 2. w is the inertial weight. Algorithm 1 describes the pseudo code of a conventional particle swarm algorithm.

It should be noted that the PSO control parameters determine the degree of balance between exploration (searching a wider space) and development (moving to local optimality), which has a significant impact on algorithm performance. For example, the acceleration coefficient c₁And c₂Playing an important role in balancing the influence of individual cognition and social cognition on guiding particles towards a target optimal solution. r is₁And r₂The random property of the method can keep the diversity of the population to a certain extent and avoid premature convergence.

It is well known that conventional particle swarm optimization algorithms generate initial particle swarm in a random manner, which may result in some infeasible solutions, thereby limiting convergence speed and search efficiency. Therefore, two local strategies, namely a greedy mechanism and a 2-opt operation, are integrated into the PSO, and the effectiveness of the algorithm is expected to be improved. In our new algorithm, greedy black boxes are used at each iteration of the particle initialization and particle generation stages. However, path crossing is easily generated because the greedy mechanism only considers the current local search case. Therefore, it is necessary to employ 2-opt to increase the probability of satisfactory combination between locally optimal segments and to eliminate path crossing.

The following takes the position points in the path as the city divided by the current administration as an example:

a greedy black box is established based on a greedy mechanism. The method has the function of establishing a local optimal solution after determining a starting city. For TSPs of N cities, C cities closest to the ith city can be found through distance ranging. Thus, a matrix A is defined_N×CThe code whose size is N × C, element Aij representing a city, is the nearest city of the jth city to the ith city A [ i]Is the code for the C city closest to the ith city.

In this work, the value of C is set to 3. Taking the TSPLIB instance burma14 as an example, the mechanism of the greedy black box can be described as follows. If the salesperson starts from a third city, then the code 14 is selected from A [3] as the next city to be visited. Then from A [14], it can be seen that the twelfth city is the closest one and should be selected as the next target city. The next target city is the sixth city, as shown by A12. By analogy, the following cities are sequentially selected to form a tour route. During this process, a repeatability check is made to ensure that each city is visited only once. Typically, three columns may satisfy the requirement of generating a new solution. However, if repetition is unavoidable, then the next target city is randomly selected for repeat checking until a closed circuit is formed. According to this principle, the original travel route may be represented by a series of city codes L ═ {3,14,12, 6,7,13,8,1,11,9,10,2,5,4}, as shown in fig. 2.

Matrix A14 × 3 is as follows:

on one hand, a great amount of infeasible solutions brought by randomness of the traditional method are eliminated by using a greedy black box, and an initial particle swarm with relatively high quality is generated. On the other hand, the greedy black box strategy enables a newly planned path to be composed of a plurality of local optimal segments, and has the advantages of short distance, low order, strong adaptability and the like. Therefore, it is feasible and reasonable to use greedy black boxes for particle initialization.

Step 2: entering an iteration loop, and dividing particles in the initial particle swarm of each iteration into a preset number of particle groups;

and step 3: for each particle group, screening out two particles O (1,2) with the lowest fitness in each particle group, and generating two new particles O' (1,2) by using a greedy black box;

and 4, step 4: comparing the fitness of O (1,2) and O' (1,2), reserving two particles with high fitness and updating the particle group;

and 5: and performing 2-opt operation on all the updated particle groups, calculating the fitness among the particles, updating the optimal positions of the particles and the optimal positions of the groups until an iteration loop stop condition is reached, outputting the optimal position of the current group, correcting the track deflection of the current course data according to the constraint factors, and finally obtaining the optimal path of the unmanned ship.

Local searches are used to improve the quality of the generated solution. In an iterative process, after the velocity and position of the particles are updated, all the particles are scrambled and divided into 4 subgroups. An "unordered" execution will avoid repeated replacement of the same particle as much as possible. At the same time, the grouping operation helps to establish a platform of contact within a subgroup or between any two subgroups. Two particles with lower fitness values are then selected in each subgroup. Their first city was used by a greedy black box to produce two new particles. If the newly generated particles have a higher fitness value, they will replace the old particles; otherwise, no replacement will be done. The operation leads to the elimination of advantages and disadvantages to a certain extent, and prevents the particle swarm from being completely generated by greedy black boxes, which is also called a greedy selection strategy. In addition, since only locally optimal codes, choices, and irreproducibility are considered by greedy black-box generation of particles, certain route crossings may occur, for example, intersections of a sixth city, as shown in fig. 2.

