Multi-satellite task allocation method based on minimum weight vertex coverage of conflict graph

Abstract

The invention relates to a multi-satellite task allocation method based on minimum weight vertex coverage of a conflict graph, and designs a centralized optimization algorithm by utilizing minimum weight vertex coverage and neighborhood search in a graph theory aiming at a multi-satellite task allocation problem with filtering constraint, pairwise constraint and accumulation constraint. A feasible observation window of a satellite is regarded as a vertex, pairwise constraint conflicts are regarded as edges, observation benefits are regarded as vertex weights, a pairwise constraint conflict graph is constructed, and an original problem is converted into iterative optimization solution containing conflict graph vertex coverage and accumulation constraint satisfaction; on the basis of a neighborhood search technology, a minimum weight vertex coverage solving algorithm and an accumulative constraint resolution operator are designed, and rapid calculation of a task allocation scheme can be effectively guaranteed.

Description

Technical Field

[0001] The invention relates to the field of multi-satellite task allocation, and in particular to a multi-satellite task allocation method based on minimum weight vertex coverage of a conflict graph. Background Art

[0002] With the growing demand for earth observation, the space satellite remote sensing operation paradigm has undergone a transformation from a single large spacecraft to a distributed satellite constellation. A group of satellite platforms operating in different orbits are equipped with heterogeneous remote sensing equipment to collect specific observation target information based on customer observation requests. Distributed remote sensing systems bring significant advantages, including shorter revisit cycles, larger coverage, better robustness, and greater flexibility to environmental and request changes, but they also bring many challenges to task allocation and scheduling.

[0003] Task allocation and scheduling aims to assign appropriate observation task sequences to a group of remote sensing satellites, maximize global evaluation indicators under the premise of complying with a series of pairwise constraints and cumulative constraints, and maximize the utilization of limited resources. Since this problem is essentially an NP-hard combinatorial optimization with complex aerospace constraints, centralized optimization methods provide a powerful solution. Among them, the centralized control node obtains the information of all individual satellites through network communication, and calculates feasible solutions or global approximate optimal solutions according to centralized optimization or search strategies. For example, The observation scheduling problem of agile satellites is studied, modeled as a constraint satisfaction problem (CSP) and the performance of typical solvers is compared. Although this method can well describe the constraints and evaluation indicators, different constraint satisfaction models need to be constructed in different specific mission scenarios. Gabrel focuses on the task allocation of a single earth observation satellite, describing it as a multi-criteria path process of finding a loop-free path in a directed graph, and proposes a two-stage algorithm to calculate a satisfactory global task allocation scheme. Although these graphs can appropriately represent the timing and conflict relationships between multiple observations, existing models have difficulties in describing the constraints and optimization objectives between multiple satellites. In addition, satellite task allocation has been simplified to various classical problems. For example, some scholars have proposed a 0-1 knapsack model for satellite task scheduling. Barkaouia and Berger proposed a VRPTW-inspired hybrid genetic algorithm with the help of the vehicle routing problem with time windows (VRPTW). Although many methods can be used for modeling and algorithm design, the scalability of existing work is poor and it is difficult to cope with more complex observation scenarios. In addition, although meta-heuristic swarm intelligence has made some improvements in dealing with NP-hard problems, including evolutionary algorithms, ant colony algorithms, particle swarm algorithms, etc., most existing methods only include simple or even naive generalizations of basic operators, failing to fully utilize the problem structure and related domain knowledge, resulting in disadvantages such as time-consuming computation and poor optimization performance when solving large-scale satellite task allocation problems.

[0004] Minimum Weight Vertex Cover (MWVC) is a classic NP-hard combinatorial optimization problem in graph theory. Consider any given simple undirected graph G = {V, E}, where each node v i With positive weight ω i . If the point set S contains at least one vertex of all edges, it is called a vertex cover. The goal of this problem is to find a vertex cover solution with the smallest sum of weights. If each feasible observation window is regarded as a vertex, the pairwise constraints between feasible observation windows are regarded as undirected edges, and the observation benefits of the tasks corresponding to the feasible observation windows are regarded as vertex weights, a global pairwise constraint conflict graph can be constructed and a task allocation solution that satisfies the pairwise constraints can be obtained through the calculation of MWVC.

[0005] In summary, the model representation methods used by most existing satellite task allocation methods have difficulties in describing multi-satellite constraints and optimization objectives, have poor scalability, and are difficult to cope with complex observation scenarios; at the same time, the optimization algorithms used fail to fully utilize the problem structure and related domain knowledge, have disadvantages such as time-consuming calculations and poor optimization performance, and cannot be applied to solving large-scale constellation task allocation problems. The MWVC problem provides a new perspective for modeling and algorithm design of satellite task allocation, especially for large-scale constellations. Summary of the invention

[0006] In order to solve the technical problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a multi-star task allocation method based on minimum weight vertex cover of conflict graph to solve the problem of large-scale star cluster task allocation.

[0007] To achieve the above-mentioned object of the invention, the present invention provides a multi-satellite task allocation method based on minimum weight vertex cover of conflict graph, comprising the following steps:

[0008] Step S1: Calculate the visible window set of each satellite in the planning period according to the observation meta-task attributes and requirements, satellite parameters and platform / payload constraints;

[0009] Step S2, considering filtering constraints, pairwise constraints and cumulative constraints, constructing a constrained global optimization model for multi-satellite task allocation;

[0010] Step S3, considering feasible observation windows as vertices, pairwise constraints between feasible observation windows as undirected edges, and observation benefits of tasks corresponding to visible windows as weights, to construct a global pairwise constraint conflict graph;

[0011] Step S4, dividing the conflict graph into an actual conflict subgraph and a non-conflict subgraph, initializing the vertex cover solution of the actual conflict subgraph, and converting it into a global task allocation solution;

[0012] Step S5: using the cumulative constraint satisfaction operator, generate an initial global feasible allocation scheme that satisfies the cumulative constraints, and define it as the current optimal solution;

[0013] Step S6: for the vertex cover solution of the current actual conflict subgraph, remove the two selected nodes in turn and update the corresponding vertex and edge attribute values;

[0014] Step S7: adding nodes to the current vertex cover scheme until all edges are covered, and updating corresponding parameters;

[0015] Step S8: Remove redundant nodes to obtain a new vertex cover solution, and update the attribute parameters of the corresponding vertices and edges;

[0016] Step S9, converting the vertex cover scheme into a global task allocation scheme, and using the cumulative constraint satisfaction operator to obtain a new feasible allocation scheme;

[0017] Step S10, using the elite strategy to obtain the current optimal global task allocation solution, based on the vertex cover solution in step S8, jump back to step S6, loop iterate until the stop criterion is met, and output the centralized task allocation result.

