Large-area marine target search task allocation method and system

Abstract

The invention provides a large-area maritime target search task allocation method and system, and the method comprises the steps: firstly constructing a task allocation model, allocating an observable task sequence set for each satellite, carrying out the reasonable decomposition of a problem, introducing a distributed solving heuristic algorithm, and obtaining an observable task sequence set; the problem that a traditional task allocation method needs more computing resources is solved, and the method has good optimization and adjustment capability.

Description

Technical Field

[0001] The invention belongs to the field of imaging satellites, and in particular relates to a method and system for allocating tasks for searching for targets at sea in a large area. Background Art

[0002] In recent years, space-based information systems composed of large-scale imaging satellite constellations are gradually becoming the most important way to obtain earth observation data. Compared with air-based earth observation systems based on drones or high-altitude airships, space-based information systems have the advantages of a large sensing range and are not restricted by terrain and deployment areas. There are many studies on task allocation algorithms for large-scale imaging satellite constellations. The following six are more representative: Reference 1 "Lian Zhenyu, Tan Yuejin, He Renjie, et al. Design of a general task planning framework for high-resolution earth observation system [J]. Computer Integrated Manufacturing Systems. 2013, 19 (5): 981–989." proposed a general framework for solving spacecraft task allocation problems. The framework decomposes earth observation tasks into multiple basic actions, and uses an intelligent optimization algorithm to solve the sequence of actions. Then, a timeline-based constraint management mechanism is used to summarize conflicts and generate the final planning sequence. The main disadvantage of this algorithm is that when solving large-scale task allocation problems, manual intervention is required to achieve effective utilization of imaging satellite resources. Reference 2 "Bianchessi N, Cordeau JF, Desrosiers J, et al. A heuristic for the multi-satellite, multi-orbit and multi-user management of Earth observation satellites [J]. European Journal of Operational Research. 2007, 177 (2): 750–762." built a multi-satellite multi-orbit centralized task allocation algorithm for the French PLEIADES satellite cluster system. The algorithm models the constellation multi-task allocation problem as a three-stage problem. The first stage of the problem is mainly to maximize the total planning revenue, the second stage is to allocate the system time available to customers, and the third stage is to redistribute the remaining available time. Reference 2 also proposed a hybrid intelligent optimization algorithm based on taboo search and column generation algorithm to solve the problem. Reference 3 "Chen Yingwu, Fang Yanshen, Li Jufang, et al. Constraint programming model for satellite mission scheduling problem [J]. Journal of National University of Defense Technology. 2006, 28 (005): 126-132." First, the constellation task allocation problem is modeled as a constraint satisfaction problem, and a taboo search algorithm is designed to solve the problem.Reference 4 “Cai Derong. Research on multi-satellite joint imaging mission planning based on ant colony algorithm [D]. Master's thesis of University of Electronic Science and Technology of China, 2012.” The star cluster task allocation problem is modeled as a type of loading and unloading problem with special time window constraints, and a heuristic algorithm based on the "Max-Min Ant Systems" algorithm is designed to solve the model. Reference 5 “He L, Liu X, Xing L, et al. Cloud Avoidance Scheduling Algorithm for Agile Optical Satellites [C]. Gong M, Pan L, Song T, et al. Bio-Inspired Computing - Theories and Applications -10th International Conference, BIC-TA 2015, Hefei, China, September 25-28,2015, Proceedings. 2015: 161–172.” A two-stage heuristic algorithm based on ant colony algorithm and iterative repair strategy is proposed to solve the star cluster task allocation problem. The algorithm can effectively improve the efficiency of task allocation between star clusters. Reference 6 "Valicka CG, Garcia D, Staid A, et al. Mixed-integer programming models for optimal constellation scheduling given cloud cover uncertainty [J]. European Journal of Operational Research. 2019, 275." modeled the constellation task allocation problem as a mixed integer programming problem considering random factors and solved it using CPLEX. However, this method can only solve constellation task allocation problems with a smaller problem scale, and its effectiveness in large-scale allocation problems remains to be verified. Summary of the invention

[0003] The main problem solved by the present invention is how to improve the task allocation efficiency of a satellite constellation when performing a large-area maritime target search task, realize online rapid allocation of search tasks, and provide a large-area maritime target search task allocation method and system.

