Multi-water unmanned surface vehicle regional cooperative exploration task allocation method based on genetic algorithm

Abstract

The present application relates to a kind of multi-water surface unmanned vehicle regional cooperative detection task allocation algorithm based on genetic algorithm, it is characterized in that, including the following steps: obtaining detection task area parameter;Obtain the task constraint related to regional detection task;Based on the target function obtained from detection task area parameter and the task constraint related to regional detection task;Using genetic algorithm to solve the target function, final solution is the task allocation result of each water surface unmanned vehicle.The present application is aimed at multi-water surface unmanned vehicle regional detection task, considers the navigation performance and detection ability of different types of water surface unmanned vehicle, uses genetic algorithm, and allocates detection area for each water surface unmanned vehicle, and ensures detection efficiency, so that the whole multi-water surface unmanned vehicle formation detection efficiency is highest.

Description

Technical Field

[0001] This invention relates to a method for allocating collaborative exploration tasks among multiple surface unmanned surface vessels (USVs), and more particularly to a method for allocating collaborative exploration tasks among multiple surface USVs based on a genetic algorithm, which belongs to the field of surface unmanned surface vessel mission planning. Background Technology

[0002] Unmanned surface vessels (USVs) are small, autonomous surface mission platforms primarily used for dangerous tasks or missions unsuitable for manned vessels. With their rapid development, USVs have shown promising prospects in both military and civilian applications. Mission allocation among multiple USVs is a crucial component of USV mission planning and a vital link in the collaborative completion of missions by multiple USV swarms.

[0003] Multi-surface unmanned surface vessel (USV) area reconnaissance missions refer to the use of various USVs carrying detection payloads to detect targets within a given unknown area. How to allocate the detection area to each USV and the number of detection attempts within that area to ensure that the area detection coverage and target detection probability of the multiple USVs are not lower than the mission requirements, while simultaneously ensuring optimal time efficiency, is a crucial issue that needs to be studied in the planning of multi-surface unmanned surface vessel area collaborative reconnaissance missions.

[0004] Current task allocation algorithms for regional detection do not consider the detection blind spots of the detection payload and the detection probability of targets, thus failing to meet the needs of actual mission execution. Furthermore, current regional cooperative search methods all employ balanced coverage search of the task area and assume a 100% detection probability for the task payload. However, in actual mission execution, different types of unmanned surface vessels (USVs) are deployed for cooperative operations, each carrying payloads with varying detection capabilities and target detection probabilities. For example, different sonars have different detection capabilities for underwater targets, and the probability of detecting a target in a single detection is not 100%. Genetic algorithms, a swarm intelligence optimization method, are based on the genetic theory of organisms and the natural selection theory of "survival of the fittest," forming an algorithm for finding optimal solutions. They possess advantages such as good robustness and resistance to getting trapped in local optima. Summary of the Invention

[0005] The technical problem to be solved by this invention is: when allocating tasks for collaborative area detection missions involving multiple surface unmanned surface vessels (USVs), considering constraints such as the detection range, blind zone, and detection probability of the mission payload, to find the optimal task allocation strategy for USVs and improve the efficiency of collaborative area detection by multiple surface unmanned surface vessels.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is to provide a multi-surface unmanned surface vessel area cooperative exploration task allocation algorithm based on genetic algorithm, characterized by including the following steps:

[0007] Step 1: Obtain the parameters of the exploration mission area;

[0008] Step 2: Obtain the relevant task constraints for the region exploration mission;

[0009] Step 3: Obtain the objective function based on the detection task area parameters and the relevant task constraints of the area detection task. The final objective is to find the detection areas of n surface unmanned surface vessels and the number of times to detect the area, so that the n different types of surface unmanned surface vessels can complete the detection of the task area in the shortest time, and the detection coverage and the probability of discovering the target are greater than the preset values ​​in the detection task area parameters.

[0010] Step 4: Use a genetic algorithm to solve the objective function of Step 3. The final solution is the task assignment result for each unmanned surface vessel.