[0018] According to a technical solution of the present invention, in step S1, a remote sensing system S consisting of n earth observation satellites is composed of n earth observation satellites. i} i∈N , where N = {1, 2, ..., n}, each satellite s i Defined as:

[0019] s i ={o i , δ i , oc i , msa i , rt i , sp i , ast i , sut i , ρ i ,br i},

[0020] Among them, i represents the satellite orbit, δ i Indicates the load field of view, oc i Indicates the load type, msa i Indicates the maximum side swing angle, rt i Indicates shutdown delay time, sp i Indicates attitude sway speed, ast i Indicates attitude stabilization time, sut i Indicates the load startup preset time, ρ iIndicates the storage space consumed by the load per unit working time, br i represents the best observation resolution;

[0021] The observation task space R consists of m meta-tasks = {r j} j∈M , M = {1, 2, ..., m}, each meta-task r j Defined as

[0022] r j ={ω j , lc j , roc j , st j , et j , lr j},

[0023] Among them, ω j Indicates the weight or priority of the task, lc j Indicates the latitude and longitude coordinates of the task, roc j Indicates the observation type, st j Indicates the observation start time, et j Indicates the observation end time, lr j Indicates the lowest resolution;

[0024] For each satellite and meta-task group {s i , r j}, calculate the visible window set according to the satellite orbit parameters and the mission geographical location;

[0025] The kth observation window is called aow ijk ={[ws ijk , we ijk ],θ ijk}, where [ws ijk , we ijk ] indicates satellite s i Target j The observation time, θ ijk Indicates the observed side swing angle.

[0026] According to a technical solution of the present invention, in step S2, the filtering constraints at least include observation type, time, swing angle and resolution, which are represented by C1-C4 respectively, and then:

[0027] C1:oc i =roc j

[0028] C2:st j <ws ijk , we ijk <et j

[0029] C3:θ ijk ∈[-msa i , msa i ]

[0030] C4:bs i ·cos(θ ijk )>lr j

[0031] If constraints C1-C4 are satisfied, then the time window ow ijk is called a feasible observation window; denoted by K ij For satellites i Meta-task r j The number of feasible observation windows is is the sum of the number of feasible observation windows of all satellites;

[0032] For each feasible observation window ow ijk , let a ijk is the corresponding decision variable, where a ijk =1 means satellite s i Select in the observation window ijk Perform observation tasks j , otherwise a ijk =0;

[0033] Arrange all observation windows in the order of satellite index and start time, where each feasible observation window ow ijk The serial number i and {i, j, k} can be converted to each other, then ow ijk Abbreviated as

[0034] Define the global task allocation result as a length of The binary vector in represents the feasible search space, with a ijk Equivalently, it indicates whether the i-th feasible observation window is selected;

[0035] Pairwise constraints are constraints related to the attributes of two adjacent observation windows. and The observation start time meets

[0036] If i1=i2=i, then the two windows belong to the same satellite s i , then the time interval between windows must satisfy the following pairwise constraint C5:

[0037]

[0038] Considering that each meta-task only needs to be executed once, when j1=j2, that is, the two windows are for the same target task, the following pairwise constraint C6 must be satisfied:

[0039]

[0040] Cumulative constraints are constraints associated with historical mission sequences. i The following cumulative constraint C7 must be satisfied:

[0041]

[0042] When constraints C1-C7 are all satisfied, a is called a feasible task allocation scheme;

[0043] Then a constrained global optimization model for multi-satellite task allocation is constructed, which is expressed as Find the optimal feasible task allocation solution that satisfies all constraints, then the goal is to maximize the following global objective function:

[0044]

[0045] Among them, η is the number of unsatisfied constraints among the three constraints C5-C7, b j (a) represents the task assignment success flag function: when the meta-task r j When successfully allocated, b j (a) = 1; otherwise, b j (a)=0.

[0046] According to a technical solution of the present invention, step S3 specifically includes:

[0047] Each feasible observation window ow ijk As vertices, the pairwise constraints between feasible observation windows as undirected edges, and the observation benefits of the tasks corresponding to the feasible observation windows as weights, a global pairwise constraint conflict graph G = {V, E} is constructed;

[0048] Vertex v i Corresponding to the i-th feasible observation window in the arrangement, its weight is the weight or priority of the corresponding observation task. For each vertex pair {v i , v j}, if constraint C5 or constraint C6 cannot be satisfied, there is a conflicting edge e ij ∈E; for each vertex v i ∈V,Ω i = {v j ∈V|e ij ∈E} represents its neighbor set, d i is the number of its neighbors.

[0049] According to a technical solution of the present invention, the step S4 specifically includes:

[0050] Eliminate the node set V0 whose number of neighbors is 0 and obtain the actual conflict subgraph G r = {V r , E r};

[0051] For G r = {V r , E r}, let the number of nodes be K, and the degree-weight ratio of each node be

[0052] In G r = {V r , E r}, consider V r A subset S of

[0053] Opposite side ij For example, if S contains at least one vertex, then edge e ij covered;

[0054] If G r All edges of are covered, then S is a vertex cover;

[0055] Based on the calculation process of the weighted vertex cover solution, an initial feasible task allocation scheme is obtained.

