A low earth orbit satellite-based multi-target scheduling method, device and medium

Abstract

The application relates to the field of aerospace technology, and particularly discloses a multi-target scheduling method and device based on low-orbit satellites and a medium, the method comprising the following steps: establishing a three-dimensional visibility matrix of multiple satellites and space multi-targets at discrete time steps, and constructing a to-be-optimized model based on the three-dimensional visibility matrix; generating an initial scheduling decision solution set population and an initial scheduling decision matrix based on the to-be-optimized model; performing an asymmetric multi-index performance measurement on the solution set in the initial scheduling decision solution set population, thereby updating the initial scheduling decision matrix to obtain multiple scheduling decision matrices; and selecting an optimal scheduling decision scheme from multiple scheduling decision schemes corresponding to the multiple scheduling decision matrices based on decision weights.

Description

Technical Field

[0001] This invention relates to the field of aerospace technology, specifically to a multi-objective scheduling method, device, and medium based on low-Earth orbit satellites in the field of aerospace mission planning technology. Background Technology

[0002] With the development of aerospace technology, low-Earth orbit (LEO) satellite constellations have been widely used in the field of space target detection due to their advantages such as global coverage, short revisit cycles, and low deployment costs. As the number of on-orbit space targets increases dramatically and the constellation size continues to expand, it becomes necessary to develop efficient and reasonable multi-target scheduling schemes under limited computing resources, strict on-board constraints, and dynamically changing target visibility windows.

[0003] In related technologies, intelligent optimization algorithms, as a multi-objective scheduling method based on low-Earth orbit satellites, are highly versatile and have outstanding global search capabilities. However, these algorithms are prone to premature convergence or getting trapped in local optima, and the Pareto solution set obtained is unevenly distributed in the target space, making it difficult to balance the algorithm's convergence and distribution. In contrast, the multi-modal multi-objective satellite mission scheduling method based on ring topology disclosed in patent document CN121860249A has advantages in balancing the algorithm's convergence and distribution. However, this algorithm still has the following shortcomings: 1. Scheduling schemes obtained under complex constraints are inefficient. When combined constraints exist, such as "dual-satellite collaborative observation," the scheduling schemes obtained by the algorithm tend to take longer to process, resulting in low efficiency when observing multiple targets. 2. Low engineering feasibility. The algorithm only focuses on theoretical performance and does not fully consider practical needs such as limited on-board resources, resulting in poor executability of the generated scheduling scheme in practice.

[0004] Therefore, how to obtain an efficient and feasible multi-objective scheduling method based on low-Earth orbit satellites is an urgent technical problem to be solved. Summary of the Invention

[0005] This invention discloses a multi-objective scheduling method, apparatus, and medium based on low-Earth orbit satellites, aiming to improve the efficiency and executability of multi-objective scheduling methods based on low-Earth orbit satellites.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, embodiments of this application provide a multi-objective scheduling method based on low-Earth orbit satellites, including: Based on the low-Earth orbit constellation distribution and satellite kinematics model, a three-dimensional visibility matrix of multiple satellites and multiple targets in space is established at discrete time steps. Based on the three-dimensional visibility matrix, an optimization model is constructed. The optimization space of the optimization model includes conflicting targets, including comprehensive observation benefits, tracking accuracy, and constellation system load balance. Based on the model to be optimized, a discrete solution space random mapping that satisfies the rigid constraint of binary star cooperation is performed in the space to be optimized to generate an initial scheduling decision solution set population, and an initial scheduling decision matrix corresponding to the initial scheduling decision solution set population is generated simultaneously. Asymmetric multi-index performance measurement is performed on the solutions in the initial scheduling decision solution set population: the number of non-dominated individuals in the initial scheduling decision solution set population is counted to establish the original dominance benchmark; an offset vector is introduced into the target evaluation space to calculate the SDE offset density estimate; the original dominance benchmark and the SDE offset density estimate are fused to obtain the mixed fitness value; and adaptive truncation screening is performed on the individuals based on the mixed fitness value. Based on the hybrid fitness value, a dominant parent interaction set is constructed. Based on the multidimensional exponential discrete memristor chaotic mapping, a nonlinear chaotic perturbation sequence with global ergodicity is generated. A dual-mode mutation mechanism is introduced to execute either the "individual perturbation mode" or the "external elite guidance mode" with probabilistic adaptive execution. The chaotic attraction points and historical non-dominated solutions are sorted. Based on the sorted chaotic attraction points and historical non-dominated solutions, the initial scheduling decision matrix is ​​updated to obtain multiple scheduling decision matrices. From the approximate Pareto optimal scheduling decision matrix, the optimal scheduling decision scheme is selected from the scheduling decision schemes corresponding to multiple scheduling decision matrices based on the decision weights, wherein the decision weights are used to characterize the importance of the task observation benefits, the tracking accuracy, and the constellation system load balancing.

[0007] Secondly, this application provides a multi-objective scheduling device based on low-Earth orbit (LEO) satellites. The device includes a model building module, a generation module, a metric and solution set balancing module, a search and update module, and a decision module. The model building module is used to establish a three-dimensional visibility matrix of multiple satellites and multiple space targets at discrete time steps based on the LEO constellation distribution and satellite kinematics model. Based on the three-dimensional visibility matrix, it constructs a model to be optimized, wherein the optimization space of the model to be optimized integrates conflicting targets, including comprehensive observation benefits, tracking accuracy, and constellation system load balancing. The generation module is used to perform a discrete solution space random mapping that satisfies the rigid constraints of dual-satellite cooperation within the optimization space, generating an initial scheduling decision solution set population, and simultaneously generating an initial scheduling decision matrix corresponding to the initial scheduling decision solution set population. The metric and solution set balancing module is used to perform asymmetric multi-index performance measurement on the mixed individual solution set to be evaluated. Specifically, it counts the number of non-dominated individuals in the initial scheduling decision solution set population within the solution set to establish the original dominance benchmark, introduces an offset vector in the target evaluation space to calculate the SDE offset density estimate, and integrates the original dominance. The baseline and the estimated SDE offset density are used to obtain a hybrid fitness value. Based on the hybrid fitness value, adaptive truncation screening is performed on the individual. The search and update module is used to construct a dominant parent interaction set based on the hybrid fitness value, generate a nonlinear chaotic perturbation sequence with global ergodicity based on a multidimensional exponential discrete memristor chaotic map, introduce a dual-modal mutation mechanism, and probabilistically and adaptively execute either an "individual perturbation mode" or an "external elite-guided mode". The chaotic attraction points and historical non-dominated solutions are sorted, and the initial scheduling decision matrix is ​​updated based on the sorted chaotic attraction points and historical non-dominated solutions to obtain multiple scheduling decision matrices. The decision module is used to select the optimal scheduling decision scheme from the scheduling decision schemes corresponding to the multiple scheduling decision matrices based on decision weights from the approximately Pareto optimal scheduling decision matrices. The decision weights are used to characterize the importance of the task observation benefits, the tracking accuracy, and the constellation system load balancing.

