Multi-aircraft task allocation method and system based on collaborative multi-population optimization

Abstract

The invention relates to the technical field of multi-objective optimization, and discloses a multi-aircraft task allocation method and system based on collaborative multi-population optimization, and the method comprises the steps: obtaining the target coordinates of each flight task and the flight parameters of each aircraft; m populations are generated; any individual in any group corresponds to one task allocation strategy; setting iteration parameters for each population, and controlling iteration of each population; identifying an evolution rhythm state of each population; performing cross-population information exchange and population iterative optimization based on the evolution rhythm state; when a preset iteration termination condition is reached, obtaining a candidate solution of each population; and screening the comprehensive recommendation individuals based on the candidate solution of each population, and obtaining a corresponding task allocation strategy. According to the method, the multi-population collaborative optimization capability is improved, and a high-quality multi-aircraft task allocation strategy can be effectively generated.

Description

Technical Field

[0001] This application relates to the technical field of multi-objective optimization, specifically to a multi-aircraft mission allocation method and system based on cooperative multi-swarm optimization. Background Technology

[0002] With the development of multi-UAV technology, systems based on multi-aircraft platforms for mission execution have been widely applied in various fields. Compared to single-aircraft systems, multi-aircraft systems offer advantages such as wider coverage, faster response speed, and stronger redundancy and fault tolerance. However, in complex environments with high mission density and limited resources, how to efficiently and rationally allocate multiple tasks to multiple aircraft to achieve overall system performance optimization remains one of the key challenges in this field. When the number of aircraft and tasks is large, the task allocation combinations grow exponentially, making traditional exhaustive or heuristic algorithms ineffective. In recent years, researchers have proposed intelligent optimization algorithms such as genetic algorithms, particle swarm optimization, and ant colony optimization to search for approximate optimal solutions in the complex solution space of multi-aircraft task allocation. However, these methods still have some shortcomings when facing high-dimensional, multi-objective task allocation problems.

[0003] Traditional optimization methods are prone to getting trapped in local optima during the search process, lack diversity maintenance mechanisms, and struggle to cover the entire solution space. Existing optimization algorithms often employ fixed iteration rules and parameter settings, failing to fully consider the differences in search needs of the population at different evolutionary stages, leading to insufficient early exploration or excessively rapid convergence in later stages. In multi-population optimization frameworks, the communication mechanisms between populations are often designed to be relatively simple, such as fixed-frequency migration, failing to dynamically adjust information flow strategies according to the actual evolutionary state, thus limiting the efficiency of the system's collaborative search.

[0004] For example, Chinese patent CN107330560B discloses a method for collaborative allocation of multiple tasks for heterogeneous aircraft considering time constraints. The method includes: obtaining the aircraft type based on the task and extracting flight performance parameters based on the aircraft type; establishing a collaborative allocation model for multiple tasks of heterogeneous aircraft considering time constraints; and solving the collaborative allocation model using a genetic algorithm considering time constraints, thereby obtaining a collaborative allocation scheme for multiple tasks of heterogeneous aircraft that satisfies task time constraints and maximizes efficiency. This scheme can effectively shorten task execution time and improve task efficiency.

[0005] For example, patent application CN117273346A discloses an integrated optimization method and apparatus for multi-aircraft cooperative task allocation and trajectory planning. The method includes constructing a multi-aircraft cooperative trajectory planning model considering various constraints; constructing a multi-aircraft cooperative attack task allocation model considering various constraints; constructing an integrated multi-aircraft cooperative task allocation and trajectory planning model based on the multi-aircraft cooperative trajectory planning model and the multi-aircraft cooperative attack task allocation model; solving the integrated model using an optimization algorithm; and outputting the optimal integrated cooperative task allocation scheme and cooperative trajectory planning result. This scheme completes target allocation during the trajectory planning process, resulting in a target allocation and trajectory planning scheme with better overall efficiency, thereby avoiding the iterative calculation problem of task allocation and trajectory planning in traditional hierarchical planning methods.

[0006] All of the above patents suffer from the problems mentioned in this background technology: they are prone to getting trapped in local optima during the search process, lack diversity maintenance mechanisms, and are difficult to cover the entire solution space.

[0007] The information disclosed in this background section is intended only to enhance the understanding of the overall background of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0008] The technical problem to be solved by this application is to overcome the defects of the prior art and provide a multi-aircraft task allocation method and system based on cooperative multi-swarm optimization, improve the multi-swarm cooperative optimization capability, and effectively generate high-quality multi-aircraft task allocation strategies.

[0009] To solve the above-mentioned technical problems, this application provides the following technical solution:

[0010] On the one hand, this application provides a multi-aircraft mission allocation method based on cooperative multi-swarm optimization, including the following steps:

[0011] Obtain the target coordinates for each flight mission and the flight parameters for each aircraft;

[0012] Generate M populations, where M is a positive integer; each individual in any population corresponds to a task allocation strategy; each task allocation strategy includes the aircraft assigned to each flight task and the execution order;

[0013] Each population is assigned iteration parameters, and the iteration of each population is controlled based on the iteration parameters and the target coordinates and flight parameters.