[0004] In order to solve the above technical problems, the technical solution adopted is: A method for allocating tasks for searching targets at sea in a large area comprises the following steps: Step 1: Get the set of task locations N to be searched and the shortest search time sm to complete each task i , i represents the i-th task location, where For the actual mission location, is the virtual search mission start location of any imaging satellite in the constellation, and n is the number of actual mission locations; Step 2: Calculate the conversion time required for the individual imaging satellite to transfer between two mission locations and form a conversion time matrix C. ,in , Indicates the mission location i Mission location j The satellite switching distance between Indicates the mission location i Mission location j The satellite switching time between the two missions, v represents the switching speed between imaging satellite missions; Step 3: Construct a task allocation model based on the conversion time matrix C, the satellite constellation K, and the search task location set N. , km represents the maximum number of imaging satellites contained in the constellation, and k represents the kth imaging satellite in the satellite constellation K; Step 4: Solve and output the task allocation model. The solution obtained is a set of task allocation sequences for all individual imaging satellites in the satellite constellation K. The task allocation sequence of each satellite forms a search path according to the observation time.

[0005] Furthermore, the task allocation model is: Optimization goal:

[0006]

[0007]

[0008] Constraints:

[0009]

[0010]

[0011]

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019] Formula (1) indicates the shortest total conversion time, formula (2) indicates the least number of imaging satellites, and formula (3) indicates the longest total search time; is the decision variable, , which means: when , representing the imaging satellite k From the mission location i Transfer to the mission location j , and complete the task at the location i and mission location j Target search; otherwise, ; Formula (4) indicates that the total number of imaging satellites in use cannot exceed the upper limit of the number of available imaging satellite constellations; Formulas (5) and (6) indicate that any imaging satellite must return to the starting point after completing its mission, while ensuring the continuity of the search path of any imaging satellite; Formula (7) indicates that any location to be searched can only be searched once by one imaging satellite; Formula (8) indicates that the power resources consumed by the search task assigned to a single imaging satellite cannot exceed the upper limit of its power resources. , indicating the completion of The power required to search for each task location is the amount of power consumed, and E is the power set consumed when searching for each task location. Formulas (9), (10), and (11) indicate that the search time for the task location must be arranged within the expected search time window for the task location. Indicates search The earliest time a mission location can be searched, Indicates The latest time a mission location can be searched, Indicates The actual search start time for each mission location, Indicates The actual search end time of the mission location; Formula (12) indicates that the imaging satellite’s search time at the mission location cannot be less than the minimum time required to complete the search at the location, Indicates completion of The shortest time required to search for a mission location; Formula (13) ensures the continuity of the time schedule of a single imaging satellite. Indicates that from Mission location to The time required to change the task location, Indicated in The waiting time at the mission location is the time required for the imaging satellite to start searching immediately after arriving at the mission location within the time allowed.

[0020] Furthermore, the method for solving the task allocation model is a hybrid multi-objective heuristic optimization algorithm based on a multi-ant colony algorithm and a neighborhood search algorithm.

[0021] Furthermore, the hybrid multi-objective heuristic optimization algorithm refers to: Step 4.1: Initialize the Pareto front condition and the global pheromone matrix, and set the number of iterations , the Pareto front condition refers to decomposing the task allocation model into multiple independent single-objective optimization problems with additional constraints, and the additional constraints refer to additional constraints added to each single-objective sub-problem while inheriting the constraints of the current task allocation model; Step 4.2: Use the Pareto frontier condition to decompose the multiple optimization objectives of the task allocation model into multiple single-objective sub-problems, establish an ant colony algorithm for each single-objective sub-problem and solve it separately. The initial pheromone matrix of each ant colony algorithm is the same as the global pheromone matrix; Step 4.3: After each single-objective subproblem is solved using the ant colony algorithm, the solution of the single-objective subproblem is first compared with the optimal value stored by the ant colony algorithm. If the solution of the current single-objective subproblem is better than the optimal value stored by the ant colony algorithm, a new local optimal solution is found, and then jump to step 4.4, otherwise jump to step 4.8; Step 4.4: Compare the newly obtained local optimal solution with each solution in the existing global optimal solution set. If the local optimal solution is the new global optimal solution, jump to step 4.5, otherwise jump to step 4.7; Step 4.5: Add the new global optimal solution to the global optimal solution set; Step 4.6: Update the global pheromone matrix; Step 4.7: Number of iterations Add 1 and update the Pareto front condition; Step 4.8: Use the neighborhood search algorithm to further optimize the current global optimal solution set; Step 4.9: For each solution in the global optimal solution set after optimization, use a multi-objective aggregation strategy to aggregate according to user preferences to obtain a feasible solution that meets user needs; Step 4.10: Output a feasible solution that meets user requirements.