[0056] According to a technical solution of the present invention, the step S5 specifically includes:

[0057] Step S51, calculate the cumulative constraint satisfaction index of the current allocation scheme, if it is not satisfied, then loop through step S52-step S53; otherwise, jump to step S54;

[0058] Step S52: for each observation window with strategy 1, calculate the cumulative constraint increments generated by it and arrange them in order from small to large;

[0059] Step S53, select the observation window with the smallest cumulative constraint increment, set its decision variable to 0, and jump to step S5;

[0060] Step S54: record the processed global task allocation scheme a 1 For x 1 , and record the current optimal solution o 1 =x 1 ;

[0061] x 1 Represents a feasible allocation plan.

[0062] According to a technical solution of the present invention, step S6 specifically includes:

[0063] Step S61: record the uncovered edge set as E U Set it to an empty set, record the taboo node set as tabu and set it to an empty set, and set the config value of all nodes to 1;

[0064] Step S62: for each node v i , denoted by i The score function value score is calculated by summing the weights of the edges formed by it and its 0 neighbors. i ;like but otherwise,

[0065] Step S63: When removing the first node, let the removable node set onf be all the current 1 nodes, where: Nodes with a value of 1; otherwise, Randomly select half of the nodes whose value is 1 and whose taboo value is 0 to form the removable node set onf;

[0066] Step S64: The node with the largest evaluation function value and the oldest age is removed, and its The value is 0;

[0067] Step S65, set the evaluation function value of the removed node to the opposite of the original value, and update the evaluation function value of its neighbors;

[0068] Step S66, update the config value of the removed node to 0, and update the config value of the neighbor to 1;

[0069] Step S67: Add the node set consisting of the node remove and its zero neighbors to E U , to update the uncovered edge set E U ;

[0070] Step S68: After removing two nodes in sequence based on steps S63 to S67, the taboo node set tabu is reset to an empty set.

[0071] According to a technical solution of the present invention, step S7 specifically includes:

[0072] Step S71: Record The node set with 0 and config 1 is fnode, and the node with the largest evaluation function value is add;

[0073] Step S72: add the node Set to 1, and update its tabu value to 1;

[0074] Step S73, update the config value of all neighbors of the node add to 1;

[0075] Step S74: Remove the edge containing the node add to update the uncovered edge set E. U ;

[0076] Step S75: For each uncovered edge, its weight is increased by 1, and the config values ​​of its two endpoints are set to 1;

[0077] Step S76, updating the evaluation function values ​​of the node add and its neighbors;

[0078] Step S77: For all endpoints of uncovered edges, update their evaluation function values ​​using the updated uncovered edge weights;

[0079] Step S78: If all edges are covered, the process ends; otherwise, the process goes to step S71 to loop;

[0080] Step S79: record the new vertex cover solution currently obtained as

[0081] According to a technical solution of the present invention, step S8 specifically includes:

[0082] Step S81: randomly select a redundant node v each time i ;

[0083] Step S82: Set the value to 0, set its evaluation function to the opposite of the original value, and update the evaluation function value of its neighbors;

[0084] Step S83: Update the selected node v i The config value of the updated neighbor is 1.

[0085] Step S84: If there are redundant nodes, go to step S81; otherwise, output the current vertex cover solution.

[0086] Step S85: Compare the original vertex cover solution With the new vertex cover scheme Increase the age of all nodes whose strategies have not changed by 1, that is, The ages of the remaining nodes are set to zero, that is,

[0087] According to a technical solution of the present invention, the step S9 specifically includes:

[0088] Step S91, construct a The binary vector The global task allocation scheme represented by V0 is 1, V r Midpoint strategy and on the contrary;

[0089] Step S92, calculate the cumulative constraint satisfaction index of the current allocation scheme; if not satisfied, perform the loop in step S93-step S94 until all cumulative constraints are satisfied; if all indicators are satisfied, jump to step S95;

[0090] Step S93: for each observation window with strategy 1, calculate the cumulative constraint increments generated by it and arrange them in order from large to small;

[0091] Step S94, select the observation window with the smallest cumulative constraint increment, set its decision variable to 0, and jump to step S92;

[0092] Step S95: record the current new constraint satisfaction solution as

[0093] According to a technical solution of the present invention, the step S10 specifically includes:

[0094] Step S101: If but Otherwise, t+1 =o t ;

[0095] Step S102: If the stopping criterion is met, output the current global optimal solution; otherwise, t=t+1, go to step f to perform an iterative loop.

[0096] Compared with the prior art, the present invention has the following beneficial effects:

[0097] The present invention proposes a multi-satellite task allocation method based on the minimum weight vertex cover of the conflict graph. For the multi-satellite task allocation problem with filtering constraints, pairwise constraints and cumulative constraints, a centralized optimization algorithm is designed by using the minimum weight vertex cover and neighborhood search in graph theory. The feasible observation windows of the satellites are regarded as vertices, the pairwise constraint conflicts are regarded as edges, and the observation benefits are regarded as vertex weights. The pairwise constraint conflict graph is constructed, and the original problem is converted into an iterative optimization solution including conflict graph vertex cover and cumulative constraint satisfaction; based on the neighborhood search technology, the minimum weight vertex cover solution algorithm and the cumulative constraint elimination operator are designed, which can effectively ensure the rapid calculation of the task allocation plan.

[0098] The present invention overcomes the shortcomings of traditional optimization algorithms, such as long calculation time and poor optimization performance due to the existence of multiple hard constraints. Through vertex cover solving and cumulative constraint satisfaction operations, it can realize the rapid calculation of feasible task allocation schemes, greatly improving the global index optimization performance and the utilization efficiency of limited observation resources. In addition, the present invention can not only be used to solve satellite task allocation problems, but its modeling method and solution mechanism can also be transplanted to task allocation and decision-making problems in other fields, such as unmanned vehicle path planning, multi-robot task allocation, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0100] Figure 1 A flowchart schematically shows a multi-star task allocation method based on minimum weight vertex cover of a conflict graph provided in an implementation case of the present invention;

[0101] Figure 2 A schematic diagram showing an example of pairwise constraint resolution based on vertex cover in an implementation case according to the present invention;

[0102] Figure 3 Schematically showing a convergence curve of a global evaluation function f in an implementation case according to the present invention;

[0103] Figure 4 Schematically showing the task allocation result of multiple satellites in an implementation case according to the present invention;