[0014] Identify the evolutionary rhythm state of each population; perform cross-population information exchange and iterative optimization of the population based on the evolutionary rhythm state;

[0015] When the preset iteration termination condition is met, obtain the candidate solutions for each population.

[0016] Individuals are comprehensively recommended based on the candidate solutions of each population, and the corresponding task allocation strategy is obtained.

[0017] As a preferred embodiment of the multi-aircraft mission allocation method based on cooperative multi-swarm optimization described in this application, each flight mission corresponds to a target coordinate; the flight parameters of each aircraft include rated speed and rated power.

[0018] Each individual corresponds to one chromosome; each chromosome contains multiple genes, each gene corresponding to a flight mission, and the gene arrangement order is the same on chromosomes corresponding to different individuals; each flight mission and each aircraft is assigned a number, and the number of each gene is inherited from the corresponding flight mission; each gene contains two attributes, namely aircraft number and execution order; where the aircraft number of the gene indicates that the corresponding flight mission is executed by the aircraft with the corresponding number, and the execution order of the gene indicates the order of the flight mission in the mission queue of the corresponding aircraft;

[0019] In population iteration, the number and order of genes on the chromosome corresponding to any individual remain unchanged, and the attributes of genes are iteratively updated through crossover and mutation operations in the genetic algorithm.

[0020] As a preferred embodiment of the multi-aircraft task allocation method based on cooperative multi-population optimization described in this application, the iteration parameters of any population include crossover probability, mutation probability, and fitness function; the iteration process of any population is as follows: crossover operation is performed on the chromosomes of different individuals in the population based on the crossover probability; mutation operation is performed on the chromosomes of different individuals in the population based on the mutation probability; the fitness of each individual is calculated based on the fitness function, and a specified number of individuals with the highest fitness are selected to enter the next iteration.

[0021] The fitness function for each population is a weighted sum of time factor, energy factor, and load balancing factor, with the reciprocal of the sum yielding the fitness. The weights of the time factor, energy factor, and load balancing factor range from [value missing]. The sum of these factors is 1; the time factor is the maximum total duration of flight missions performed by different aircraft; the energy consumption factor is the total energy consumption of all aircraft performing flight missions; and the load balancing factor is the variance of the energy consumption of different aircraft performing flight missions.

[0022] As a preferred embodiment of the multi-aircraft task allocation method based on cooperative multi-swarm optimization described in this application, the evolution rhythm state includes an initial state, a first convergence state, and a second convergence state; the evolution rhythm state at the beginning of the first iteration of any swarm is the initial state;

[0023] The method for identifying the evolutionary rhythm state of any population is as follows:

[0024] At the start of each iteration, the evolutionary pace index is calculated based on the chromosome corresponding to each individual and the fitness of each individual.

[0025] The evolutionary rhythm indicators include fitness fluctuation indicators, population diversity indicators, and optimization trend indicators; a first threshold and a second threshold are set for the fitness fluctuation indicators, and the first threshold is greater than the second threshold; a population diversity threshold and an optimization trend threshold are set.

[0026] Identifying the evolutionary rhythm of a population based on the evolutionary rhythm index specifically includes: for any population in the initial evolutionary rhythm state, if the fitness fluctuation index is less than the corresponding first threshold and the optimization trend index is less than the optimization trend threshold, then the evolutionary rhythm state of the population is marked as the first convergence state; for any population in the initial or first convergence state, if the fitness fluctuation index is less than the corresponding second threshold and the population diversity index is less than the population diversity threshold, then the evolutionary rhythm state of the population is marked as the second convergence state.

[0027] As a preferred embodiment of the multi-aircraft mission allocation method based on cooperative multi-population optimization described in this application, the step of cross-population information exchange and iterative optimization includes iterative optimization of the population in the first convergence state; specifically, it includes:

[0028] If the population is marked as the first convergent state from the initial state in the current iteration, then the crossover probability of the population is reduced and the mutation probability is locally adjusted. The local adjustment of the mutation probability specifically includes: sorting the fitness of each individual by size; marking the N individuals with the highest fitness as elite individuals; N is a positive integer; and increasing the mutation probability of elite individuals.

[0029] As a preferred embodiment of the multi-aircraft task allocation method based on cooperative multi-population optimization described in this application, the iterative optimization of the population in the first convergence state further includes:

[0030] Any population in the first convergence state is marked as the first target population, and each population other than the first target population is marked as a migrating population.

[0031] Elite individuals from each migrating population are selected as candidate migrating individuals;

[0032] The fitness of each candidate migrant in the first target population is calculated based on the fitness function of the first target population, and is used as the target fitness of each candidate migrant.

[0033] Set a target fitness threshold; remove candidate migrants whose target fitness is less than the target fitness.

[0034] The gene arrangement order and attributes on the chromosomes of each individual in the first target population and each candidate migrant individual are encoded into the feature vector of the corresponding individual.