Therefore, a 2-opt operation is employed in each iteration to eliminate the path crossing phenomenon. Our desire is to keep the solution vector with locally optimal segments and to increase the probability of a satisfactory combination between these segments.

Croes first proposed a 2-opt operation in 1958. The mechanism is shown in fig. 3. Two edges are randomly deleted, and the closed path is divided into two parts. The ends of the two parts are then reconnected in another suitable manner, thereby creating a new solution. That is, from the perspective of path coding, two non-adjacent city nodes are randomly selected. The path segments between them will be completely reversed and concatenated back to the original encoded string, resulting in a new solution. This process will repeat until the shortest path is found.

Theoretically, on the one hand, the 2-opt operation ensures that the newly generated solution will retain the excellent encoding string fragment of the old solution. On the other hand, the greedy selection strategy ensures the unidirectionality of the optimized solution vector, namely, the particles continuously move towards a higher fitness direction.

Algorithm 2 gives pseudo code that improves the particle swarm algorithm. Firstly, a greedy black box is established to generate an initial particle swarm, and the initial particle swarm is composed of a plurality of local optimal solutions. During each iteration, the fitness of each particle will be calculated to update the individual and the optimal solution. Then, the velocity and position of each particle are updated according to equations (1) and (2). Furthermore, a greedy selection strategy is applied to the local search along with the 2-opt operation. The algorithm will be terminated when the maximum number of iterations is reached or a sufficiently short path is found.

And converting the constraint factors by weather and sea wave information of the real-time environment where the unmanned ship is located.

Collecting weather and sea wave information of the environment where the unmanned ship is located through an ultrasonic weather sensor, wherein the weather and sea wave information comprises the height of sea waves, the flow velocity of the sea waves and the wavelength of the sea waves; and weather and sea wave information is converted into constraint factors, and the constraint factors are applied to track correction of the unmanned ship.

The constraint factor is mainly sea wave acting force, and because the acting force of the ship in water is mainly the sea wave acting force, the sea wave acting force model is used as the constraint factor to correct track deflection, and the constraint factor has the functions as follows:

wherein h is the height of the sea wave, V_lIs the flow velocity of sea waves, λ₀Is the wavelength of sea waves, M₀Is the mass of the ocean waves.

To ensure that 2-op can completely eliminate path intersections, the number of city points is typically chosen empirically. Meanwhile, a preliminary relationship between the number of city points and the reference maximum number of iterations is obtained through an early test, as shown in fig. 4.

Two existing algorithms, the traditional genetic algorithm (CGA) and the ant colony optimization Algorithm (ACO), are selected to be compared with the traditional PSO (CPSO) and the greedy mechanism-based particle swarm optimization (IPSO). Eight instances from TSPLIB were used: eil51, rat99, kroa100, lin105, ch150, kroa200, tsp225 and lin 318. To eliminate

Randomness of the algorithm in the operating environment, we performed 100 monte carlo simulations to obtain a data set of optimal path distances (D) for each instance and each algorithm. In addition, the control parameters of each algorithm are asTable 1, to facilitate repeatability of this work. It is worth emphasizing that, in order to pursue a balance between complete convergence and economic time consumption, the maximum number of iterations (max) depends on the number of urban points, which is determined by preliminary experiments. For these 8 examples, the maximum values are 300, 400, 500, 600, 700, 800, 900 and 1000, respectively.

Table 1: traditional particle swarm optimization (CPSO), traditional genetic algorithm (CGA), Ant Colony Optimization (ACO), and greedy-based mechanism particle swarm optimization (IPSO) parameterization.