[0022] Furthermore, the ant colony algorithm refers to: Step 4.2.1: Define the single-objective sub-problem to be solved by the ant colony algorithm OMa , a=1, 2, 3, is the serial number of the single-objective subproblem, initializes the local pheromone matrix and the optimization target value, and sets the number of iterations t The local pheromone matrix and the optimization target value are initialized as follows:

[0023]

[0024] in, is the initial value of the local pheromone matrix, is the current value of the global pheromone matrix, is the ath single-objective subproblem, is the latest local optimal result of the a-th single-objective subproblem; Step 4.2.2: Place the artificial ants at the departure point of the imaging satellite cluster to start a round of search; Step 4.2.3: The artificial ant selects the task location to be searched next based on the probability matrix calculated by formula (20);

[0025] in, is the heuristic factor, is the probability of selecting the jth task location as the next search location;

[0026] in, For the t The remaining path at the iteration ij The amount of pheromones on , Respectively indicated by iThe conversion time required to turn from the jth task location to the jth task location and the search path ij The degree of importance of the pheromone concentration on the deposit; Step 4.2.4: After the artificial ant selects the task location for the next search, it compares the current time with the latest searchable time of all remaining task locations. If there are still task locations that can be searched, it returns to step 4.2.3 to start the next round of selection; otherwise, it goes to step 4.2.5; Step 4.2.5: The artificial ant returns to the initial task location and completes the construction of a task location search path. At this time, if there are still task locations that have not been searched, it returns to step 4.2.2 and starts a new round of search; otherwise, it completes the construction of a new solution, and the number of iterations is t Increase 1 ， Go to step 4.2.6; Step 4.2.6: For a new solution, first determine whether it is a feasible solution that satisfies all constraints. If so, go to step 4.2.7; otherwise, go to step 4.2.9; Step 4.2.7: For a new feasible solution, update the pheromone according to formula (22);

[0027] in, is the pheromone evaporation coefficient, is the pheromone residue constant; Step 4.2.8: If the optimization objective value of the feasible solution is better than , then update , and go to step 4.2.10; otherwise, go directly to step 4.2.10; Step 4.2.9: For a new infeasible solution, update the pheromone according to formula (23).

[0028] (twenty three) Step 4.2.10: After completing the pheromone update, make a judgment. If the number of iterations is t Less than the maximum number of iterations of the ant colony algorithm , then go to step 4.2.11, otherwise go to step 4.2.12; Step 4.2.11: If it is a subproblem OM3 , then the search time extension operator is called to complete the shortest time set required to search for the task location Make adjustments and go to step 4.2.2; otherwise, go directly to step 4.2.2; the search time extension operator refers to an operator used to increase the search time of the i-th task location based on probability: Step 4.2.12: After the search is completed, output and the corresponding feasible solution.

[0029] Furthermore, the operator that increases the search time of the i-th task location based on probability is:

[0030] in,

[0031]

[0032] It indicates the probability of increasing the search time when the i-th task location is selected. When the i-th task location is selected, the increase in the search time at the task location is ,

[0033] represents the congestion of the search time window of the i-th task location, represents the i-th task location and the j The length of overlap of the search time windows for each task location.

[0034] Furthermore, the global pheromone update method is: When any ant colony algorithm finds a new global optimal solution, the global pheromone matrix is ​​updated using formula (28);

[0035] in, For the t In the iteration, the i-th task location is j The amount of global pheromone remaining on the search path of the task location, is the residual amount of pheromone, = .

[0036] Furthermore, in step 4.8, the method of using the neighborhood search algorithm to further optimize the current global optimal solution set is: For each solution in the global optimal solution set, suppose there is a solution rn task location search paths, denoted as , the number of task locations contained in each path is , the neighborhood search algorithm is used for each search path to adjust the order of searching for task locations in the search path. If a better path than the current search path is found after the adjustment, the current search path is used to replace it, and the neighborhood search is performed again on the search path until all search paths have been searched through the neighborhood search and no better search path can be found.