[0035] Calculate the structural heterogeneity of each candidate migrant individual based on the feature vector; mark the m candidate migrant individuals with the highest structural heterogeneity as migrant individuals; m is a positive integer;

[0036] Copy m migrating individuals to the first target population, and remove the m individuals with the lowest fitness from the first target population.

[0037] As a preferred embodiment of the multi-aircraft task allocation method based on cooperative multi-population optimization described in this application, the step of cross-population information exchange and iterative optimization further includes iterative optimization of the population in the second convergence state, specifically including:

[0038] Any population in the second convergence state is marked as the second target population, and each population other than the second target population is marked as an interaction population.

[0039] Elite individuals from each interaction population are selected as candidate crossover individuals; candidate gene fragments are extracted from the chromosomes corresponding to the candidate crossover individuals.

[0040] Calculate the structural variability and local superiority of each candidate gene fragment; calculate the interaction priority of each candidate gene fragment based on the structural variability and local superiority.

[0041] Select the n candidate gene fragments with the highest interaction priority as crossover gene fragments; n is a positive integer; select n individuals from the second target population as individuals to be crossovered;

[0042] Assign a crossover gene segment to each individual to be crossovered; replace the corresponding gene segment on the chromosome of the individual to be crossovered with the crossover gene segment.

[0043] As a preferred embodiment of the multi-aircraft mission allocation method based on cooperative multi-population optimization described in this application, the method for calculating the structural variability of any candidate gene fragment is as follows:

[0044] The gene sequence and attributes in the candidate gene fragments are encoded into the feature vectors of the corresponding candidate gene fragments.

[0045] The similarity between the feature vector of the candidate gene fragment and the feature vector of each individual in the second target population is calculated by dynamic time warping and the average value is taken as the structural variability of the candidate gene fragment.

[0046] The method for calculating the local superiority of any candidate gene segment is as follows:

[0047] The fitness of candidate gene fragments is calculated based on the fitness function of the second target population, and is used as the local superiority of the candidate gene fragments.

[0048] As a preferred embodiment of the multi-aircraft task allocation method based on cooperative multi-population optimization described in this application, the method for obtaining candidate solutions for each population is as follows: setting a fitness threshold for each population; marking individuals in each population with a fitness greater than the corresponding fitness threshold as candidate solutions.

[0049] The method for selecting individuals for comprehensive recommendations is as follows:

[0050] Calculate the time factor, energy consumption factor, and load balancing factor for each candidate solution.

[0051] Normalize the time factor, energy consumption factor and load balancing factor of each individual corresponding to each candidate solution and form a multi-objective vector for each candidate solution.

[0052] Based on the multi-objective vector of each candidate solution, a fast non-dominated sorting algorithm is used to filter the non-dominated solutions among the candidate solutions, and each non-dominated solution is marked as a comprehensive recommendation individual.

[0053] Secondly, this application provides a multi-aircraft mission allocation system based on cooperative multi-swarm optimization, including a data processing module, an algorithm module, an identification module, an optimization module, and a strategy module; wherein:

[0054] The data processing module is configured with the target coordinates for each flight mission and the flight parameters for each aircraft;

[0055] The algorithm module generates a population for exploring task allocation strategies, sets iteration parameters for each population, and controls the iteration of each population based on the iteration parameters;

[0056] In each iteration, the algorithm module controls the crossover and mutation operations; the data processing module simulates the execution of the task allocation strategy and calculates the individual fitness based on the target coordinates and flight parameters; the algorithm module selects individuals to enter the next iteration based on the individual fitness.

[0057] The identification module is used to identify the evolutionary rhythm state of each population;

[0058] The optimization module performs cross-population information exchange and iterative optimization of the population based on the evolutionary rhythm state;

[0059] The algorithm module is also configured with an iteration termination condition; when the preset iteration termination condition is reached, the algorithm module obtains the candidate solutions for each population.

[0060] The strategy module filters and recommends individuals based on the candidate solutions of each population, and obtains the corresponding task allocation strategy.

[0061] Compared with the prior art, the beneficial effects achieved by this application are as follows:

[0062] This application identifies the evolutionary rhythm state of various groups and dynamically adjusts the crossover rate, mutation density, and information exchange strategy. This effectively avoids premature convergence, enhances local search capabilities, improves the overall efficiency and stability of the optimization process, and improves the quality of multi-aircraft mission allocation schemes.

[0063] By structurally adapting through the migration of elite individuals and genetic recombination among different populations, the effectiveness and diversity of cross-population information exchange can be improved, which helps to more comprehensively explore the scheduling strategy space and enhance the global optimization capability of flight mission allocation. Attached Figure Description

[0064] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0065] Figure 1 A flowchart of the multi-aircraft mission allocation method based on cooperative multi-swarm optimization provided in this application;

[0066] Figure 2 A schematic diagram of the structure of the multi-aircraft mission allocation system based on cooperative multi-swarm optimization provided in this application. Detailed Implementation

[0067] The technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of this application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0068] Example 1

[0069] This embodiment introduces a multi-aircraft mission allocation method based on cooperative multi-swarm optimization, referring to... Figure 1 The method includes the following steps:

[0070] Obtain the target coordinates for each flight mission and the flight parameters for each aircraft;

[0071] Each flight mission corresponds to a target coordinate; flight missions include, but are not limited to, high-resolution imaging of the location corresponding to the target coordinate, coverage search of the area corresponding to the target coordinate, and transporting resources to the area where the target coordinate is located. Different flight missions have spatial and temporal competition or coupling relationships, have independent execution attributes, and face pressure for coordinated path scheduling.