[0011] Preferably, in step 1, the obtained detection mission area parameters include the detection area S, the area detection coverage p, and the expected probability p of the detection target. t .

[0012] Preferably, in step 2, the task constraints related to the regional exploration mission include:

[0013] The number of unmanned surface vessels (USVs) performing the mission, n;

[0014] Maximum operating time t for unmanned surface vessels performing missions max ;

[0015] Maximum detection speed of the unmanned surface vessel performing the mission: V 1max , 2max ,…, nmax ;

[0016] The detection range of the unmanned surface vessel performing the mission: L1, L2, ..., L n ;

[0017] Detection blind spots of unmanned surface vessels performing missions: D1, D2, ..., D n ;

[0018] The probability of an unmanned surface vessel (USV) detecting a target during a mission: p t1 ,p t2 ,…,p tn .

[0019] Preferably, in step S3, the objective function obtained is:

[0020] N1+N2+…+N n ≥S*p

[0021]

[0022]

[0023] Where: N i This represents the size of the detection area of ​​the i-th unmanned surface vessel, where i = 1, 2, ..., n; m di This represents the number of probes required to satisfy the desired probability of detecting the target. This indicates the number of detections required to fill the detection blind spots; m i This indicates that the i-th unmanned surface vessel is detecting the area N. i Total number of detections, m i Round the value up; Represents the combination number, i.e., from m i The number of combinations of selecting one element from a set of elements.

[0024] Preferably, in step 3, the ultimate goal of the objective function is to find the detection areas [N1, N2, ..., N] of n unmanned surface vessels. n and the number of probes to this area [m1,m2,…,m] n The goal is to enable n different types of unmanned surface vessels to complete the detection of the mission area in the shortest time, with a detection coverage rate greater than or equal to p and a target detection probability greater than or equal to p. t .

[0025] Preferably, in step 4, the final solution is the task assignment result for each unmanned surface vessel, i.e., the detection area of ​​each unmanned surface vessel [N1, N2, ..., N]. n and the number of probes to this area [m1,m2,…,m] n ].

[0026] Preferably, step 4 includes the following steps:

[0027] Step 401: Randomly generate the initial population;

[0028] Step 402: Calculate the fitness of individuals in the initial population, and measure the fitness of individuals in the population by the probe time;

[0029] Step 403: Calculate and preserve the best initial population individuals:

[0030] Based on the calculated value of the fitness function, the individual with the largest fitness value is selected as the best individual;

[0031] S404. Propagate and iterate to find the optimal solution.

[0032] Preferably, in step 401, g individuals are randomly generated from the initial population, and the genes in each individual are encoded sequentially, with each individual satisfying:

[0033] N1+N2+…+N n≥S*p

[0034]

[0035]

[0036] V i ≤V imax

[0037]

[0038] In the formula, V i This represents the detection speed of the i-th unmanned surface vessel.

[0039] The parent population is obtained as shown in the following formula:

[0040]

[0041] Where, [N1,N2,…,N n ] and [m1,m2,…,m n ] represents a single chromosome, i.e., an individual, N i and m i It represents a gene on a chromosome.

[0042] Preferably, in step 402, the fitness of the i-th gene is:

[0043] f(i) = 1 / (N i *m i ) / (V i *L i ))

[0044] An individual's fitness is the minimum of the fitness of all genes within that individual, i.e.:

[0045] f=min(f(1),f(2),…,f(n)).

[0046] Preferably, step 404 includes the following steps:

[0047] Step 4041, Select Operation:

[0048] The selection process, based on the principles of betting odds, involves the following steps:

[0049] a) Sum the fitness values ​​of all individuals calculated previously to obtain the total fitness value;

[0050] b) Randomly select a number such that it falls within the range of zero to the total fitness.

[0051] c) Start by sequentially increasing the fitness of the first individual, and continue until you find a fitness value that is higher than the selected value.