[0037] Furthermore, the multi-objective aggregation strategy refers to:

[0038] f is a single solution in the global optimal solution set, , , They are the user's preference for the three optimization objectives, and are adjusted according to user needs. is the maximum value of optimization objective 1 in the global optimal solution set, The maximum value of the optimization objective 2 is the global optimal solution. The maximum value of optimization objective 3 is the global optimal solution. To optimize the minimum value of objective 3, To optimize the minimum value of objective 2, is the minimum value of optimization objective 1.

[0039] The present invention also provides a large-area maritime target search task allocation system, which uses the steps of the large-area maritime target search task allocation method described above.

[0040] By adopting the above technical solution, the present invention has the following beneficial effects: The present invention provides a large-area maritime target search task allocation method and system, which constructs a task allocation model, allocates an observable task sequence set to each satellite, and solves the problem that traditional task allocation methods usually require a large number of computing resources by reasonably decomposing the problem and introducing a heuristic algorithm that can be solved in a distributed manner. The system has good optimization and adjustment capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of the system of the present invention; Figure 2 for IMAS Algorithm flow chart; Figure 3 is the Pareto frontier condition; Figure 4 This is the flow chart of the ant colony algorithm; Figure 5 This is a schematic diagram of the 2-opt neighborhood search algorithm; Figure 6Schematic diagram of the 3-opt neighborhood search algorithm. DETAILED DESCRIPTION

[0042] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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.

[0043] Figures 1 to 6 A specific embodiment of a large-area marine target search task allocation method is shown, such as Figure 1 As shown, the following steps are included: Step 1: Get the set of task locations N to be searched and the shortest search time smi to complete each task. i represents the i-th task location, where For the actual mission location, is the virtual search mission starting location of any imaging satellite in the constellation, and n is the number of actual mission locations.

[0044] Step 2: Calculate the conversion time required for the individual imaging satellite to transfer between two mission locations and form a conversion time matrix C. ,in , Indicates the mission location i Mission location j The satellite switching distance between Indicates the mission location i Mission location j is the satellite switching time between missions, and v is the switching speed between imaging satellite missions.

[0045] Step 3: Construct a task allocation model based on the conversion time matrix C, the satellite constellation K, and the search task location set N. , km represents the maximum number of imaging satellites contained in the constellation, and k represents the kth imaging satellite in the satellite constellation K.

[0046] Task Allocation Model OM The detailed description is as follows.

[0047] First, define the model OM The decision variables involved are as follows: , which means: when , representing the imaging satellite k From the mission area i Transfer to the mission area j , and complete the task in the mission areai and mission area j Target search; otherwise, ; For the mission location i Waiting time; For the mission location i The time when the search actually started; For the location i The time when the search actually ended.

[0048] Model OM It includes the following three optimization goals:

[0049]

[0050]

[0051] Among them, (1) represents the shortest total conversion time, (2) represents the least number of imaging satellites used, and (3) represents the longest total search time. is the decision variable, , which means: when , representing the imaging satellite k From the mission location i Transfer to the mission location j , and complete the task at the location i and mission location j Target search; otherwise, .

[0052] Model OM The following constraints are included:

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] Among the above constraints, formula (4) indicates that the total number of imaging satellites in use cannot exceed the upper limit of the number of available imaging satellite constellations; formulas (5) and (6) indicate that any imaging satellite must return to the starting point after completing the mission, while ensuring the continuity of the search path of any imaging satellite; formula (7) indicates that any location to be searched can only be searched once by one imaging satellite; formula (8) indicates that the power resources consumed by the search mission assigned to a single imaging satellite cannot exceed the upper limit of its power resources. , indicating the completion of The power required to search for each task location is the amount of power consumed, and E is the power set consumed when searching for each task location. Formulas (9), (10), and (11) indicate that the search time for the task location must be arranged within the expected search time window for the task location. Indicates search The earliest time a mission location can be searched, Indicates The latest time a mission location can be searched, Indicates The actual search start time for each mission location, Indicates The actual search end time of the mission location; Formula (12) indicates that the imaging satellite’s search time at the mission location cannot be less than the minimum time required to complete the search at the location, Indicates completion of The shortest time required to search for a mission location; Formula (13) ensures the continuity of the time schedule of a single imaging satellite. Indicates that from Mission location to The time required to change the task location, Indicated in The waiting time at the mission location is the time required for the imaging satellite to start searching immediately after arriving at the mission location within the time allowed.