[0072] The flight parameters for each aircraft include rated speed and rated power. Rated speed is the stable flight speed of the aircraft under standard cruise conditions, which determines the mission execution time; rated power is the average power consumption level of the aircraft during stable flight, used to calculate the energy consumption required for mission execution.

[0073] Generate M populations, where M is a positive integer; each individual in any population corresponds to a task allocation strategy; each task allocation strategy includes the aircraft assigned to each flight task and the execution order;

[0074] Each individual corresponds to one chromosome; each chromosome contains multiple genes, each gene corresponding to a flight task, and the gene arrangement order is the same across chromosomes of different individuals; each flight task and each spacecraft is assigned a number, with each gene's number inherited from its corresponding flight task; each gene contains two attributes: spacecraft number and execution order; the spacecraft number of a gene indicates that the corresponding flight task is executed by the spacecraft with that number, and the execution order of the gene indicates the order of the flight task in the task queue of the corresponding spacecraft; for example, if the spacecraft number of any gene is p and the execution order is q, it means that the flight task corresponding to this gene is executed by the spacecraft with number p, and this spacecraft needs to execute at least p flight tasks, and the flight task corresponding to this gene is the p-th flight task it needs to execute. The gene arrangement order and gene attributes on the chromosome corresponding to any individual constitute the encoding form of the task allocation strategy for that individual.

[0075] In population iteration, the number and order of genes on the chromosome corresponding to any individual remain unchanged, while the attributes of the genes are iteratively updated through crossover and mutation operations in the genetic algorithm.

[0076] Set iteration parameters for each population, and control the iteration of each population based on the iteration parameters;

[0077] The iteration parameters for any population include crossover probability, mutation probability, and fitness function. The iteration process for any population is as follows: crossover operation is performed on the chromosomes of different individuals in the population based on the crossover probability; mutation operation is performed on the chromosomes of different individuals in the population based on the mutation probability; the fitness of each individual is calculated based on the fitness function, and a specified number of individuals with the highest fitness are selected to enter the next round of iteration.

[0078] The crossover probability is the probability that an individual's chromosomes will undergo a crossover operation; the mutation probability is the probability that an individual's chromosomes will undergo a mutation operation. The number of individuals selected to enter the next iteration in each iteration is set by those skilled in the art based on actual needs. For example, it can be set to always be equal to the size of the initial population, or gradually decrease as the number of iterations increases.

[0079] The fitness function for each population is a weighted sum of time factor, energy factor, and load balancing factor, with the reciprocal of the sum yielding the fitness. The weights of the time factor, energy factor, and load balancing factor range from [value missing]. The sum of these factors is 1; the time factor is the maximum total duration of flight missions performed by different aircraft; the energy consumption factor is the total energy consumption of all aircraft performing flight missions; and the load balancing factor is the variance of the energy consumption of different aircraft performing flight missions.

[0080] In this embodiment, the weights of the time factor, energy consumption factor, and load balancing factor in the fitness function of each population are different to form differentiated exploration directions. In any iteration, new individuals are generated through crossover and mutation operations. Then, the task allocation strategy corresponding to each individual is obtained and simulated. Based on the simulation results, the time factor, energy consumption factor, and load balancing factor of each individual are calculated, and the fitness of each individual is then calculated. Specifically, this includes: simulating the flight path of each aircraft based on the target coordinates of each flight task in the task allocation strategy, the flight tasks to be performed by each aircraft, and the execution order; calculating the time required for each aircraft to complete all its assigned flight tasks based on the aircraft's rated speed; the maximum task time for different aircraft is the time factor, which determines the time point when all flight tasks are completed. Based on the task time of each flight and the rated power of the aircraft, the energy consumption of each aircraft is calculated, and then the energy consumption factor and load balancing factor are calculated.

[0081] In this embodiment, the crossover or mutation probabilities of different populations can be set uniformly, or differentiated by those skilled in the art based on the weight distribution characteristics of the fitness function of each population. For example, if the path balance factor has a large weight in the fitness function of a certain population, then that population focuses more on exploring the task allocation strategy with the smallest path balance factor. Its crossover probability can be set to a larger value to frequently recombine the task allocation structure of different individuals, increase the ways in which tasks can be combined among various aircraft, and help to discover more balanced task allocation strategy patterns and improve load balancing effect.