[0008] Thirdly, embodiments of this application provide a readable storage medium storing a computer-readable program, which, when executed by a processor, is used to implement the method described in the first aspect of this application.

[0009] Compared with the prior art, the technical solution of this application has the following beneficial technical effects: This invention integrates mission observation benefits, tracking accuracy, and constellation system load balancing into the optimization space of the optimization model, and introduces a dual-satellite cooperative rigid constraint as a strongly coupled boundary of the optimization model. Based on this optimization model, it generates an initial scheduling decision matrix, performs multi-index performance measurement and solution set balancing based on non-dominated strength and offset density, conducts external elite-guided memristor chaotic topology search and updates the initial scheduling decision matrix, and generates the optimal scheduling decision scheme. This scheme, on the one hand, ensures the observation efficiency of the optimal scheduling decision scheme by using the comprehensive observation benefits index of the optimization space while maintaining tracking accuracy; on the other hand, it ensures that the optimal decision scheme achieves balance among multiple different satellite resources by using the constellation system load balancing index of the optimization space. This avoids the problem of decision schemes being difficult to execute due to ignoring the actual situation of satellite resources through theoretical analysis alone, thus improving the efficiency and executability of multi-objective scheduling methods based on low-Earth orbit satellites. Attached Figure Description

[0010] Figure 1 One of the schematic diagrams illustrating the workflow of a multi-target scheduling method based on low-Earth orbit satellites, provided for some embodiments of this application; Figure 2 A second schematic diagram illustrating the workflow of a multi-target scheduling method based on low-Earth orbit satellites, provided for some embodiments of this application; Figure 3 A schematic diagram showing the comparison between a multi-objective scheduling method based on low-Earth orbit satellites and other algorithms provided for some embodiments of this application; Figure 4 A scheduling instruction Gantt chart for a multi-objective scheduling method based on low-Earth orbit satellites, provided for some embodiments of this application; Figure 5 A schematic diagram of the structure of a multi-target scheduling device based on low-Earth orbit satellites is provided for some embodiments of this application; Figure 6 Schematic diagrams of electronic devices provided for some embodiments of this application. Detailed Implementation

[0011] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0012] The terms "at least one," "at least one of," etc., used in the specification and claims of this application refer to any one, any two, or a combination of two or more of the included items. For example, at least one of a, b, and c can mean: "a," "b," "c," "a and b," "a and c," "b and c," and "a, b, and c," where a, b, and c can be single or multiple. Similarly, "at least two" refers to two or more items, and its meaning is similar to that of "at least one."

[0013] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0014] In this invention, mission observation benefits, tracking accuracy, and constellation system load balancing are integrated into the optimization space of the optimization model. A dual-satellite cooperative rigid constraint is introduced as a strongly coupled boundary of the optimization model. An initial scheduling decision matrix is ​​then generated based on this model. Subsequently, multi-indicator performance metrics and solution set balancing are performed in sequence, followed by external elite-guided memristor chaotic topology search and initial scheduling decision matrix update. Finally, the optimal scheduling decision scheme is generated. Because the observation efficiency of the optimal scheduling decision scheme is guaranteed by the comprehensive observation benefits index of the optimization space while ensuring tracking accuracy, and the optimal decision scheme achieves balance among multiple different satellite resources by using the constellation system load balancing index of the optimization space, the problem of decision schemes being difficult to execute due to ignoring the actual situation of satellite resources through theoretical analysis is avoided. Therefore, the efficiency and executability of the multi-objective scheduling method based on low-Earth orbit satellites are improved.

[0015] Reference Figure 1 and Figure 2 The multi-target scheduling method based on low-Earth orbit satellites provided in this application embodiment may include the following steps 201 to 205: Step 201: Construct the model to be optimized: In some embodiments of this application, step 201 specifically includes: establishing a three-dimensional visibility matrix of multiple satellites and multiple targets in space at discrete time steps based on the low-Earth orbit constellation distribution and satellite kinematics model; constructing an optimization model based on the three-dimensional visibility matrix; wherein the optimization space of the optimization model includes conflicting targets, and the conflicting targets include comprehensive observation benefits, tracking accuracy, and constellation system load balance.

[0016] In some embodiments of this application, the aforementioned multiple satellites refer to multiple satellites in a low-Earth orbit.

[0017] In some embodiments of this application, the aforementioned multiple targets in space refer to multiple targets observed by satellites in low Earth orbit.

[0018] In some embodiments of this application, the step of constructing the model to be optimized based on the three-dimensional visibility matrix includes using the dual-star cooperative rigid constraint and the onboard payload capacity as the strongly coupled boundary of the model to be optimized to construct the model to be optimized.

[0019] In some embodiments of this application, the establishment of a three-dimensional visibility matrix for multiple satellites and multiple targets in space at discrete time steps based on the low-Earth orbit constellation distribution and satellite kinematics model may include the following steps 201a and 201b: Step 201a: Based on the low-Earth orbit constellation distribution model and satellite kinematics model, simulate and generate each discrete time step within the scheduling cycle. The first position vector and first velocity vector of all satellites in the geocentric inertial frame, and the second position vector and second velocity vector of all targets.

[0020] In some embodiments of this application, all of the above objectives are the above-described spatial multi-objectives.

[0021] In some embodiments of this application, the above-described low-Earth orbit constellation distribution model is used to characterize the satellite set in a low-Earth orbit constellation.

[0022] In some embodiments of this application, the above-described satellite kinematic model is used to characterize the kinematic laws of each of the above-described satellites.

[0023] In some embodiments of this application, the above objective is achieved at each discrete time step. The position vector and velocity vector in the geocentric inertial frame.