[0104] Figure 5 Schematically showing the allocation result of each task in an implementation case according to the present invention;

[0105] Figure 6 The figure schematically shows the cumulative constraint satisfaction results of each satellite in an implementation case according to the present invention. DETAILED DESCRIPTION

[0106] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0107] like Figures 1 to 6 As shown, t represents the number of iterations, Y represents the condition is met (yes), and N represents the condition is not met (no). A, B, C, D, E, and F are node names, corresponding to feasible observation time windows one by one. The numbers on the right are the observation benefits of the corresponding observation windows; S represents the set of candidate points for vertex cover, a i =1 indicates node a i Belongs to the set S, a i = 0 otherwise; edges indicate that there are pairwise constraint conflicts between windows. f represents the global evaluation index of task assignment, and when all constraints are satisfied, it represents the sum of the observation benefits corresponding to the task assignment. The X-axis coordinate corresponding to the left side of the rectangular box is the start time of the observation time window, and the X-axis coordinate corresponding to the right side is the end time of the observation time window. The numbers in the box represent the observation target ID and the visible window number of the satellite to the target; T represents the actual time, Cpct represents the cumulative observation time of the satellite, full represents the upper limit of the cumulative observation time, and actual represents the actual cumulative observation time corresponding to the allocation result.

[0108] The multi-satellite task allocation method based on the minimum weight vertex cover of the conflict graph of the present invention belongs to the intersection of graph theory, operations research and aerospace, and is proposed by using the minimum weight vertex cover and local search optimization theory. The method first constructs a global pairwise constraint conflict graph using the feasible observation window set of each satellite, and then performs iterative optimization search on the minimum weight vertex cover solution through the steps of random node removal, uncovered node addition, redundant node removal, etc., and generates a global task allocation plan based on this; then, a feasible allocation plan is obtained by using the cumulative constraint satisfaction operator; finally, jump back to the node removal step for loop iteration until the termination condition is met, and output the task allocation plan.

[0109] The method of the present invention is described in detail below with a specific implementation mode, and the experimental scenario includes n=10 satellites, m=279 observation tasks, and the priority of each observation task is an integer between 1 and 100. The experimental environment is MATLAB 2016a version, the operating system is Win10.0, 3.07Ghz, and 4G memory.

[0110] like Figure 1 As shown, a multi-star task allocation method based on minimum weight vertex cover of a conflict graph of the present invention comprises the following steps:

[0111] Step S1: Calculate the visible window of each star

[0112] According to the observation meta-task attributes and requirements, satellite parameters and platform / payload constraints, the set of visible windows of each satellite in the planning period is calculated; the sum of all feasible observation windows It is 1548.

[0113] The remote sensing system S consists of n earth observation satellites. i} i∈N , where N = {1, 2, ..., n}, each satellite s i Defined as:

[0114] s i ={o i , δ i , oc i , msa i , rt i , sp i , ast i , sut i , ρ i ,br i},

[0115] Among them, i represents the satellite orbit, δ i Indicates the load field of view, oc i Indicates the load type, msa i Indicates the maximum side swing angle, rt i Indicates shutdown delay time, sp i Indicates attitude sway speed, ast i Indicates attitude stabilization time, sut i Indicates the load startup preset time, ρ i Indicates the storage space consumed by the load per unit working time, br i represents the best observation resolution;

[0116] The observation task space R consists of m meta-tasks = {r j} j∈M , M = {1, 2, ..., m}, each meta-task r j Defined as

[0117] r j ={ω j , lc j , roc j , st j , et j , lr j},

[0118] Among them, ω j Indicates the weight or priority of the task, lc j Indicates the latitude and longitude coordinates of the task, roc j Indicates the observation type, st j Indicates the observation start time, et j Indicates the observation end time, lr j Indicates the lowest resolution;

[0119] For each satellite and meta-task group {s i , r j}, calculate the visible window set according to the satellite orbit parameters and the mission geographical location;

[0120] The kth observation window is called aow ijk ={[ws ijk , we ijk ],θ ijk}, where [ws ijk , we ijk ] indicates satellite s i Target j The observation time, θ ijk Indicates the observed side swing angle.

[0121] Step S2: Establishing a global optimization model

[0122] Considering filtering constraints, pairwise constraints and cumulative constraints, a constrained global optimization model for multi-satellite task allocation is constructed.

[0123] The filtering constraints include at least the observation type, time, sway angle and resolution, which are represented by C1-C4 respectively, and then:

[0124] C1:oc i =roc j

[0125] C2:st j <ws ijk , we ijk <et j

[0126] C3:θ ijk ∈[-msa i , msa i ]

[0127] C4:bs i ·cos(θ ijk )>lr j

[0128] If constraints C1-C4 are satisfied, then the time window ow ijk is called a feasible observation window; denoted by K ij For satellites i Meta-task r j The number of feasible observation windows is is the sum of the number of feasible observation windows of all satellites;

[0129] For each feasible observation window ow ijk , let a ijk is the corresponding decision variable, where aijk =1 means satellite s i Select in the observation window ijk Perform observation tasks j , otherwise a ijk =0;

[0130] Arrange all observation windows in the order of satellite index and start time, where each feasible observation window ow ijk Serial number and {i, j, k} can be converted to each other, then ow ijk Abbreviated as

[0131] Define the global task allocation result as a length of The binary vector in represents the feasible search space, with a ijk Equivalently, it indicates whether the i-th feasible observation window is selected;

[0132] Pairwise constraints are constraints related to the attributes of two adjacent observation windows. and The observation start time meets

[0133] If i1=i2=i, then the two windows belong to the same satellite s i , then the time interval between windows must satisfy the following pairwise constraint C5:

[0134]

[0135] Considering that each meta-task only needs to be executed once, when j1=j2, that is, the two windows are for the same target task, the following pairwise constraint C6 must be satisfied:

[0136]

[0137] Cumulative constraints are constraints associated with historical mission sequences. i The following cumulative constraint C7 must be satisfied:

[0138]

[0139] When constraints C1-C7 are all satisfied, a is called a feasible task allocation scheme;

[0140] Then a constrained global optimization model for multi-satellite task allocation is constructed, which is expressed as Find the optimal feasible task allocation solution that satisfies all constraints, then the goal is to maximize the following global objective function:

[0141]

[0142] Among them, η is the number of unsatisfied constraints among the three constraints C5-C7, b j (a) represents the task assignment success flag function: when the meta-task r j When successfully allocated, b j (a) = 1; otherwise, b j (a)=0.