Fig. 5(a) -5 (h) are comparison result graphs of the maximum number of comparison iterations of the Conventional Genetic Algorithm (CGA), the ant colony optimization Algorithm (ACO), the conventional PSO algorithm (CPSO), and the improved PSO algorithm (IPSO) of 300, 400, 500, 600, 700, 800, 900, and 1000, respectively; wherein the comparison results are shown in the form of block diagrams and line graphs. The explanations of Spill were mentioned. In each plot, a box is drawn to represent the quartile range of the data set. The method can reflect the data dispersion degree or the robustness of the algorithm to a certain degree. In addition, a red line and a plus sign are drawn in the bar graph to identify the median and mean values of the data set. There are also two whiskers on either side of the bar, the ends of which represent the optimum and worst values, respectively. The results of ACO and IPSO increase in the upper right corner due to their slight differences. It should be noted that the SD is calculated to display all data points (D)_k) Distance from the mean of the algorithm robustness quantification. In addition, RPE is defined to reflect the gap between the average solution and KOS for TSPLIB. In the four algorithms of these eight examples, the optimal values of AVG are shown in bold and highlighted in gray. SD and RPE are defined as:

in general, CPSO and CGA give a large average and large dispersion difference in the eight cases. Their effectiveness has greatly increased space. In contrast, ACO and IPSO exhibit similar and satisfactory performance, particularly where more urban points are considered. By observing the magnified image, IPSO can find a shorter path, albeit relatively less robust. Using lin105 as an example, CPSO gave an exaggerated solution of mean 60625m and standard deviation 2978m, but with a greedy mechanism and 2-opt IPSO effectively reduced the mean by 74.9% and the standard deviation by 89.3%. For tsp225, the average optimal routing distance for ACO and IPSO is 4300m and 4188m, respectively. The SD value of IPSO is only 0.6% greater than ACO. Furthermore, it was found that in all cases the difference between the mean value of IPSO and the KOS of TSPLIB was less than 10%.

Further, fig. 6(a) -6 (h) are iteration histories of the optimal path distance (D) and the iteration (m) corresponding to fig. 5(a) -5 (h), respectively; in general, the curves for each algorithm have similar trends. As the number of iterations increases, the value of D decreases significantly until a critical number of iterations (Mcri) is reached, and then the solution converges. The present embodiment utilizes the Mcri criterion to evaluate the convergence speed and computational efficiency of each algorithm. The results show that CPSO converges rapidly, terminating at the minimum Mcri. Obviously, this results in a fall into local optimality. The convergence speed of CGA is very slow in all cases compared to the other three algorithms. This behavior helps to find a better solution than CPSO. Meanwhile, IPSO and ACO are co-driving in convergence speed and converged optimal path distance. Taking kroa100 as an example, CGA requires 446 iterations to achieve convergence, which is 7 times that of CPSO. The results show that the algorithm can reduce the average optimal path distance by about 42.9%. In addition, the number of critical iterations for IPSO and CPSO is 113 and 62, respectively. The improvement in Mcri is 82% so that the average value is reduced from 85243.19 to 22981.74m, and it is necessary to slow the convergence rate appropriately in order to obtain a high quality solution.

Since CPSO and CGA will present a chaotic path of 8 TSPLIB instances, fig. 7(a) -7 (h) are the optimal trajectories generated by ant colony optimization Algorithm (ACO) corresponding to the 8 TSPLIB instances of this embodiment; fig. 8(a) -8 (h) show the best trajectories generated by the greedy machine granulation subgroup algorithm (IPSO) for the 8 TSPLIB instances of this example. The paths of the ant colony algorithm have different degrees of intersection, especially when considering more planning points. However, there is no path crossing in the path of IPSO. The application of the 2-opt operation and the greedy selection strategy is beneficial to avoiding path crossing and effectively reducing the complexity of the path. This is also why ACOs get longer paths than IPSOs under the same conditions. The result shows that the method is crucial to improving the quality of the solution, avoiding falling into local optimum and delaying the convergence rate.

Example two

The embodiment provides an unmanned ship path planning system based on a greedy mechanism particle swarm algorithm, which comprises:

(1) the data acquisition module is used for acquiring the current position, the course data and the target position of the unmanned ship;

(2) and the path planning module is used for calculating an optimal position point of the unmanned ship from the current position to the target position based on a greedy mechanism particle swarm algorithm, correcting the flight path deflection of the unmanned ship course data according to a preset constraint factor, and finally obtaining the optimal path of the unmanned ship.

establishing a greedy black box, initializing particle swarm parameters based on the position data of the unmanned ship and generating an initial particle swarm; the initial particle swarm is composed of a plurality of initial paths of the unmanned ship, and the particles are position points in the paths;

and performing 2-opt operation on all the updated particle groups, calculating the fitness among the particles, updating the optimal positions of the particles and the optimal positions of the groups until an iteration loop stop condition is reached, outputting the optimal positions of the current groups, correcting the track deflection of the current course data according to the constraint factors, and finally obtaining the optimal path of the unmanned ship.