[0024] Step 202: Generate the initial scheduling decision matrix: In some embodiments of this application, step 202 specifically includes: based on the model to be optimized, performing a discrete solution space random mapping that satisfies the binary star cooperative rigid constraint in the space to be optimized, generating an initial scheduling decision solution set population, and simultaneously generating an initial scheduling decision matrix corresponding to the initial scheduling decision solution set population.

[0025] In some embodiments of this application, a convergence iteration limit of the optimization space is first set, and then, based on the convergence iteration limit, a discrete solution space random mapping that satisfies the binary star cooperative rigid constraint is performed in the optimization space.

[0026] In some embodiments of this application, the initial scheduling decision matrix is ​​isomorphic to the size of the initial scheduling decision solution set population and is used to retain non-dominant frontier solutions across generations.

[0027] In some embodiments of this application, the initial scheduling decision matrix described above is used to characterize the observation status of different satellites on each target at each discrete time step.

[0028] In some embodiments of this application, the above observation scenarios include: observation and no observation.

[0029] As an example, suppose a low-Earth orbit constellation contains If there are 10 satellites, then the set of satellites in the low Earth orbit constellation is 100. Suppose there are a total of space targets to be observed. If there are 1, then the set of spatial targets to be observed is 1. Suppose the scheduling period is discretized as If there are 12 equal time steps, then the set of discretized scheduling cycles is denoted as . Define the initial scheduling decision matrix. ,in, Indicates at time step Arrange satellites For the target Conduct observations. Indicates at time step No satellites will be deployed. For the target Observations were conducted, among which, For any item in the satellite set S, Let M be any item in the target set M of the set space. Let be any item in the set of satellite discretization scheduling cycles P.

[0030] Step 201b: Based on the first position vector, first velocity vector, second position vector, second velocity vector and satellite sensor parameters mentioned above, calculate the comprehensive visibility of each satellite to each target at each time step to form a three-dimensional visibility matrix.

[0031] In some embodiments of this application, the above-mentioned satellite sensor parameters include one or more of the following: field of view, maximum effective range, spatial geometric relationship, or other parameters, which are not limited in this application.

[0032] In some embodiments of this application, there are multiple sets of satellite sensor parameters, with one set of satellite sensor parameters corresponding to one satellite.

[0033] In some embodiments of this application, the three-dimensional visibility of each satellite to each target at each time step is calculated based on the above-mentioned satellite sensor parameters, and the comprehensive visibility is obtained based on the three-dimensional visibility.

[0034] In some embodiments of this application, the aforementioned three-dimensional visibility matrix includes a set of multiple integrated visibilitys, with different integrated visibilitys corresponding to different combinations of time steps, satellites, and targets.

[0035] In some embodiments of this application, the aforementioned three-dimensional visibility includes geometric visibility, optical visibility, and device visibility, with each comprehensive visibility corresponding to a geometric visibility, an optical visibility, and a device visibility.

[0036] In some embodiments of this application, the aforementioned geometric visibility is used to determine whether the line of sight between the satellite and the target is blocked by the Earth.

[0037] In some embodiments of this application, geometric visibility is obtained by calculating the perpendicular distance from the Earth's center to the line connecting the satellite and the target, comparing the perpendicular distance with the Earth's radius, and determining whether the line of sight between the satellite and the target is blocked by the Earth based on the set visibility.

[0038] As an example, in the ECI coordinate system, let the satellite... and target The position vectors are respectively and , point to The line-of-sight vector is From the Earth's core Towards the line of sight Or, draw a perpendicular line from its extension, and denote the foot of the perpendicular as . ,but and The angle between This can be expressed by the following formula 1: (Formula 1); Among them, the above Geometric visibility refers to the visibility of a target when the line of sight between the satellite and the target is not obstructed by the Earth. Formula 2 must be satisfied: (Formula 2); In some embodiments of this application, the aforementioned optical visibility is used to characterize whether the target is within the field of view of the sensor pointing towards the sun.

[0039] In some embodiments of this application, it is possible to simultaneously determine whether the angle between the line connecting the satellite and the target, and between the satellite and the sun, is greater than a preset threshold to determine whether the target is within the field of view of the sensor pointing to the sun.

[0040] As an example, in the ECI coordinate system, satellites are calculated using the following formula 3. Pointing to the target respectively unit vector of the sun and The angle between : (Formula 3); Among them, the above Optical visibility refers to the visibility of a target when the line of sight between the satellite and the target is not obstructed by the Earth. Satisfy the following formula 4: (Formula 4); Understandably, to avoid the influence of the sun as a strong background light source on the observation results, it is necessary to ensure that the target is not within the field of view of the sensor pointing towards the sun during observation.

[0041] In some embodiments of this application, the device visibility described above is used to determine whether the target is within the field of view of the satellite sensor.

[0042] Based on the above example, the detection range of the optical sensor can be quantized as a unit vector pointing from the optical axis. Azimuth Pitch angle and the effective distance For a commonly defined rectangular view frustum, the device is visible when the target is located within the view frustum. In this case, the above parameters must satisfy the following formula 5: (Formula 5); In formula 5 above, To be aligned with the optical axis Orthogonal unit vectors.

[0043] In some embodiments of this application, the aforementioned device visibility can be obtained through a device visibility model.

[0044] In some embodiments of this application, the aforementioned device visibility model is a set of all target locations in space that satisfy the frustum criterion. This device visibility model includes two modes: a ground-oriented mode and a near-edge mode. In the ground-oriented mode, the optical axis is perpendicular to the Earth's surface; in the near-edge mode, the optical axis is tangent to the Earth's radius.

[0045] Based on the above examples, for each satellite For the target In time step The overall visibility of a satellite is determined by sequentially assessing geometric visibility, optical visibility, and equipment visibility; if all three are deemed visible, then the satellite is considered to be visible. .

[0046] It should be noted that, in order to reduce the difficulty of solving the problem and to construct a feasible mathematical model, this application assumes that all satellites are functioning normally when constructing the above three-dimensional visibility matrix.

[0047] In some embodiments of this application, the aforementioned dual-satellite cooperative rigid constraint includes: for each observation mission, there are at least two satellites that simultaneously observe the target under visibility conditions.

[0048] In some embodiments of this application, each of the above observation tasks can be used to observe one target or multiple targets.