[0143] Step S3: Construct a global pairwise constraint conflict graph

[0144] The feasible observation windows are regarded as vertices, the pairwise constraints between the feasible observation windows are regarded as undirected edges, and the observation benefits of the tasks corresponding to the visible windows are regarded as weights. A global pairwise constraint conflict graph is constructed, which includes:

[0145] Each feasible observation window ow ijk As vertices, the pairwise constraints between feasible observation windows as undirected edges, and the observation benefits of the tasks corresponding to the feasible observation windows as weights, a global pairwise constraint conflict graph G = {V, E} is constructed;

[0146] Vertex v i Corresponding to the i-th feasible observation window in the arrangement, its weight is the weight or priority of the corresponding observation task. For each vertex pair {v i , v j}, if constraint C5 or constraint C6 cannot be satisfied, there is a conflicting edge e ij ∈E; for each vertex v i ∈V,Ω i = {v j ∈V|e ij ∈e} represents its neighbor set, d i is the number of its neighbors.

[0147] Step S4: Initialize the task allocation scheme

[0148] The conflict graph is divided into actual conflict subgraph and non-conflict subgraph, the vertex cover solution of the actual conflict subgraph is initialized, and it is converted into a global task allocation scheme, which includes:

[0149] Eliminate the node set V0 whose number of neighbors is 0 and obtain the actual conflict subgraph G r = {V r , E r};

[0150] For G r = {V r , E r}, let the number of nodes be K, and the degree-weight ratio of each node be

[0151] In G r = {V v , E r}, consider V r A subset S of

[0152] Opposite side ij For example, if S contains at least one vertex, then edge e ij covered;

[0153] If G r All edges of are covered, then S is a vertex cover;

[0154] Based on the calculation process of the weighted vertex cover solution, an initial feasible task allocation scheme is obtained, which includes:

[0155] Step S401: Initialize V r The vertex cover solution of

[0156] Step S402: r In the example, randomly select an uncovered edge e ij , select nodes with larger degree weight to join S;

[0157] Step S403, repeat step S402 until all edges are covered;

[0158] Step S404: randomly select redundant nodes and remove them until there are no redundant nodes;

[0159] Step S405: Encode S into a binary vector of length 1548

[0160] Step S406: The strategies of all nodes in the , are reversed to obtain the global task allocation scheme a that satisfies the pairwise constraints. 1 ;

[0161] Step S407: define G r The weights of all edges in are The age of all nodes is

[0162] Step S5: Initialize the current optimal solution

[0163] The cumulative constraint satisfaction operator is used to generate an initial global feasible allocation solution that satisfies the cumulative constraints and is defined as the current optimal solution, including:

[0164] Step S51, calculate the cumulative constraint satisfaction index of the current allocation scheme, if it is not satisfied, then loop through step S52-step S53; otherwise, jump to step S54;

[0165] Step S52: for each observation window with strategy 1, calculate the cumulative constraint increments generated by it and arrange them in order from small to large;

[0166] Step S53, select the observation window with the smallest cumulative constraint increment, set its decision variable to 0, and jump to step S5;

[0167] Step S54: record the processed global task allocation scheme a 1 For x 1 , and record the current optimal solution o 1 =x 1 ;

[0168] x 1 Represents a feasible allocation plan.

[0169] Step S6: Removal of conflicting subgraph nodes

[0170] For the vertex cover solution of the current actual conflict subgraph, remove the two selected nodes in turn and update the corresponding vertex and edge attribute values, including:

[0171] Step S61: record the uncovered edge set as E U Set it to an empty set, record the taboo node set as tabu and set it to an empty set, and set the config value of all nodes to 1;

[0172] Step S62: for each node v i , denoted by i The score function value score is calculated by summing the weights of the edges formed by it and its 0 neighbors. i ;like but otherwise,

[0173] Step S63: When removing the first node, let the removable node set onf be all the current 1 nodes, where: Nodes with a value of 1; otherwise, Randomly select half of the nodes whose value is 1 and whose taboo value is 0 to form the removable node set onf;

[0174] Step S64: record the node with the largest evaluation function value and the oldest age as remove, remove it and update its The value is 0;

[0175] Step S65, set the evaluation function value of the removed node to the opposite of the original value, and update the evaluation function value of its neighbors;

[0176] Step S66: Update the config value of the removed node to 0, and update the config value of the neighbor to 1;

[0177] Step S67: Add the node set consisting of the node remove and its zero neighbors to E U , to update the uncovered edge set E U ;

[0178] Step S68, after removing two nodes in sequence based on the steps S63 to S67, the taboo node set tabu is reset to an empty set;

[0179] Step S7: Add conflict subgraph nodes

[0180] Add nodes to the current vertex cover scheme until all edges are covered and update the corresponding parameters, including:

[0181] Step S71: Record The node set with 0 and config 1 is fnode, and the node with the largest evaluation function value is add;

[0182] Step S72: add the node Set to 1, and update its tabu value to 1;

[0183] Step S73, update the config value of all neighbors of the node add to 1;

[0184] Step S74: Remove the edge containing the node add to update the uncovered edge set E. U ;

[0185] Step S75: For each uncovered edge, its weight is increased by 1, and the config values ​​of its two endpoints are set to 1;

[0186] Step S76, updating the evaluation function values ​​of the node add and its neighbors;

[0187] Step S77: For all endpoints of uncovered edges, update their evaluation function values ​​using the updated uncovered edge weights;

[0188] Step S78: If all edges are covered, the process ends; otherwise, the process goes to step S71 to loop;

[0189] Step S79: record the new vertex cover solution currently obtained as

[0190] Step S8: Removal of redundant nodes in conflicting subgraphs

[0191] Remove redundant nodes to obtain a new vertex cover solution, and update the attribute parameters of the corresponding vertices and edges, including:

[0192] Step S81: randomly select a redundant node v each time i ;

[0193] Step S82: Set the value to 0, set its evaluation function to the opposite of the original value, and update the evaluation function value of its neighbors;

[0194] Step S83: Update the selected node v i The config value of the updated neighbor is 1.