[0052] d) Add all individuals at the cutoff position to the breeding population;

[0053] Step 4042, Cross Operation

[0054] The crossover operation is performed using a multi-point crossover method: two individuals are randomly selected for pairing. If the population size is odd, the last remaining individual is not crossovered. A crossover point is randomly selected in the individual's gene coding, and the contents after the crossover point of the paired individuals are exchanged. Then, it is determined whether the crossovered individuals meet the constraints. If not, the crossover contents are mutated until the constraints are met.

[0055] Step 4043, Mutation Operation

[0056] A gene is randomly selected from an individual as the individual's variant gene. For the variant gene, it is replaced with another randomly generated gene. After mutation, it is determined whether the individual meets the constraints. If not, the gene is mutated again until the constraints are met.

[0057] Step 4044: Calculate the fitness of each individual in the offspring: Calculate the fitness of the offspring individuals according to the method in step 402;

[0058] Step 4045: Update the total fitness of the population for selection using roulette wheel selection, where the total fitness of the population is the sum of the fitness values ​​of each individual in the population;

[0059] Step 4046: Preserve the best offspring individual;

[0060] Step 4047: Update the child generation content to the parent generation, and this child generation will become the parent generation of the next generation;

[0061] Step 4048: Update the best individual and update the breeding generation.

[0062] Step 4049: Determine whether the termination condition has been reached. If the termination condition has been reached, output the best individual. If the termination condition has not been reached, continue with the process of step S4041.

[0063] This invention addresses the task of regional exploration for multiple surface unmanned surface vessels (USVs). It considers the navigation performance and detection capabilities of different types of USVs and uses a genetic algorithm to allocate a detection area to each USV while ensuring detection efficiency, thereby maximizing the detection efficiency of the entire USV fleet. Attached Figure Description

[0064] Figure 1A flowchart of the overall algorithm for regional collaborative detection task allocation for multiple surface unmanned surface vessels in this invention;

[0065] Figure 2 This is a detailed flowchart of the algorithm for a multi-surface unmanned surface vessel area collaborative exploration task allocation method based on genetic algorithm according to the present invention. Detailed Implementation

[0066] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0067] This embodiment discloses a multi-surface unmanned surface vessel (USV) regional cooperative exploration task allocation algorithm based on a genetic algorithm, including the following steps:

[0068] S1: Obtain the latitude and longitude coordinates of the endpoints of the area detection mission region and calculate the area S of the detection region. Based on the mission instructions, obtain the area detection coverage p and the expected probability p of the detected target. t Target detection expected probability p t This represents the expected probability of finding the target.

[0069] S2: Obtain the task constraints related to the region exploration mission, including:

[0070] The number of unmanned surface vessels (USVs) performing the mission, n;

[0071] Maximum operating time of unmanned surface vessels (USVs) performing missions (based on endurance) t max ;

[0072] Maximum detection speed of the unmanned surface vessel performing the mission: V 1max V 2max ,…,V nmax ;

[0073] The detection range of the unmanned surface vessel performing the mission: L1, L2, ..., L n ;

[0074] Detection blind spots of unmanned surface vessels performing missions: D1, D2, ..., D n ;

[0075] The probability of an unmanned surface vessel (USV) detecting a target during a mission: p t1 ,p t2 ,…,p tn .

[0076] S3: Based on the exploration mission area and the execution constraints of the unmanned surface vessel mission, the objective function is obtained:

[0077] N1+N2+…+N n ≥S*p

[0078]

[0079]

[0080] Where: N i This represents the size of the detection area of ​​the i-th unmanned surface vessel, where i = 1, 2, ..., n; m di This represents the number of probes required to satisfy the desired probability of detecting the target. This indicates the number of detections required to fill the detection blind spots; m i This indicates that the i-th unmanned surface vessel is detecting the area N. i Total number of detections, m i Round the value up; Represents the combination number, i.e., from m i The number of combinations of selecting one element from a set of elements. The probability of detecting a target is obtained through a binomial distribution, meaning that the probability of detecting the target in each detection is p. ti Detecting m i The probability of detecting the target is [number] times.