[0049] Understandably, for each observation mission, by requiring two satellites that meet the visibility requirements to simultaneously observe the target, positioning accuracy and mission reliability are improved.

[0050] In some embodiments of this application, the above-mentioned spaceborne payload capacity limitation includes: the spaceborne payload capacity during scheduling does not exceed the maximum number of simultaneous observation targets of the satellite.

[0051] Specifically, the above-mentioned satellite-borne payload capacity not exceeding the satellite's maximum number of simultaneously observed targets can be expressed by the following formula 6: (Formula 6); In formula 6, Representative satellite The maximum number of targets that can be observed simultaneously.

[0052] In some embodiments of this application, the above-mentioned observation comprehensive benefit is used to characterize the total value of the task that successfully completes the observation within the scheduling period.

[0053] In some embodiments of this application, the above-mentioned model to be optimized is used to perform multiple objective optimization tasks, including: maximizing the overall observation benefit, maximizing the tracking accuracy, and maximizing the constellation system load balancing.

[0054] In some embodiments of this application, the above-mentioned observation comprehensive benefit is used to characterize the total value of the task that successfully completes the observation within the scheduling period.

[0055] For example, let the overall observation benefit be... The total value of tasks that successfully complete observations within the scheduling period can be calculated using the following formula 7: (Formula 7); In formula 7, The goal At time step The normalized task weight is determined by multiple factors, including but not limited to: task priority and target threat level. It is a task completion indicator function, when the target At time step When observed by two satellites, It is 1 if it is true, otherwise it is 0.

[0056] In some embodiments of this application, the above-mentioned tracking accuracy is used to characterize the geometric positioning accuracy of binary star observation.

[0057] As an example, let the tracking accuracy be... ,but It can be calculated using the following formula 8: (Formula 8); In formula 8, It refers to all successfully executed binary satellite observation missions in the scheduling scheme, i.e. The total number of tasks. It is the first In a dual-satellite observation mission, the angle between the line-of-sight vectors of the two observation satellites relative to the target is ideally 90 degrees. It can be calculated using the position vectors of the satellite and the target. The value range is [0,1]. The larger the value, the better the average observation geometry and the higher the positioning accuracy.

[0058] In some embodiments of this application, the aforementioned constellation system load balancing degree is used to characterize the degree of balance of the workload of each satellite.

[0059] As an example, suppose the load balancing degree of the constellation system is... Then, it can be calculated using the following formula 9: (Formula 9); In formula 9, It is a satellite Total number of observations during the entire scheduling cycle. It is the average load of all satellites. The value range is (0,1). When all satellites have identical payloads, the variance is zero. The greater the load difference, The smaller.

[0060] Understandably, maximizing the load balance of the constellation system can prevent some satellites from being overused while others remain idle, which is beneficial for the long-term operation of the constellation and the rational use of resources.

[0061] In some embodiments of this application, the above-described model to be optimized can be used to handle the multi-objective optimization problem in Equation 10 below: (Formula 10); In some embodiments of this application, the above-mentioned model to be optimized is used to perform the above-mentioned multiple objective optimization tasks by a hybrid multi-objective chaotic evolutionary optimization algorithm.

[0062] In some embodiments of this application, the parameters of the above-described hybrid multi-objective chaotic evolutionary optimization algorithm include population size. External elite save file size limit Maximum number of function evaluations Or other parameters, the embodiments of this application do not limit this.

[0063] In some embodiments of this application, step 202 specifically includes steps 202a, 202b, and 202c: Step 202a: Determine the required number of observation tasks, generate an observation value sequence based on the number of observation tasks, and set the initial scheduling decision solution set population according to the number of observation tasks. population size ; Step 202b: Configure storage An external archive of elite Pareto solutions; As an example, it can be based on the three-dimensional visibility matrix Randomly generate an initial feasible scheduling decision solution set population that satisfies the dual-satellite collaborative observation constraint and the load capacity constraint. .

[0064] Step 202c: Based on the above initial scheduling decision solution set population Obtain the initial scheduling decision matrix .

[0065] In some embodiments of this application, in the above-mentioned initial scheduling decision matrix middle, Indicates at time step Arrange satellites For the target Conduct observations. This indicates that the observation will not be scheduled.

[0066] In some embodiments of this application, step 202 above further includes calculating the population of the initial scheduling decision solution set. The three objective function values ​​for each individual. Initialize an empty external elite archive. .

[0067] Step 203: Multi-index performance measurement and solution set balancing of non-dominated intensity and offset density: In some embodiments of this application, step 203 specifically includes: performing an asymmetric multi-index performance measurement on the mixed individual solution set to be evaluated.

[0068] Specifically, the above-mentioned asymmetric multi-index performance measurement of the mixed individual solution set to be evaluated includes: counting the number of non-dominated individuals in the initial scheduling decision solution set population to establish the original dominance benchmark; introducing an offset vector in the target evaluation space to calculate the SDE offset density estimate; fusing the original dominance benchmark and the SDE offset density estimate to obtain the mixed fitness value; and performing adaptive truncation screening on the individuals based on the mixed fitness value.

[0069] In some embodiments of this application, step 203 specifically includes: selecting individuals with a mixed fitness value less than the mixed fitness threshold from the initial scheduling decision solution set population to obtain an initial candidate solution set; iterating the initial candidate solution set based on the population size N to obtain a new generation population.

[0070] In some embodiments of this application, the above-mentioned hybrid fitness threshold can be 1 or other values, and this application does not limit this.

[0071] In some embodiments of this application, step 203 can be specifically implemented through the following steps 203a, 203b, 203c, 203d, 203e and 203f: Step 203a: Calculate individuals Intensity value That is, the number of other individuals it dominates in the current merged population.

[0072] Step 203b: Based on the above strength values Calculate individual primitive fitness That is, all domination The sum of the intensity values ​​of the individuals.

[0073] Specifically, the aforementioned initial fitness It can be calculated using the following formula 11: (Formula 11); Step 203c: Calculate individuals Offset density estimate .

[0074] In some embodiments of this application, the above... This is a density estimate based on the offset distance.