In a specific implementation, in the path planning controller, initializing particle swarm parameters including an initialization population number, a maximum iteration number, a calculation function of fitness and a speed correlation coefficient of particles; the fitness calculation function is the inverse of the path length.

In a specific implementation, in the path planning controller, the process of performing the 2-opt operation on all the updated particle groups is as follows:

for each updated set of particles, two non-adjacent location points are randomly selected, and the path segment between them will be completely reversed and concatenated back into the original path, thereby creating a new path, and this process will repeat until the shortest path in each updated set of particles is found.

In specific implementation, the length of the USV model is 1.8 meters, and the width of the USV model is 0.9 meters, and a schematic structural diagram of an unmanned ship path planning system based on a greedy mechanism particle swarm algorithm is given as shown in fig. 9. When the planning point is selected, the path planning controller generates a feasible track for USV tracking by using a greedy mechanism-based particle swarm algorithm. The planned track is transmitted to an automatic driving controller together with navigation information of the direction of the bow of the ship and USV position data acquired by a data acquisition device consisting of a plurality of sensors such as an electronic compass, a GPS and the like, and then the heading and the speed of the vehicle are determined by closed-loop control until the next target point is reached.

Table 3: simulation result of three planning points by ACO and IPSO

When the conventional particle swarm optimization algorithm is applied to the TSP of the USV or the path planning problem, the population is usually initialized in a random manner, which may generate a large number of infeasible solutions and limit the computational efficiency. Thus, the present embodiment introduces a greedy mechanism and a 2-opt operation based on a combination strategy. The method aims to improve the probability of obtaining the optimal solution and improve the quality of the solution while optimizing the traditional algorithm. To verify the effectiveness and reliability of the improved method, Monte Carlo simulations were performed on 8 TSPLIB instances, and application tests were performed on self-developed USVs.

EXAMPLE III

The present embodiment provides a computer-readable storage medium, on which a computer program is stored, which when being executed by a processor, implements the steps of the method for planning an unmanned ship path based on a greedy-based particle swarm algorithm according to the first embodiment.

Example four

The present embodiment provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the program, the processor implements the steps in the method for unmanned ship path planning based on greedy-based particle swarm optimization according to the first embodiment.

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An unmanned ship path planning method based on a greedy mechanism particle swarm algorithm is characterized by comprising the following steps:

2. The unmanned ship path planning method based on greedy mechanism particle swarm optimization algorithm according to claim 1, wherein initializing particle swarm parameters comprises initializing population quantity, maximum iteration times, calculation functions of fitness and velocity correlation coefficients of particles.

3. The unmanned ship path planning method based on the greedy mechanism particle swarm optimization algorithm according to claim 2, wherein the iteration loop stop condition is that a maximum iteration number is reached or the fitness of the current population is minimum.

4. The unmanned ship path planning method based on greedy mechanism particle swarm optimization algorithm according to claim 1, wherein the constraint factor is wave acting force, and the wave acting force is a known function of wave height and wave flow velocity.

5. The method for unmanned ship path planning based on greedy-based particle swarm optimization according to claim 1, wherein the process of performing 2-opt operations on all updated particle groups is as follows:

for each updated set of particles, two non-adjacent location points are randomly selected, the path segment between them will be completely reversed and concatenated back into the original path, resulting in a new path, and the process will repeat until the shortest path in each updated set of particles is found.

6. An unmanned ship path planning system based on a greedy mechanism particle swarm algorithm is characterized by comprising the following steps:

7. The unmanned ship path planning system based on greedy mechanism particle swarm optimization of claim 6, wherein initializing particle swarm parameters in the path planning module comprises initializing population number, maximum iteration number, calculation function of fitness and velocity correlation coefficient of particles.

8. The greedy-based particle swarm algorithm-based unmanned ship path planning system of claim 6,

in the path planning module, the process of performing the 2-opt operation on all the updated particle groups is as follows:

9. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method for unmanned ship path planning based on greedy-based particle swarm optimization according to any of claims 1-5.

10. A computer arrangement comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor when executing the program implements the steps in the method for unmanned ship path planning based on greedy-based particle swarm optimization according to any of claims 1-5.