[0082] Identify the evolutionary rhythm state of each population; perform cross-population information exchange and iterative optimization of the population based on the evolutionary rhythm state;

[0083] The evolutionary rhythm state includes the initial state, the first convergence state, and the second convergence state; the evolutionary rhythm state at the beginning of the first iteration of any population is the initial state;

[0084] The method for identifying the evolutionary rhythm state of any population is as follows:

[0085] At the start of each iteration, the evolutionary pace index is calculated based on the chromosome corresponding to each individual and the fitness of each individual.

[0086] The evolutionary rhythm indicators include fitness fluctuation indicators, population diversity indicators, and optimization trend indicators; a first threshold and a second threshold are set for the fitness fluctuation indicators, and the first threshold is greater than the second threshold; a population diversity threshold and an optimization trend threshold are set.

[0087] Identifying the evolutionary rhythm of a population based on the evolutionary rhythm index specifically includes: for any population in the initial evolutionary rhythm state, if the fitness fluctuation index is less than the corresponding first threshold and the optimization trend index is less than the optimization trend threshold, then the evolutionary rhythm state of the population is marked as the first convergence state; for any population in the initial or first convergence state, if the fitness fluctuation index is less than the corresponding second threshold and the population diversity index is less than the population diversity threshold, then the evolutionary rhythm state of the population is marked as the second convergence state.

[0088] Those skilled in the art can set a first threshold and a second threshold for the fitness fluctuation index, as well as a population diversity threshold and an optimization trend threshold, based on actual needs. Optionally, the fitness fluctuation index is the variance of the fitness of each individual; the optimization trend index is the rate of change of the highest fitness, i.e., the ratio of the change in the highest fitness to the highest fitness in the previous iteration; the change in fitness is the difference between the highest fitness among all individuals at the beginning of the current iteration and the highest fitness among all individuals in the previous iteration.

[0089] Optionally, the method for calculating the population diversity index is as follows: the gene arrangement order and attributes on the chromosome of each individual are encoded into the feature vector of the corresponding individual; any two individuals are extracted as a comparison group, and the Hamming distance between the feature vectors of the two individuals in the comparison group is calculated as the diversity index of the comparison group; the mean of the diversity indices of all comparison groups in the population is calculated as the population diversity index.

[0090] The cross-population information exchange and iterative optimization includes iterative optimization of the population in the first convergence state; specifically, it includes:

[0091] If the population is marked as the first convergent state from the initial state in the current iteration, then the crossover probability of the population is reduced and the mutation probability is locally adjusted. The local adjustment of the mutation probability specifically includes: sorting the fitness of each individual by size; marking the N individuals with the highest fitness as elite individuals; N is a positive integer; and increasing the mutation probability of elite individuals.

[0092] A population in the first convergence state exhibits a low optimization trend index during the iteration process, and the speed of exploring the optimal solution gradually slows down. By reducing the crossover probability and locally adjusting the mutation probability, for example, setting the crossover probability of the population to 50% of its initial value and increasing the mutation probability of elite individuals by 50%, the probability of the structure of high-quality individuals being destroyed can be reduced, and the population can be guided to deeply explore around the current high-quality individuals, thereby improving the fine-grained search capability of the genetic algorithm and accelerating iterative convergence.

[0093] The iterative optimization of the population in the first convergence state also includes:

[0094] Any population in the first convergence state is marked as the first target population, and each population other than the first target population is marked as a migrating population.

[0095] Elite individuals are extracted from each migrating population as candidate migrating individuals; in this embodiment, elite individuals in any population include the N individuals with the highest fitness in the population at the start of the current iteration.

[0096] The fitness of each candidate migrant in the first target population is calculated based on the fitness function of the first target population, and is used as the target fitness of each candidate migrant.

[0097] Set a target fitness threshold; remove candidate migrants whose target fitness is less than the target fitness; optionally, set the target fitness threshold to the mean fitness of all individuals in the first target population.

[0098] The gene arrangement order and attributes on the chromosomes of each individual in the first target population and each candidate migrant individual are encoded into the feature vector of the corresponding individual.

[0099] The structural heterogeneity of each candidate migrant individual is calculated based on the feature vector; the m candidate migrant individuals with the highest structural heterogeneity are marked as migrant individuals; m is a positive integer; optionally, for any candidate migrant individual, its structural heterogeneity is the mean of the Hamming distance between its feature vector and the feature vector of each individual in the first target population.

[0100] Copy m migrating individuals to the first target population, and remove the m individuals with the lowest fitness from the first target population.

[0101] A population in the first convergence state experiences slower convergence and is prone to getting trapped in local optima. Introducing high-quality individuals from other populations can guide the population towards a new high-quality solution space, breaking the inefficient convergence state. This embodiment avoids optimization strategy incompatibility and information exchange failure caused by blindly introducing individuals from other populations by screening candidate migrants and further screening them.

[0102] The cross-population information exchange and iterative optimization also includes iterative optimization of the population in the second convergence state, specifically including:

[0103] Any population in the second convergence state is marked as the second target population, and each population other than the second target population is marked as an interaction population.