[0075] In some embodiments of this application, the above... It can be calculated using the following formula 12: (Formula 12); in, , representing the individual being calculated Target vector and offset vector The Euclidean distance between them, for each individual Calculate its relationship with the merged population All other individuals of The SDE distances of the individual individuals are then sorted in ascending order to obtain an ordered sequence. and select By using the minimum distance as a reference, the density estimate can be obtained.

[0076] It is understandable that the above Used to measure the solution The smaller the value of its sparsity in its neighborhood, the better the sparsity of the above-mentioned... The higher the quality.

[0077] Step 203d: Calculation The mixed fitness value is obtained. This is used to map the solution from the decision space to the evaluation space; the smaller its value, the higher the quality of the solution.

[0078] In some embodiments of this application, it is possible to [do something related to the current situation]. Generation population and its offspring population The mixed population formed by merging For each individual in the dataset, calculate its mixed fitness value. .

[0079] Step 203e: From the initial scheduling decision solution set population Selecting mixed fitness values Individuals whose fitness level is below the above-mentioned mixed fitness threshold.

[0080] Understandably, this initial candidate solution set It is a high-quality population obtained after truncation and screening.

[0081] Step 203f: Based on the population size N, the above... Through iteration, a new generation of population is obtained. .

[0082] like equal to the preset population size This would lead to the next generation of the species ;like Less than Then, select the individual with the smallest mixed fitness value from the remaining individuals to join. until its size equals ,constitute ;like Greater than Then iteratively from Remove the individual with the smallest current SDE distance (i.e., the densest distribution) until its size equals [the value of the SDE]. To form a new generation of population .

[0083] Step 204: External elite-guided memristor chaotic topology search and initial scheduling decision matrix update: In some embodiments of this application, step 204 specifically includes: constructing a dominant parent interaction set based on the mixed fitness value; generating a nonlinear chaotic perturbation sequence with global ergodicity based on a multidimensional exponential discrete memristor chaotic mapping; introducing a dual-modal mutation mechanism; probabilistically adaptively executing an "individual perturbation mode" or an "external elite-guided mode"; sorting the chaotic attraction points and historical non-dominated solutions; and updating the initial scheduling decision matrix based on the sorted chaotic attraction points and historical non-dominated solutions to obtain multiple scheduling decision matrix columns.

[0084] In some embodiments of this application, step 204 can be specifically implemented through the following steps S204a, S204b, S204c, S204d, and S204e: S204a: Based on an elite solution-guided mutation mechanism, it performs a memristor chaotic topology search task.

[0085] As an example, the formula for performing the memristor chaotic topology search task is as follows: ,in, For the mutation vector, The scaling factor is a random factor that is uniformly distributed within the interval [0,1]. The chaotic attraction point obtained after mapping the chaotic sequence back to the decision space. The parent individual; the basis vector The value ranges include: Individual guidance model: At this point, the mutation is centered on the parent individual; Elite Guided Mode: ,in It is an elite solution that is randomly selected uniformly from the current external elite archive.

[0086] S204b: The next generation of population Add the current external elite archive to form a new merged solution set; S204c: Perform fast non-dominated sorting on the newly merged solution set to identify non-dominated solutions; S204d: Retain the first level of non-dominated solutions after non-dominated sorting, and remove duplicate solutions to form a new candidate archive; S204e: Based on the preset maximum archive size The new candidate archive is updated to obtain multiple scheduling decision matrices.

[0087] In some embodiments of this application, the aforementioned external elite archive corresponds to the aforementioned initial scheduling decision matrix.

[0088] In some embodiments of this application, it can be determined whether the size of the candidate archive exceeds a preset archive size limit. To update the above new candidate archive.

[0089] In some embodiments of this application, the above-described determination of whether the size of the candidate archive exceeds a preset archive size upper limit is performed. ,include: If the size of the candidate archive is less than or equal to If so, the candidate archive will be directly used as the updated external elite archive; If the size of the candidate archive is greater than Then, a truncation strategy based on offset density estimation distance is adopted to calculate the offset density estimation distance from each individual in the candidate archive to its nearest neighbor individual; Iteratively remove the individuals that contribute the least to the distribution of the solution set (i.e., the closest individuals) from the candidate archive. After each removal, recalculate the offset density estimate distance of the remaining individuals until its size equals the distance of the remaining individuals. The truncated solution set is used as the updated initial scheduling decision matrix.

[0090] In some embodiments of this application, a binary tournament selection mechanism is employed, using a base-calculated mixed fitness value to select from the merged population. Select dominant individuals to form a mating pool Subsequently, a chaotic sequence is generated using a two-dimensional exponential discrete memristor chaotic map, introducing randomness and ergodicity into the evolutionary process. This map can be calculated using the following formula 13: (Formula 13); Next, from the mating pool Randomly select parent individuals and The evolutionary operation is performed as follows: for the parent individual Randomly select a point from its corresponding set of chaotic attraction points. Then, a mutation strategy based on external elite archives is adopted: using probability... Elite solutions from external archives are selected as mutation basis vectors; otherwise, the individual itself is used as the basis vector, and a perturbation term guided by a chaotic sequence is superimposed to generate an experimental vector. This strategy dynamically balances global exploration and local exploitation capabilities. The formula for calculating this experimental vector is shown in Formula 14. (Formula 14); In formula 13, It is a scaling factor that controls the search step size. In individual-guided mode, it refers to the basis vectors. The mutations centered on the parent generation, perturbing towards the chaotic attraction point. This model emphasizes local exploration, aiming to conduct a fine search of the neighborhood where the parent individual is located, which helps to uncover potential solution space characteristics, maintain population diversity, and avoid premature convergence.

[0091] In some embodiments of this application, vectors are used in elite-guided mode. ,in From the current external elite archive An elite solution was selected from the pool. The algorithm dynamically maintains the set of non-dominated solutions discovered so far and selects elite individuals from it at a uniform and random rate as guides. In essence, it directly positions the initial point of subsequent searches near the known high-quality solution region, thereby effectively guiding the population to converge faster toward the true Pareto front.

[0092] Further, alternatively, the mutation vector can be modified through a crossover operation. and parental individuals Perform binomial crossover to generate test vectors. For the d-th dimension decision variable, corresponding components This leads to the generation of offspring populations. , The calculation process is shown in Formula 15: (Formula 15); In the formula, Represents the mutation vector The d-th component; Representing the parent generation The d-th component; The crossover probability is generated randomly.