[0195] Step S84: If there are redundant nodes, go to step S81; otherwise, output the current vertex cover solution.

[0196] Step S85: Compare the original vertex cover solution With the new vertex cover scheme Increase the age of all nodes whose strategies have not changed by 1, that is, The ages of the remaining nodes are set to zero, that is,

[0197] Step S9: Generate a new feasible allocation plan

[0198] The vertex cover scheme is transformed into a global task allocation scheme, and a new feasible allocation scheme is obtained by using the cumulative constraint satisfaction operator, including:

[0199] Step S91, construct a The binary vector The global task allocation scheme represented by V0 is 1, V r Midpoint strategy and on the contrary;

[0200] Step S92, calculate the cumulative constraint satisfaction index of the current allocation scheme; if not satisfied, perform the loop in step S93-step S94 until all cumulative constraints are satisfied; if all indicators are satisfied, jump to step S95;

[0201] Step S93: for each observation window with strategy 1, calculate the cumulative constraint increments generated by it and arrange them in order from large to small;

[0202] Step S94, select the observation window with the smallest cumulative constraint increment, set its decision variable to 0, and jump to step S92;

[0203] Step S95: record the current new constraint satisfaction solution as

[0204] Step S10: Update the current global optimal solution

[0205] The elite strategy is used to obtain the current optimal global task allocation solution. Based on the vertex cover solution in step S8, the process jumps back to step S6, iterates until the stopping criterion is met, and outputs the centralized task allocation result, which specifically includes:

[0206] Step S101: If but Otherwise, t+1 =o t ;

[0207] Step S102: If the stopping criterion is met, output the current global optimal solution; otherwise, t=t+1, go to step f to perform an iterative loop.

[0208] Figure 2 This is an example of pairwise constraint resolution based on vertex cover, where each node corresponds to a feasible observation window and the edge represents the pairwise constraint conflict between windows. The red nodes form a minimum weight vertex cover. By removing nodes A, C, and D, we can get task allocation results B, E, and F that satisfy the pairwise constraints. If the result also satisfies the cumulative constraint, it is the global optimal task allocation solution.

[0209] Figure 3 is the convergence curve of the global evaluation index generated by the algorithm operation. Figure 4 Assign tasks to each satellite. Figure 5 is the allocation result of each observation task, Figure 6 The output task assignment result corresponds to the cumulative constraint satisfaction of each star. The experimental results show that the multi-star task assignment method algorithm based on the minimum weight vertex cover of the conflict graph proposed in the present invention can achieve the rapid solution of the nearest optimal task assignment solution that satisfies the pairwise constraints and cumulative constraints.

[0210] According to one aspect of the present invention, an electronic device is provided, comprising: one or more processors, one or more memories, and one or more computer programs; wherein the processor is connected to the memory, and the one or more computer programs are stored in the memory. When the electronic device is running, the processor executes the one or more computer programs stored in the memory, so that the electronic device executes a multi-star task allocation method based on minimum weight vertex cover of a conflict graph as described in any of the above technical solutions.

[0211] According to one aspect of the present invention, a computer-readable storage medium is provided for storing computer instructions. When the computer instructions are executed by a processor, a multi-star task allocation method based on minimum weight vertex cover of a conflict graph as described in any of the above technical solutions is implemented.

[0212] Computer-readable storage media may include any medium capable of storing or transmitting information. Examples of computer-readable storage media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. Code segments may be downloaded via computer networks such as the Internet, intranets, etc.

[0213] The present invention provides a multi-satellite task allocation method based on the minimum weight vertex cover of the conflict graph, which uses the minimum weight vertex cover to solve the multi-satellite task allocation problem. The main purpose is to:

[0214] First, in the multi-satellite multi-task allocation process, due to the sliding characteristics of the observation window, each satellite has a large number of feasible observation windows; at the same time, due to the limitations of physical constraints such as sway speed and shutdown delay time, the tasks corresponding to some feasible observation windows cannot be executed sequentially, and there are a large number of pairwise constraint conflicts. These conflicts can be expressed using a simple undirected graph, and by removing any vertex cover solution on it, the pairwise conflict constraints can be quickly satisfied.

[0215] Therefore, vertex cover provides an effective model representation for the rapid resolution of pairwise conflicts;

[0216] Secondly, as a classic NP-hard problem in graph theory, the minimum weight vertex cover can be solved quickly and efficiently with the help of a variety of heuristic or local search algorithms. At the same time, considering the physical constraints such as the number of single-cycle startups, startup time, and storage space, there are situations where a given task sequence cannot be executed together. Therefore, through the design of the cumulative constraint satisfaction operator, the efficient solution of feasible task allocation schemes can be achieved based on pairwise conflict resolution.

[0217] Based on the above reasons, the present invention uses minimum weight vertex cover as a modeling tool, transforms multi-satellite task allocation into a vertex cover scheme optimization and cumulative constraint satisfaction process, constructs a pairwise constraint conflict graph, designs a vertex cover optimization algorithm based on neighborhood search, and gives a feasible task allocation scheme calculation process based on cumulative constraint satisfaction, providing a more efficient solution for large-scale satellite mission planning.

[0218] The present invention can realize the rapid resolution of pairwise constraint conflicts. Since any vertex cover scheme touches all pairwise constraint conflict edges, by removing the vertex cover window set in all feasible observation windows, the obtained task set and sequence do not have pairwise conflicts.

[0219] The present invention can resolve cumulative constraint conflicts. For any task allocation scheme that satisfies pairwise constraints, a cumulative constraint satisfaction operator is used to remove corresponding feasible observation windows in sequence to quickly obtain a feasible task allocation scheme that satisfies all constraints.