[0104] Elite individuals from each interaction population are selected as candidate crossover individuals; candidate gene segments are extracted from the chromosomes corresponding to the candidate crossover individuals; for example, the extraction length is set to t, and multiple candidate gene segments are extracted from the chromosomes of different candidate crossover individuals, with each candidate gene segment containing t genes.

[0105] Calculate the structural variability and local superiority of each candidate gene fragment; calculate the interaction priority of each candidate gene fragment based on the structural variability and local superiority; optionally, the interaction priority is obtained by weighted summation of the structural variability and local superiority, and the sum of their weights is 1; those skilled in the art can set the weights of structural variability and local superiority based on actual needs.

[0106] Select the n candidate gene fragments with the highest interaction priority as crossover gene fragments; n is a positive integer; select n individuals from the second target population as individuals to be crossovered;

[0107] Each individual to be crossovered is assigned a crossover gene segment; the corresponding gene segment on the chromosome of the individual to be crossovered is replaced with the crossover gene segment. For example, if the crossover gene segment corresponds to flight missions 1-5, then the gene segment on the chromosome of the individual to be crossovered corresponding to flight missions 1-5 is replaced with the crossover segment. In this embodiment, when selecting individuals to be crossovered from the second target population, elite individuals are not selected to prevent the destruction of high-quality solutions.

[0108] The method for calculating the structural variability of any candidate gene fragment is as follows:

[0109] The gene sequence and attributes in the candidate gene fragments are encoded into the feature vectors of the corresponding candidate gene fragments.

[0110] The similarity between the feature vector of the candidate gene fragment and the feature vector of each individual in the second target population is calculated by dynamic time warping and the average value is taken as the structural variability of the candidate gene fragment.

[0111] The method for calculating the local superiority of any candidate gene segment is as follows:

[0112] The fitness of candidate gene fragments is calculated based on the fitness function of the second target population, and is used as the local superiority of the candidate gene fragments.

[0113] In this embodiment, the fitness function of any population is calculated by the allocation results of all flight missions on the complete chromosome; the candidate gene fragment is not a complete chromosome, but the fitness can still be calculated according to the fitness function. The only difference is that the number of flight missions involved in the calculation is smaller, and the fitness calculation process remains the same.

[0114] In a population in the second convergence state, the genetic structure of individuals tends to be uniform, and the population lacks a basis for further optimization. This application addresses this by injecting gene fragments from other populations into a population in the second convergence state, stimulating new combinations of gene sequences and enabling it to escape the path convergence dilemma. The higher the structural variability, the richer the genetic diversity that the fragment can introduce into the second target population. Through further screening based on local superiority, meaningless perturbations introduced into conventional structures can be avoided, enhancing the heuristic value of gene injection.

[0115] When the preset iteration termination condition is met, obtain the candidate solutions for each population.

[0116] Optionally, the iteration termination condition is reaching the maximum number of iterations; the method for obtaining candidate solutions for each population is as follows: set a fitness threshold for each population; mark individuals in each population whose fitness is greater than the corresponding fitness threshold as candidate solutions.

[0117] Individuals are comprehensively recommended based on the candidate solutions of each population, and the corresponding task allocation strategy is obtained.

[0118] The method for selecting individuals for comprehensive recommendations is as follows:

[0119] Calculate the time factor, energy consumption factor, and load balancing factor for each candidate solution.

[0120] Normalize the time factor, energy consumption factor and load balancing factor of each individual corresponding to each candidate solution and form a multi-objective vector for each candidate solution.

[0121] Based on the multi-objective vector of each candidate solution, a fast non-dominated sorting algorithm is used to filter non-dominated solutions from the candidate solutions, and each non-dominated solution is marked as a comprehensive recommendation individual. For example, the NSGA-II algorithm is used to filter non-dominated solutions from the candidate solutions.

[0122] Optionally, based on actual needs, any one of the time factor, energy consumption factor, and load balancing factor can be set as a key factor, and the non-dominated solution with the largest key factor is marked as the unique comprehensive recommended individual. The genetic attributes of the chromosome records of the comprehensive recommended individual are converted into the allocation results of the corresponding flight tasks to obtain the final recommended task allocation strategy.

[0123] Multi-population optimization can lead to multiple policy orientations, such as time-priority and energy-priority optimization, and enable information exchange among populations, promoting the fusion of local optima towards the global optimum across different policy directions. However, populations often exhibit asynchronous convergence trends, and traditional schemes use a uniform iteration rhythm, ignoring the differences in convergence states among different populations. This application identifies the evolutionary rhythm state, enabling differentiated iterative optimization for populations in different convergence states, improving computational resource utilization, and preventing premature convergence or resource waste.

[0124] Example 2

[0125] Based on the same inventive concept as Embodiment 1, refer to Figure 2 This embodiment introduces a multi-aircraft mission allocation system based on cooperative multi-swarm optimization, including a data processing module, an algorithm module, a recognition module, an optimization module, and a strategy module; wherein:

[0126] The data processing module is configured with the target coordinates for each flight mission and the flight parameters for each aircraft; the flight parameters for each aircraft include rated speed and rated power.