[0093] Further, alternatively, the parent population and its offspring population merged into Based on mixed fitness values Environmental selection strategies are used to select the next generation of populations. The specific steps are as follows: 1) If Then directly order ; 2) If This indicates that there are not enough high-quality solutions. The individual with the lowest fitness value from the remaining individuals should be selected and added. Together constitute ; 3) If This indicates an overabundance of high-quality solutions, requiring selection based on distribution. In each iteration, the following calculations are performed. Calculate the SDE distance from each individual to all other individuals in the set, sort these distances in ascending order, and determine and remove the individual with the densest current distribution. Repeat this process until... Ultimately, it was designated as the new generation population. .

[0094] Understandably, this strategy prioritizes retaining high-quality non-dominated solutions with fitness values ​​less than the mixed fitness threshold. If the number exceeds the population size, it iteratively removes individuals with the smallest offset density estimation distance, i.e., individuals in the most crowded region, thereby maintaining the population size and ensuring the uniformity of the solution set distribution.

[0095] Further, alternatively, the new generation of population Non-dominated solutions and current external elite save files Merge, perform fast non-dominated sorting, retain only the first level of non-dominated solutions, whose upper limit is . Update according to the following steps to obtain an external elite archive with good convergence and uniform distribution.

[0096] As an example, the above-mentioned external elite archive with good convergence and uniform distribution can be achieved in the following way: 1) Candidate set generation: The new population selected by the environment will be generated. Add to the current save file to form a new merged class. ; 2) Quick Non-Dominated Sort, for the new merged solution set All individuals are sorted using a fast non-dominated sort, and only the first-level non-dominated solutions are retained to form a new candidate archive. ; 3) Remove duplicate solutions. Repeated solutions with identical decision vectors; 4) Truncation operation, if If the environment selection fails, a truncation operation is required. Truncation strategies typically prioritize maintaining the distribution of the solution set over environment selection. A truncation method based on offset distance (SDE) iteratively removes individuals that contribute the least to the distribution of the solution set until the archive size equals [the desired size]. Then update the save file, making .

[0097] Alternatively, the above steps can be repeated until a preset termination condition is met. After the algorithm terminates, an external elite archive is output. As the obtained approximate Pareto optimal solution set, each solution in the solution set corresponds to a scheduling decision scheme that achieves different trade-offs among the three objectives of mission observation benefit, tracking accuracy and constellation system load balance. Each scheduling decision scheme corresponds to a scheduling decision matrix.

[0098] Step 205: Generate the optimal scheduling decision scheme: In some embodiments of this application, step 205 includes: obtaining multiple decision schemes based on the multiple decision matrices; selecting the optimal scheduling decision scheme from the scheduling decision schemes corresponding to the multiple scheduling decision matrices from the near Pareto optimal scheduling decision matrices based on preset decision weights, wherein the decision weights are used to characterize the importance of the task observation benefits, the tracking accuracy, and the constellation system load balancing.

[0099] Alternatively, decision-makers can set a set of normalized decision weight vectors for mission observation benefits, tracking accuracy, and constellation system load balancing, based on the specific needs of the current mission phase. For example, in emergency situations, priority should be given to ensuring the observation and tracking accuracy of high-value targets, while in normal times, more attention should be paid to the balanced operation of the system. ,in .

[0100] In some embodiments of this application, for each scheduling scheme in the Pareto solution set, its comprehensive score is calculated. The scheduling decision scheme corresponding to the scheduling decision matrix with the highest comprehensive score is selected as the final scheduling scheme to be executed.

[0101] In some embodiments of this application, the final scheduling decision scheme is represented by its corresponding scheduling decision matrix, which is a three-dimensional scheduling instruction matrix. ,in, Indicates at time step Arrange satellites For the target Observation should be conducted, otherwise This indicates that the observation will not be scheduled. This command can be directly issued to the satellite constellation's ground control system or the onboard autonomous mission planning module for execution.

[0102] In some embodiments of this application, step 205 further includes: when the current iteration count reaches a preset termination condition, outputting the optimal scheduling decision matrix as an approximate Pareto optimal scheduling scheme set; otherwise, incrementing the current iteration count by 1 and returning to step 203. Finally, based on the preset decision weight vector... ,in Calculate the overall score of the scheme. The overall scores of all schemes are compared, and the scheme with the highest score is selected as the optimal scheduling decision scheme.

[0103] In some embodiments of this application, the decision weight vector can be a preset weight vector or a decision weight vector input in real time.

[0104] Understandably, by inputting decision weight vectors in real time, users can perform multi-precision weighted energy measurements on multiple scheduling decision schemes based on actual task preferences, which is beneficial for generating an engineering scheduling instruction matrix that can be directly issued and executed by satellites.

[0105] As an example, this embodiment describes a scenario where a Walker-δ constellation containing 60 satellites schedules 50 space targets for 2 hours. The software implementation of this embodiment is based on the MATLAB platform.

[0106] In the scenario described above, the satellite constellation uses six orbital planes, with 10 satellites in each plane, at an altitude of approximately 1500 km and an inclination of 68°, forming a Walker-δ constellation. There are 50 space targets with orbital altitudes ranging from 500 km to 900 km and random inclinations. The scheduling cycle is 2 hours, with a time step of 60 seconds. One time step. Satellite sensor parameters are set as follows: azimuth and elevation angles. Satellite payload capacity This means that each satellite can track a maximum of two targets at any given time. (Task weight) exist Randomly generated within an interval to simulate different priorities for different objectives.

[0107] In the algorithm parameter settings of the above example, the population size External archive size Maximum number of function evaluations Elite guidance probability in chaotic mutation Crossover probability .

[0108] Next, a multi-objective chaotic evolutionary optimization algorithm is run. After the algorithm finishes running, a non-dominated Pareto solution is obtained, and the solution set has good distribution across all three objective dimensions. Assume the current task has weights set. After calculation, the scheduling decision matrix with the highest comprehensive score is selected, and the corresponding scheduling decision scheme set is the optimal scheduling decision scheme. The objective function value of this optimal scheme is compared with other algorithms as follows: Figure 3 As shown, the scheduling decision matrix is ​​decoded to generate the final scheduling instruction Gantt chart. Figure 4 As shown.