[0081] The ultimate goal is to find the detection areas of n unmanned surface vessels [N1, N2, ..., N]. n and the number of probes to this area [m1,m2,…,m] n The goal is to enable n different types of unmanned surface vessels to complete the detection of the mission area in the shortest time, with a detection coverage rate greater than or equal to p and a target detection probability greater than or equal to p. t .

[0082] S4: A genetic algorithm is used to solve the problem, and the final solution is the task assignment result for each unmanned surface vessel, that is: the detection area of ​​each unmanned surface vessel [N1, N2, ..., N]. n and the number of probes to this area [m1,m2,…,m] n Specifically, it includes the following steps:

[0083] S4.1 Randomly generate the initial population (i.e., the parent generation), randomly generate g individuals, and encode the genes in each individual in sequence, such that each individual satisfies:

[0084] N1+N2+…+N n ≥S*p

[0085]

[0086]

[0087] V i ≤V imax

[0088]

[0089] In the formula, V i This represents the detection speed of the i-th unmanned surface vessel.

[0090] Obtain the parent population:

[0091]

[0092] The population consists of individuals, [N1, N2, ..., N n ] and [m1,m2,…,m n ] represents a single chromosome, i.e., an individual, N i and m i It represents a gene on a chromosome.

[0093] S4.2 Calculate the fitness of parent individuals. The fitness of individuals in the population is measured by the detection time; that is, the less time an individual spends detecting, the higher its fitness, and the higher the probability of it being passed on to the next generation. The fitness of the i-th gene is:

[0094] f(i) = 1 / (N i *m i ) / (V i *L i ))

[0095] An individual's fitness is the minimum of the fitness of all genes within that individual, i.e.:

[0096] f = min(f(1), f(2), ..., f(n))

[0097] S4.3 Preserve the best parent individual

[0098] Based on the calculated value of the fitness function, the individual with the largest fitness value is selected as the best individual.

[0099] S4.4 performs reproduction and iteratively searches for the optimal solution, specifically including the following steps:

[0100] S4.4.1 Selection Operation

[0101] The selection operator is equivalent to performing a survival-of-the-fittest operation on individuals in a population. It is based on the calculation of individual fitness in the previous step and transmits individuals in the population to the offspring population with a certain probability. Individuals with high fitness have a greater probability of being transmitted to offspring, while individuals with low fitness have a smaller probability of being transmitted to offspring.

[0102] A proportional selection operator is used in the selection operation.

[0103] The process of implementing the selection operation of a genetic algorithm using the principle of betting odds is as follows:

[0104] (1) Sum the fitness values ​​of all individuals calculated previously to obtain the total fitness value.

[0105] (2) Randomly select a number such that it falls within the range of zero to the total fitness. This step is equivalent to determining the position where the betting pointer stops.

[0106] (3) Start by gradually increasing the fitness of the first individual in sequence until the fitness of the accumulated individual is higher than the selected number.

[0107] (4) Move all individuals at the cutoff position to the breeding population. After the breeding population, crossover, mutation, and other operations need to be performed.

[0108] This method aligns with the mechanism of natural selection, preserving a large number of superior individuals while maintaining the diversity and richness of individuals within the population.

[0109] S4.4.2 Cross Operation

[0110] In genetic algorithms, crossover is the process of exchanging gene segments between paired chromosomes to generate entirely new individuals. Crossover is an important way to generate new individuals and plays a crucial role in expanding the search range of genetic algorithms.

[0111] The crossover operation is performed using a multi-point crossover method. Two individuals are randomly selected for pairing. If the population size is odd, the last remaining individual is not crossovered. A crossover point is arbitrarily chosen in the individual's genetic code, and the contents after the crossover point are exchanged between the paired individuals. Then, it is determined whether the crossovered individuals meet the constraints. If not, the crossover contents are mutated until the constraints are met.

[0112] S4.4.3 Mutation Operation

[0113] Mutation is an auxiliary method for generating new individuals, but it is also indispensable because mutation operators can enhance the local search performance of genetic algorithms, while also maintaining the richness of individuals in the population and preventing premature convergence.