[0220] The present invention can realize the efficient solution of the approximate optimal task allocation scheme. Since the sum of the weights of all observation windows is fixed, the nearest optimal solution can be calculated in the feasible solution space that satisfies the constraints by solving the approximate minimum weight vertex cover and accumulating constraint satisfaction operations. In addition, since heuristic information related to the problem is used, the method is particularly suitable for the calculation of large-scale multi-satellite multi-task allocation problems.

[0221] The present invention can also be transplanted to other task allocation problems involving pairwise constraints and cumulative constraints, including but not limited to logistics order delivery, etc.

[0222] In addition, it should be noted that the present invention can be provided as a method, an apparatus or a computer program product. Therefore, the embodiments of the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0223] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0224] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0225] It should also be noted that, in this article, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0226] Finally, it should be noted that the above is a preferred embodiment of the present invention. It should be pointed out that although the preferred embodiment of the present invention has been described, for those skilled in the art, once the basic creative concept of the present invention is known, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention. Therefore, the attached claims are intended to be interpreted as including the preferred embodiment and all changes and modifications that fall within the scope of the embodiments of the present invention.

Claims

1. A multi-star task allocation method based on minimum weight vertex cover of conflict graph, characterized in that: The following steps are involved: Step S1: Calculate the visible window set of each satellite in the planning period according to the observation meta-task attributes and requirements, satellite parameters and platform / payload constraints; Step S2, considering filtering constraints, pairwise constraints and cumulative constraints, constructing a constrained global optimization model for multi-satellite task allocation; Step S3, considering feasible observation windows as vertices, pairwise constraints between feasible observation windows as undirected edges, and observation benefits of tasks corresponding to visible windows as weights, to construct a global pairwise constraint conflict graph; Step S4, dividing the conflict graph into an actual conflict subgraph and a non-conflict subgraph, initializing the vertex cover solution of the actual conflict subgraph, and converting it into a global task allocation solution; Step S5: using the cumulative constraint satisfaction operator, generate an initial global feasible allocation scheme that satisfies the cumulative constraints, and define it as the current optimal solution; Step S6: for the vertex cover solution of the current actual conflict subgraph, remove the two selected nodes in turn and update the corresponding vertex and edge attribute values; Step S7: adding nodes to the current vertex cover scheme until all edges are covered, and updating corresponding parameters; Step S8: Remove redundant nodes to obtain a new vertex cover solution, and update the attribute parameters of the corresponding vertices and edges; Step S9, converting the vertex cover scheme into a global task allocation scheme, and using the cumulative constraint satisfaction operator to obtain a new feasible allocation scheme; Step S10, using the elite strategy to obtain the current optimal global task allocation solution, based on the vertex cover solution in step S8, jump back to step S6, loop iterate until the stop criterion is met, and output the centralized task allocation result.

2. The multi-star task allocation method based on the minimum weight vertex cover of the conflict graph according to claim 1 is characterized in that: In step S1, the remote sensing system S consisting of n earth observation satellites is i } i∈N , where N = {1, 2, ..., n}, each satellite s i Defined as: s i ={o i ,d i ,oc i ,msa i ,rt i ,sp i ,ast i ,sut i ,r i ,br i }, Among them, i represents the satellite orbit, δ i Indicates the load field of view, oc i Indicates the load type, msa i Indicates the maximum side swing angle, rt i Indicates shutdown delay time, sp i Indicates attitude sway speed, ast i Indicates attitude stabilization time, sut i Indicates the load startup preset time, ρ i Indicates the storage space consumed by the load per unit working time, br i represents the best observation resolution; The observation task space R consists of m meta-tasks = {r j } j∈M , M = {1, 2, ..., m}, each meta-task r j Defined as r j ={ω j , lc j , roc j , st j , et j , lr j }, Among them, ω j Indicates the weight or priority of the task, lc j Indicates the latitude and longitude coordinates of the task, roc j Indicates the observation type, st j Indicates the observation start time, et j Indicates the observation end time, lr j Indicates the lowest resolution; For each satellite and meta-task group {s i , r j }, calculate the visible window set according to the satellite orbit parameters and the mission geographical location; The kth observation window is called aow ijk ={[ws ijk , we ijk ],θ ijk }, where [ws ijk , we ijk ] indicates satellite s i Target j The observation time, θ ijk Indicates the observed side swing angle.

3. The multi-star task allocation method based on the minimum weight vertex cover of the conflict graph according to claim 2 is characterized in that: In step S2, the filtering constraints at least include observation type, time, roll angle and resolution, which are represented by C1-C4 respectively, and then: C1:oc i =roc j C2:st j <ws ijk ,we ijk <et j C3:θ ijk ∈[-msa i ,msa i ] C4:bs i ·cos(θ ijk )>lr j If constraints C1-C4 are satisfied, then the time window ow ijk is called a feasible observation window; denoted by K ij For satellites i Meta-task r j The number of feasible observation windows is is the sum of the number of feasible observation windows of all satellites; For each feasible observation window ow ijk , let a ijk is the corresponding decision variable, where a ijk =1 means satellite s i Select in the observation window ijk Perform observation tasks j , otherwise a ijk =0; Arrange all observation windows in the order of satellite index and start time, where each feasible observation window ow ijk Serial number and {i, j, k} can be converted to each other, then ow ijk Abbreviated as Define the global task allocation result as a length of The binary vector in represents the feasible search space, with a ijk Equivalently, it indicates whether the i-th feasible observation window is selected; Pairwise constraints are constraints related to the attributes of two adjacent observation windows. and The observation start time meets If i1=i2=i, then the two windows belong to the same satellite s i , then the time interval between windows must satisfy the following pairwise constraint C5: C5: Considering that each meta-task only needs to be executed once, when j1=j2, that is, the two windows are for the same target task, the following pairwise constraint C6 must be satisfied: C6: Cumulative constraints are constraints associated with historical mission sequences. i The following cumulative constraint C7 must be satisfied: C7: When constraints C1-C7 are all satisfied, a is called a feasible task allocation scheme; Then a constrained global optimization model for multi-satellite task allocation is constructed, which is expressed as Find the optimal feasible task allocation solution that satisfies all constraints, then the goal is to maximize the following global objective function: Among them, η is the number of unsatisfied constraints among the three constraints C5-C7, b j (a) represents the task assignment success flag function: when the meta-task r j When successfully allocated, b j (a) = 1; otherwise, b j (a)=0.