[0109] pass Figure 4 It can be seen that the method of the present invention can effectively handle the low-orbit satellite collaborative observation scheduling problem with multiple constraints and multiple objectives. The generated Pareto solution set provides rich optimization trade-offs for mission command. The final output scheduling scheme has achieved good results in terms of mission benefits, tracking accuracy and system balance.

[0110] The multi-objective scheduling method based on low-Earth orbit (LEO) satellites provided in this application integrates mission observation benefit, tracking accuracy, and constellation system load balancing into the optimization space of the optimization model. It introduces a dual-satellite cooperative rigid constraint as a strongly coupled boundary of the optimization model, and then generates an initial scheduling decision matrix based on this model. Next, it sequentially performs multi-index performance measurement and solution set balancing based on non-dominated strength and offset density, external elite-guided memristor chaotic topology search, and updates the initial scheduling decision matrix, finally generating the optimal scheduling decision scheme. Because it can guarantee the observation efficiency of the optimal scheduling decision scheme by using the comprehensive observation benefit index (the optimization space index) while ensuring tracking accuracy, and ensure that the optimal decision scheme achieves balance among multiple different satellite resources by using the constellation system load balancing index (the optimization space index), it avoids the problem of decision schemes being difficult to execute due to theoretical analysis neglecting the actual situation of satellite resources. Therefore, it improves the efficiency and executability of the multi-objective scheduling method based on LEO satellites.

[0111] It should be noted that the above-described method embodiments, or the various possible implementations of the method embodiments, can be executed individually, or, provided there are no contradictions, they can be combined with each other. The specific implementation can be determined according to actual usage requirements, and this application embodiment does not impose any restrictions on this.

[0112] The multi-target scheduling method based on low-Earth orbit (LEO) satellites provided in this application can be executed by a multi-target scheduling device based on LEO satellites. This application uses an example of a multi-target scheduling device based on LEO satellites executing the multi-target scheduling method based on LEO satellites to illustrate the multi-target scheduling device based on LEO satellites provided in this application.

[0113] Figure 5 The above embodiments illustrate a multi-objective scheduling device 800 based on low-Earth orbit satellites provided in this application. The multi-objective scheduling device 800 based on low-Earth orbit satellites may include: a model building module 801, a generation module 802, a metric and solution set balancing module 803, a search and update module 804, and a decision module 805. The model construction module 801 is used to establish a three-dimensional visibility matrix of multiple satellites and multiple space targets at discrete time steps based on the low-Earth orbit constellation distribution and satellite kinematics model, and to construct an optimization model based on the three-dimensional visibility matrix. The optimization space of the optimization model integrates conflicting targets, including comprehensive observation benefits, tracking accuracy, and constellation system load balancing. The generation module 802 is used to perform a discrete solution space random mapping that satisfies the rigid constraints of dual-satellite cooperation within the optimization space based on the optimization model, generating an initial scheduling decision solution set population, and simultaneously generating an initial scheduling decision matrix corresponding to the initial scheduling decision solution set population. The measurement and solution set balancing module 803 is used to perform asymmetric multi-index performance measurement on the solutions in the initial scheduling decision solution set population: statistically analyzing the number of non-dominated individuals in the initial scheduling decision solution set population to establish the original dominance benchmark, introducing offset vectors in the target evaluation space to calculate the SDE offset density estimate, and integrating... The original dominance benchmark and the estimated SDE offset density are combined to obtain a hybrid fitness value. Based on this hybrid fitness value, adaptive truncation screening is performed on the individual. The search and update module 804 is used to construct a dominant parent interaction set based on the hybrid fitness value, generate a nonlinear chaotic perturbation sequence with global ergodicity based on a multidimensional exponential discrete memristor chaotic map, introduce a dual-modal mutation mechanism, and probabilistically adaptively execute either an "individual perturbation mode" or an "external elite-guided mode." The chaotic attraction points and historical non-dominated solutions are sorted, and the initial scheduling decision matrix is ​​updated based on the sorted chaotic attraction points and historical non-dominated solutions to obtain multiple scheduling decision matrices. The decision module 805 is used to select the optimal scheduling decision scheme from the approximately Pareto optimal scheduling decision matrices based on decision weights, where the decision weights characterize the importance of the task observation gain, the tracking accuracy, and the constellation system load balancing.

[0114] Optionally, such as Figure 6 As shown, this application embodiment also provides an electronic device 900, including a processor 901 and a memory 902. The memory 902 stores a program or instructions that can run on the processor 901. When the program or instructions are executed by the processor 901, they implement the various steps of the above-described multi-target scheduling method embodiment based on low-orbit satellites and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0115] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described multi-target scheduling method embodiment based on low-orbit satellites and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0116] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0117] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described gesture recognition method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0118] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0119] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0120] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described gesture recognition method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0121] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0122] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0124] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. The substitutions may be replacements of some structures, devices, or method steps, or they may be complete technical solutions. Equivalent substitutions or modifications made to the technical solutions and inventive concepts of the present invention should all be covered within the scope of protection of the present invention.

Claims

1. A multi-objective scheduling method based on low-Earth orbit satellites, characterized in that, include: S1. Construct the matrix L to be optimized: Based on the low-Earth orbit constellation distribution and satellite kinematics model, a three-dimensional visibility matrix of multiple satellites and multiple targets in space is established at discrete time steps. Based on the three-dimensional visibility matrix, an optimization model is constructed. The optimization space of the optimization model includes conflicting targets, including comprehensive observation benefits, tracking accuracy, and constellation system load balance. S2. Generate the initial scheduling decision matrix: Based on the model to be optimized, a discrete solution space random mapping that satisfies the rigid constraint of binary star cooperation is performed in the space to be optimized to generate an initial scheduling decision solution set population, and an initial scheduling decision matrix corresponding to the initial scheduling decision solution set population is generated simultaneously. S3. Multi-index performance measurement and solution set balance of non-dominated intensity and offset density: Asymmetric multi-index performance measurement is performed on the solutions in the initial scheduling decision solution set population: the number of non-dominated individuals in the initial scheduling decision solution set population is counted to establish the original dominance benchmark; an offset vector is introduced into the target evaluation space to calculate the SDE offset density estimate; the original dominance benchmark and the SDE offset density estimate are fused to obtain the mixed fitness value; and adaptive truncation screening is performed on the individuals based on the mixed fitness value. S4. Externally guided memristor chaotic topology search and initial scheduling decision matrix update: Based on the hybrid fitness value, a set of dominant parent interactions is constructed. Based on the multidimensional exponential discrete memristor chaotic mapping, a nonlinear chaotic perturbation sequence with global ergodicity is generated to update the initial scheduling decision matrix, resulting in multiple scheduling decision matrices. Step S5: Generate the optimal scheduling decision scheme: From the approximate Pareto optimal scheduling decision matrix, the optimal scheduling decision scheme is selected from the scheduling decision schemes corresponding to the multiple scheduling decision matrices based on the decision weights, wherein the decision weights are used to characterize the importance of the task observation benefits, the tracking accuracy, and the constellation system load balancing.