[0114] A gene is randomly selected from an individual as the individual's variant gene. For the variant gene, it is replaced with another randomly generated gene. After mutation, it is determined whether the individual meets the constraints. If not, the gene is mutated again until the constraints are met.

[0115] The probability of mutation is generally between 0.0001 and 0.1. Mutation operators play a decisive role in increasing the diversity of a population.

[0116] S4.4.4 Calculate the fitness of each individual in the offspring: Calculate the fitness of the offspring individuals according to the method in step S4.2.

[0117] S4.4.5 updates the overall fitness of the population for selection using roulette wheel.

[0118] The overall fitness of a population is the sum of the fitness values ​​of each individual in the population.

[0119] S4.4.6 Preserve the best offspring individuals.

[0120] S4.4.7 updates the child generation content to the parent generation, and the child generation becomes the parent generation of the next generation.

[0121] S4.4.8 updates the best individual and the number of breeding generations.

[0122] S4.4.9 Determine if the termination condition has been reached. If the termination condition has been reached, output the best individual. If the termination condition has not been reached, continue with step S4.4.1.

[0123] The termination condition of the genetic algorithm is the maximum number of generations. The optimal solution is output after the genetic algorithm reaches the maximum number of generations. The maximum number of generations is typically between 100 and 1000; in this invention, 200 is chosen as the maximum number of generations.

[0124] This invention considers that when multi-surface unmanned surface vessels (MSVs) perform area detection missions, there are blind spots in the mission payload and the probability of target detection is not 100%. Based on the navigation performance and detection capabilities of different types of MSVs, the optimal allocation strategy for multi-surface unmanned surface vessels to perform area detection missions is calculated by a genetic algorithm, thereby improving the collaborative detection efficiency of multi-surface unmanned surface vessels in actual combat mission environments.

Claims

1. A multi-surface unmanned surface vessel (USV) regional cooperative exploration task allocation algorithm based on genetic algorithm, characterized in that, Includes the following steps: Step 1: Obtain the detection task area parameters; Step 2: Obtain the relevant task constraints for the area detection task; Step 3: Based on the detection task area parameters and the relevant task constraints, obtain the objective function. The final objective is to find the detection areas of n different types of unmanned surface vessels and the number of times each area is detected, so that the n different types of unmanned surface vessels complete the detection of the task area in the shortest time, and the detection coverage and the probability of target discovery are greater than the preset values ​​in the detection task area parameters; Step 4: Use a genetic algorithm to solve the objective function in Step 3. The final solution is the task allocation result for each unmanned surface vessel, that is: the detection area of ​​each unmanned surface vessel [N1, N2, ... N n and the number of probes to this area [m1, m2, ,m n ], where N i Let m represent the size of the detection area of ​​the i-th unmanned surface vessel. i This indicates that the i-th unmanned surface vessel is detecting the area N. i Total number of detections, m i The numerical values ​​are rounded up, i=1,2,…,n, and include the following steps: Step 401, randomly generate an initial population, wherein g individuals are randomly generated from the initial population, and the genes in each individual are encoded sequentially, and each individual satisfies: , , , , In the formula: p is the regional detection coverage; S is the detection area; V i m represents the detection speed of the i-th unmanned surface vessel; di This represents the number of probes required to satisfy the desired probability of detecting the target. Represents the combination number, i.e., from m di The number of combinations of selecting one element from elements; p ti Let p represent the probability of the unmanned surface vessel (USV) detecting the target, i = 1, 2, ..., n; t L represents the expected probability of detecting a target; i / D i This indicates the number of detections required to fill the detection blind spots; V imax L represents the maximum detection speed of the unmanned surface vessel performing the mission. i Indicates the detection range of the unmanned surface vessel performing the mission, i=1,2,…,n; t max This represents the maximum operating time of the unmanned surface vessel performing the mission; the parent population is obtained as shown in the following formula: , where [N1,N2, N n ] and [m1,m2, ,m n ] represents a single chromosome, i.e., an individual, N i and m i This represents a gene on a chromosome; Step 402: Calculate the fitness of individuals in the initial population, measuring the fitness of individuals in the population by the probe time, where the fitness of the i-th gene is: The fitness of an individual is the minimum value of the fitness of all genes in that individual, that is: Step 403: Calculate and save the best initial population individuals: Based on the calculated value of the fitness function, select the individual with the largest fitness value as the best individual; Step 404: Perform reproduction and iteratively search for the optimal solution.