[0127] The algorithm module generates a population for exploring task allocation strategies, sets iteration parameters for each population, and controls the iteration of each population based on the iteration parameters. During each iteration, the algorithm module controls the crossover and mutation operations. The data processing module simulates the execution of the task allocation strategy based on the target coordinates and flight parameters and calculates the individual fitness. The algorithm module selects individuals to enter the next iteration based on the individual fitness.

[0128] The identification module is used to identify the evolutionary rhythm state of each population; the evolutionary rhythm state includes the initial state, the first convergence state, and the second convergence state; the evolutionary rhythm state at the beginning of the first iteration of any population is the initial state.

[0129] The optimization module performs cross-population information exchange and iterative optimization of the population based on the evolutionary rhythm state, including iterative optimization of the population in the first convergence state and the second convergence state.

[0130] The algorithm module is also configured with an iteration termination condition; when the preset iteration termination condition is reached, the algorithm module obtains the candidate solutions for each population.

[0131] The strategy module filters candidates from each population to select recommended individuals and obtains the corresponding task allocation strategy. The strategy module is equipped with a fast non-dominated sorting algorithm to select non-dominated solutions from the candidate solutions as recommended individuals. It then converts the genetic attributes recorded on the chromosomes of the recommended individuals into the allocation results for the corresponding flight tasks, thus obtaining the recommended task allocation strategy.

[0132] The specific functional implementation of each module is described in the relevant content of the multi-aircraft task allocation method based on cooperative multi-swarm optimization in Example 1, and will not be repeated here.

[0133] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0134] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of protection of this application, and these forms are all within the protection scope of this application.

Claims

1. A multi-aircraft mission allocation method based on cooperative multi-swarm optimization, characterized in that: Includes the following steps: Obtain the target coordinates for each flight mission and the flight parameters for each aircraft; Generate M populations, where M is a positive integer; each individual in any population corresponds to a task allocation strategy; each task allocation strategy includes the aircraft assigned to each flight task and the execution order; Each population is assigned iteration parameters, and the iteration of each population is controlled based on the iteration parameters and the target coordinates and flight parameters. Identify the evolutionary rhythm state of each population; perform cross-population information exchange and iterative optimization of the population based on the evolutionary rhythm state; When the preset iteration termination condition is met, obtain the candidate solutions for each population. Individuals are comprehensively recommended based on the candidate solutions of each population, and the corresponding task allocation strategy is obtained.

2. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 1, characterized in that: Each flight mission corresponds to a target coordinate; the flight parameters of each aircraft include rated speed and rated power. Each individual corresponds to one chromosome; each chromosome contains multiple genes, each gene corresponding to a flight mission, and the gene arrangement order is the same on chromosomes corresponding to different individuals; each flight mission and each aircraft is assigned a number, and the number of each gene is inherited from the corresponding flight mission; each gene contains two attributes, namely aircraft number and execution order; where the aircraft number of the gene indicates that the corresponding flight mission is executed by the aircraft with the corresponding number, and the execution order of the gene indicates the order of the flight mission in the mission queue of the corresponding aircraft; In population iteration, the number and order of genes on the chromosome corresponding to any individual remain unchanged, and the attributes of genes are iteratively updated through crossover and mutation operations in the genetic algorithm.

3. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 2, characterized in that: The iteration parameters for any population include crossover probability, mutation probability, and fitness function. The iteration process for any population is as follows: crossover operation is performed on the chromosomes of different individuals in the population based on the crossover probability; mutation operation is performed on the chromosomes of different individuals in the population based on the mutation probability; the fitness of each individual is calculated based on the fitness function, and a specified number of individuals with the highest fitness are selected to enter the next iteration. The fitness function for each population is a weighted sum of time factor, energy factor, and load balancing factor, with the reciprocal of the sum yielding the fitness. The weights of the time factor, energy factor, and load balancing factor range from [value missing]. The sum of these factors is 1; the time factor is the maximum total duration of flight missions performed by different aircraft; the energy consumption factor is the total energy consumption of all aircraft performing flight missions; and the load balancing factor is the variance of the energy consumption of different aircraft performing flight missions.

4. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 3, characterized in that: The evolutionary rhythm state includes the initial state, the first convergence state, and the second convergence state; the evolutionary rhythm state at the beginning of the first iteration of any population is the initial state; The method for identifying the evolutionary rhythm state of any population is as follows: At the start of each iteration, the evolutionary pace index is calculated based on the chromosome corresponding to each individual and the fitness of each individual. The evolutionary rhythm indicators include fitness fluctuation indicators, population diversity indicators, and optimization trend indicators; Set a first threshold and a second threshold for the fitness fluctuation index, where the first threshold is greater than the second threshold; Set population diversity thresholds and optimization trend thresholds; Identifying the evolutionary rhythm of a population based on the evolutionary rhythm index specifically includes: for any population in the initial evolutionary rhythm state, if the fitness fluctuation index is less than the corresponding first threshold and the optimization trend index is less than the optimization trend threshold, then the evolutionary rhythm state of the population is marked as the first convergence state; for any population in the initial or first convergence state, if the fitness fluctuation index is less than the corresponding second threshold and the population diversity index is less than the population diversity threshold, then the evolutionary rhythm state of the population is marked as the second convergence state.

5. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 4, characterized in that: The cross-population information exchange and iterative optimization includes iterative optimization of the population in the first convergence state; specifically, it includes: If the population is marked as the first convergent state from the initial state in the current iteration, then the crossover probability of the population is reduced and the mutation probability is locally adjusted. The local adjustment of the mutation probability specifically includes: sorting the fitness of each individual by size; marking the N individuals with the highest fitness as elite individuals; N is a positive integer; and increasing the mutation probability of elite individuals.

6. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 5, characterized in that: The iterative optimization of the population in the first convergence state also includes: Any population in the first convergence state is marked as the first target population, and each population other than the first target population is marked as a migrating population. Elite individuals from each migrating population are selected as candidate migrating individuals; The fitness of each candidate migrant in the first target population is calculated based on the fitness function of the first target population, and is used as the target fitness of each candidate migrant. Set a target fitness threshold; remove candidate migrants whose target fitness is less than the target fitness. The gene arrangement order and attributes on the chromosomes of each individual in the first target population and each candidate migrant individual are encoded into the feature vector of the corresponding individual. Calculate the structural heterogeneity of each candidate migrant individual based on the feature vector; mark the m candidate migrant individuals with the highest structural heterogeneity as migrant individuals; m is a positive integer; Copy m migrating individuals to the first target population, and remove the m individuals with the lowest fitness from the first target population.

7. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 6, characterized in that: The cross-population information exchange and iterative optimization also includes iterative optimization of the population in the second convergence state, specifically including: Any population in the second convergence state is marked as the second target population, and each population other than the second target population is marked as an interaction population. Elite individuals from each interaction population are selected as candidate crossover individuals; candidate gene fragments are extracted from the chromosomes corresponding to the candidate crossover individuals. Calculate the structural variability and local superiority of each candidate gene fragment; calculate the interaction priority of each candidate gene fragment based on the structural variability and local superiority. Select the n candidate gene fragments with the highest interaction priority as crossover gene fragments; n is a positive integer; select n individuals from the second target population as individuals to be crossovered; Assign a crossover gene segment to each individual to be crossovered; replace the corresponding gene segment on the chromosome of the individual to be crossovered with the crossover gene segment.

8. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 7, characterized in that: The method for calculating the structural variability of any candidate gene fragment is as follows: The gene sequence and attributes in the candidate gene fragments are encoded into the feature vectors of the corresponding candidate gene fragments. The similarity between the feature vector of the candidate gene fragment and the feature vector of each individual in the second target population is calculated by dynamic time warping and the average value is taken as the structural variability of the candidate gene fragment. The method for calculating the local superiority of any candidate gene segment is as follows: The fitness of candidate gene fragments is calculated based on the fitness function of the second target population, and is used as the local superiority of the candidate gene fragments.

9. The multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in claim 8, characterized in that: The method for obtaining candidate solutions for each population is as follows: set a fitness threshold for each population; mark individuals in each population whose fitness is greater than the corresponding fitness threshold as candidate solutions; The method for selecting individuals for comprehensive recommendations is as follows: Calculate the time factor, energy consumption factor, and load balancing factor for each candidate solution. Normalize the time factor, energy consumption factor and load balancing factor of each individual corresponding to each candidate solution and form a multi-objective vector for each candidate solution. Based on the multi-objective vector of each candidate solution, a fast non-dominated sorting algorithm is used to filter the non-dominated solutions among the candidate solutions, and each non-dominated solution is marked as a comprehensive recommendation individual.

10. A multi-aircraft mission allocation system based on cooperative multi-swarm optimization, used to implement the multi-aircraft mission allocation method based on cooperative multi-swarm optimization as described in any one of claims 1-9, characterized in that: It includes a data processing module, an algorithm module, a recognition module, an optimization module, and a strategy module; among which: The data processing module is configured with the target coordinates for each flight mission and the flight parameters for each aircraft; The algorithm module generates a population for exploring task allocation strategies, sets iteration parameters for each population, and controls the iteration of each population based on the iteration parameters; In each iteration, the algorithm module controls the crossover and mutation operations; the data processing module simulates the execution of the task allocation strategy and calculates the individual fitness based on the target coordinates and flight parameters; the algorithm module selects individuals to enter the next iteration based on the individual fitness. The identification module is used to identify the evolutionary rhythm state of each population; The optimization module performs cross-population information exchange and iterative optimization of the population based on the evolutionary rhythm state; The algorithm module is also configured with an iteration termination condition; when the preset iteration termination condition is reached, the algorithm module obtains the candidate solutions for each population. The strategy module filters and recommends individuals based on the candidate solutions of each population, and obtains the corresponding task allocation strategy.

Images