2. The method according to claim 1, characterized in that, The method for establishing a three-dimensional visibility matrix for multiple satellites and multiple targets in space at discrete time steps, based on the low-Earth orbit constellation distribution and satellite kinematics model, includes: Based on the low-Earth orbit constellation distribution model and satellite kinematics model, simulations generate each discrete time step within the scheduling cycle. The first position vector and first velocity vector of all satellites in the geocentric inertial frame, and the second position vector and second velocity vector of all targets; Based on the first position vector, first velocity vector, second position vector, second velocity vector and satellite sensor parameters mentioned above, the comprehensive visibility of each satellite to each target at each time step is calculated to form a three-dimensional visibility matrix; The overall visibility is based on the geometric visibility, optical visibility, and equipment visibility of each satellite for each target.

3. The method according to any one of claims 1 or 2, characterized in that, The comprehensive observation benefit is used to characterize the total value of tasks that successfully complete observations within a scheduling cycle. The tracking accuracy is used to characterize the geometric positioning accuracy of binary star observations; The constellation system load balancing degree is used to characterize the degree of balance in the workload of each satellite.

4. The method according to claim 1, characterized in that, Step S2 includes: Determine the required number of observation tasks, generate an observation value sequence based on the number of observation tasks, and set the initial scheduling decision solution set population according to the number of observation tasks. population size ; Configure for storage An external archive of elite Pareto solutions; Based on the initial scheduling decision solution set population Obtain the initial scheduling decision matrix .

5. The method according to claim 4, characterized in that, Step S3 includes: Calculate individuals Intensity value Based on the intensity value Calculate the individual primitive fitness ; Calculate the individual Offset density estimate ; Based on the above and stated Obtain the mixed fitness value ; The population is resolved from the initial scheduling decision. Selecting mixed fitness values Individuals with fitness values ​​below the mixed fitness threshold are used to obtain the initial candidate solution set. ; Based on the population size N, the above Through iteration, a new generation of population is obtained. .

6. The method according to claim 5, characterized in that, The process of updating the initial scheduling decision matrix by generating a nonlinear chaotic perturbation sequence with global ergodicity based on a multidimensional exponential discrete memristor chaotic map includes: A bimodal mutation mechanism is introduced to probabilistically adapt to either the "individual perturbation mode" or the "external elite guidance mode," sorting the chaotic attraction points and historical non-dominated solutions, and updating the initial scheduling decision matrix based on the sorted chaotic attraction points and historical non-dominated solutions.

7. The method according to claim 6, characterized in that, Step S4 further includes: Based on the mutation mechanism guided by elite solutions, a memristor chaotic topology search task is performed. The new generation population Add the current external elite archive to form a new merged solution set; Perform a fast non-dominated sort on the newly merged solution set to identify the non-dominated solutions; The first-level non-dominated solution after the fast non-dominated sort is retained, and duplicate solutions are removed to form a new candidate archive. Based on the preset maximum archive size The new candidate archive is updated to obtain the multiple scheduling decision matrices.

8. The method according to claim 7, characterized in that, The step of selecting the optimal scheduling decision scheme from the scheduling decision schemes corresponding to multiple scheduling decision matrices based on decision weights from the approximately Pareto optimal scheduling decision matrix includes: Define a set of normalized decision weight vectors for mission observation gains, tracking accuracy, and constellation system load balancing. ; For each of the multiple scheduling decision schemes, calculate its comprehensive score, and select the scheduling decision scheme corresponding to the scheduling decision matrix with the highest comprehensive score as the final scheduling decision scheme to be executed.

9. A multi-target scheduling device based on low-Earth orbit satellites, characterized in that, include: The model building module is used to establish a three-dimensional visibility matrix of multiple satellites and multiple targets in space at discrete time steps based on the low-Earth orbit constellation distribution and satellite kinematics model. Based on the three-dimensional visibility matrix, a model to be optimized is constructed. The model to be optimized integrates conflicting targets in the space to be optimized, including observational comprehensive benefits, tracking accuracy, and constellation system load balance. The generation module is used to perform a discrete solution space random mapping that satisfies the rigid constraint of binary star cooperation within the optimization space based on the model to be optimized, generate an initial scheduling decision solution set population, and simultaneously generate an initial scheduling decision matrix corresponding to the initial scheduling decision solution set population. The measurement and solution set balancing module is used to perform asymmetric multi-index performance measurement on the solution set in the initial scheduling decision solution set population: count the number of non-dominated individuals in the solution set to establish the original dominance benchmark, introduce an offset vector in the target evaluation space to calculate the SDE offset density estimate, fuse the original dominance benchmark and the SDE offset density estimate to obtain the mixed fitness value, and perform adaptive truncation screening on the individuals based on the mixed fitness value; The search and update module is used to construct a set of dominant parent interactions based on the mixed fitness value, generate a nonlinear chaotic perturbation sequence with global ergodicity based on the multidimensional exponential discrete memristor chaotic mapping, introduce a dual-modal mutation mechanism, and execute the "individual perturbation mode" or "external elite guidance mode" with probabilistic adaptive execution, sort the chaotic attraction points and historical non-dominated solutions, and update the initial scheduling decision matrix based on the sorted chaotic attraction points and historical non-dominated solutions to obtain multiple scheduling decision matrices. The decision module is used to select the optimal scheduling decision scheme from the scheduling decision schemes corresponding to multiple scheduling decision matrices based on decision weights from the near Pareto optimal scheduling decision matrix. The decision weights are used to characterize the importance of the task observation benefits, the tracking accuracy, and the constellation system load balancing.

10. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by a processor, they are used to implement the method as described in any one of claims 1-7.

Images