4. The multi-star task allocation method based on the minimum weight vertex cover of the conflict graph according to claim 3 is characterized in that: The step S3 specifically includes: Each feasible observation window ow ijk As vertices, the pairwise constraints between feasible observation windows as undirected edges, and the observation benefits of the tasks corresponding to the feasible observation windows as weights, a global pairwise constraint conflict graph G = {V, E} is constructed; Vertex v i Corresponding to the i-th feasible observation window in the arrangement, its weight is the weight or priority of the corresponding observation task. For each vertex pair {v i , v j }, if constraint C5 or constraint C6 cannot be satisfied, there is a conflicting edge e ij ∈E; for each vertex v i ∈V,Ω i = {v j ∈V|e ij ∈E} represents its neighbor set, d i is the number of its neighbors.

5. The multi-star task allocation method based on the minimum weight vertex cover of the conflict graph according to claim 3 is characterized in that: The step S4 specifically includes: Eliminate the node set V0 whose number of neighbors is 0 and obtain the actual conflict subgraph G r = {V r , E r }; For G r = {V r , E r }, let the number of nodes be K, and the degree-weight ratio of each node be In G r = {V r , E r }, consider V r A subset S of Opposite side ij For example, if S contains at least one vertex, then edge e ij covered; If G r All edges of are covered, then S is a vertex cover; Based on the calculation process of the weighted vertex cover solution, an initial feasible task allocation scheme is obtained.

6. The multi-star task allocation method based on conflict graph minimum weight vertex cover according to claim 3 is characterized in that: The step S5 specifically includes: Step S51, calculate the cumulative constraint satisfaction index of the current allocation scheme, if it is not satisfied, then loop through step S52-step S53; otherwise, jump to step S54; Step S52: for each observation window with strategy 1, calculate the cumulative constraint increments generated by it and arrange them in order from small to large; Step S53, select the observation window with the smallest cumulative constraint increment, set its decision variable to 0, and jump to step S5; Step S54: record the processed global task allocation scheme a 1 For x 1 , and record the current optimal solution o 1 =x 1 ; x 1 Represents a feasible allocation plan.

7. The multi-star task allocation method based on conflict graph minimum weight vertex cover according to claim 4 is characterized in that: The step S6 specifically includes: Step S61: record the uncovered edge set as E U Set it to an empty set, record the taboo node set as tabu and set it to an empty set, and set the config value of all nodes to 1; Step S62: for each node v i , denoted by i The score function value score is calculated by summing the weights of the edges formed by it and its 0 neighbors. i ;like but otherwise, Step S63: When removing the first node, let the removable node set onf be all the current 1 nodes, where: Nodes with a value of 1; otherwise, Randomly select half of the nodes whose value is 1 and whose taboo value is 0 to form the removable node set onf; Step S64: The node with the largest evaluation function value and the oldest age is removed, and its The value is 0; Step S65, set the evaluation function value of the removed node to the opposite of the original value, and update the evaluation function value of its neighbors; Step S66, update the config value of the removed node to 0, and update the config value of the neighbor to 1; Step S67: Add the node set consisting of the node remove and its zero neighbors to E U , to update the uncovered edge set E U ; Step S68: After removing two nodes in sequence based on steps S63 to S67, the taboo node set tabu is reset to an empty set.

8. The multi-star task allocation method based on conflict graph minimum weight vertex cover according to claim 4 is characterized in that: The step S7 specifically includes: Step S71: Record The node set with 0 and config 1 is fnode, and the node with the largest evaluation function value is add; Step S72: add the node Set to 1, and update its tabu value to 1; Step S73, update the config value of all neighbors of the node add to 1; Step S74: Remove the edge containing the node add to update the uncovered edge set E. U ; Step S75: For each uncovered edge, its weight is increased by 1, and the config values ​​of its two endpoints are set to 1; Step S76, updating the evaluation function values ​​of the node add and its neighbors; Step S77: For all endpoints of uncovered edges, update their evaluation function values ​​using the updated uncovered edge weights; Step S78: If all edges are covered, the process ends; otherwise, the process goes to step S71 to loop; Step S79: record the new vertex cover solution currently obtained as 9. The multi-star task allocation method based on minimum weight vertex cover of conflict graph according to claim 4 is characterized in that: The step S8 specifically includes: Step S81: randomly select a redundant node v each time i ; Step S82: Set the value to 0, set its evaluation function to the opposite of the original value, and update the evaluation function value of its neighbors; Step S83: Update the selected node v i The config value of the updated neighbor is 1. Step S84: If there are redundant nodes, go to step S81; otherwise, output the current vertex cover solution. Step S85: Compare the original vertex cover solution With the new vertex cover scheme Increase the age of all nodes whose strategies have not changed by 1, that is, The ages of the remaining nodes are set to zero, that is, The step S9 specifically includes: Step S91, construct a The binary vector The global task allocation scheme represented by V0 is 1, V r Midpoint strategy and on the contrary; Step S92, calculate the cumulative constraint satisfaction index of the current allocation scheme; if not satisfied, perform the loop in step S93-step S94 until all cumulative constraints are satisfied; if all indicators are satisfied, jump to step S95; Step S93: for each observation window with strategy 1, calculate the cumulative constraint increments generated by it and arrange them in order from large to small; Step S94, select the observation window with the smallest cumulative constraint increment, set its decision variable to 0, and jump to step S92; Step S95: record the current new constraint satisfaction solution as 10. The multi-star task allocation method based on minimum weight vertex cover of conflict graph according to claim 4 is characterized in that: The step S10 specifically includes: Step S101: If but Otherwise, t+1 =o t ; Step S102: If the stopping criterion is met, output the current global optimal solution; otherwise, t=t+1, go to step f to perform an iterative loop.

Images