2. The multi-water unmanned surface vehicle regional cooperative exploration task allocation algorithm based on a genetic algorithm according to claim 1, characterized in that, In step 1, the acquired detection task area parameters include a detection area size S, a region detection coverage rate p, and a detection target expected probability p t .

3. The multi-water unmanned surface vehicle regional cooperative exploration task allocation algorithm based on a genetic algorithm according to claim 2, characterized in that, In step 2, the relevant task constraints for the regional exploration mission include: the number of unmanned surface vessels (USVs) n performing the mission; and the maximum operating time t of the USVs performing the mission. max Maximum detection speed of the unmanned surface vessel performing the mission: V 1max V 2max ,…,V nmax The detection range of the unmanned surface vessel performing the mission: L1, L2, ..., L n ; Detection blind spots of unmanned surface vessels performing missions: D1, D2, ..., D n The probability of an unmanned surface vessel detecting a target during a mission: p t1 ,p t2 ,…,p tn .

4. The multi-water unmanned surface vehicle regional cooperative exploration task allocation algorithm based on a genetic algorithm according to claim 3, characterized in that, In step 3, the objective function obtained is: , , ,in: Represents the combination number, i.e., from m i The number of combinations of selecting one element from a set of elements.

5. The multi-water unmanned surface vehicle regional cooperative exploration task allocation algorithm based on a genetic algorithm according to claim 4, characterized in that, In step 3, the ultimate goal of the objective function is to find the detection areas [N1, N2, ..., ...] of n different types of unmanned surface vessels. N n and the number of probes to this area [m1, m2, ,m n The goal is to enable n different types of unmanned surface vessels to complete the detection of the mission area in the shortest time, with a detection coverage rate greater than or equal to p and a target detection probability greater than or equal to p. t .

6. The multi-surface unmanned surface vessel (USV) regional collaborative exploration task allocation algorithm based on genetic algorithm as described in claim 1, characterized in that, Step 404 includes the following steps: Step 4041, Selection Operation: The selection operation is implemented using the roulette wheel principle, specifically including the following steps: a) Accumulate the fitness values ​​of all previously calculated individuals to obtain the total fitness value; b) Randomly select a number that falls within the range of zero to the total fitness; c) Gradually accumulate the fitness values ​​starting from the first individual until the accumulated fitness value is higher than the selected number; d) Add all individuals at the cutoff position to the breeding population; Step 4042, Crossover Operation: The crossover operation is completed using a multi-point crossover method: Randomly select two individuals for pairing. If the population size is odd, the last remaining individual is not crossovered; Randomly select a crossover point in the individual's gene encoding, and exchange the contents of the paired individuals after the crossover point. Then, determine whether the crossovered individuals meet the constraints. If not, mutate the crossover contents until the constraints are met. Step 4043: Mutation operation: Randomly select a gene from an individual as the individual's mutation gene. For the mutation gene, replace it with another randomly generated gene. After mutation, determine whether the individual meets the constraints. If not, continue to mutate the gene until the constraints are met. Step 4044: Calculate the fitness of each individual in the offspring: Calculate the fitness of the offspring individuals according to the method in step 402. Step 4045: Update the total fitness of the population, which is used for selection using roulette wheel. The total fitness of the population is the sum of the fitness values ​​of each individual in the population. Step 4046: Save the best offspring individual. Step 4047: Update the offspring content to the parent generation. This offspring becomes the parent of the next generation. Step 4048: Update the best individual and update the generation number. Step 4049: Determine whether the termination condition has been reached. If the termination condition has been reached, output the best individual. If the termination condition has not been reached, continue with the process in step 4041.

Images