[0064] Step 4: Solve and output the task allocation model. The solution obtained is a set of task allocation sequences for all individual imaging satellites in the satellite constellation K. The task allocation sequence of each satellite forms a search path according to the observation time.

[0065] In this embodiment, the method for solving the task allocation model is a hybrid multi-objective heuristic optimization based on multi-ant colony algorithm and neighborhood search algorithm. IMAS Algorithm. The hybrid multi-objective heuristic optimization algorithm refers to: Step 4.1: Initialize the Pareto front condition and the global pheromone matrix, and set the number of iterations The Pareto front condition refers to decomposing the task allocation model into multiple independent single-objective optimization problems with additional constraints. The additional constraints refer to the additional constraints added to each single-objective sub-problem while inheriting the constraints of the current task allocation model.

[0066] In this embodiment, IMAS The algorithm constructs the Pareto frontier condition to model the task allocation OM Decompose into multiple independent single-objective optimization problems with additional constraints. In this embodiment, the Pareto front condition can be expressed as a triple Description, where , , The sub-problems OM1 , OM2 , OM3 The latest local optimal results of Figure 2 shown.

[0067] Step 4.2: Use the Pareto front condition to decompose the multiple optimization objectives of the task allocation model into multiple single-objective sub-problems, and establish an ant colony algorithm to solve each single-objective sub-problem separately. The initial pheromone matrix of each ant colony algorithm is the same as the global pheromone matrix.

[0068] In this embodiment, Figure 4 As shown, the ant colony algorithm refers to: Step 4.2.1: Define the single-objective sub-problem to be solved by the ant colony algorithm OMa , a=1, 2, 3, is the serial number of the single-objective subproblem, initializes the local pheromone matrix and the optimization target value, and sets the number of iterations t The local pheromone matrix and the optimization target value are initialized as follows:

[0069]

[0070] in, is the initial value of the local pheromone matrix, is the current value of the global pheromone matrix, is the ath single-objective subproblem, is the latest local optimal result of the a-th single-objective subproblem; Step 4.2.2: Place the artificial ants at the departure point of the imaging satellite cluster to start a round of search; Step 4.2.3: The artificial ant selects the task location to be searched next based on the probability matrix calculated by formula (20);

[0071] in, is the heuristic factor, is the probability of selecting the jth task location as the next search location;

[0072] in, For the t The remaining path at the iteration ij The amount of pheromones on , Respectively indicated by i The conversion time required to turn from the jth task location to the jth task location and the search path ij The degree of importance of the pheromone concentration on the deposit; Step 4.2.4: After the artificial ant selects the task location for the next search, it compares the current time with the latest searchable time of all remaining task locations. If there are still task locations that can be searched, it returns to step 4.2.3 to start the next round of selection; otherwise, it goes to step 4.2.5; Step 4.2.5: The artificial ant returns to the initial task location and completes the construction of a task location search path. At this time, if there are still task locations that have not been searched, it returns to step 4.2.2 and starts a new round of search; otherwise, it completes the construction of a new solution, and the number of iterations is t Increase 1 ， Go to step 4.2.6; Step 4.2.6: For a new solution, first determine whether it is a feasible solution that satisfies all constraints. If so, go to step 4.2.7; otherwise, go to step 4.2.9; Step 4.2.7: For a new feasible solution, update the pheromone according to formula (22);

[0073] in, is the pheromone evaporation coefficient, is the pheromone residue constant; Step 4.2.8: If the optimization objective value of the feasible solution is better than , then update , and go to step 4.2.10; otherwise, go directly to step 4.2.10; Step 4.2.9: For a new infeasible solution, update the pheromone according to formula (23).

[0074]

[0075] Step 4.2.10: After completing the pheromone update, if the number of iterations t Less than the maximum number of iterations of the ant colony algorithm , then go to step 4.2.11, otherwise go to step 4.2.12; Step 4.2.11: If it is a subproblem OM3 , then the search time extension operator is called to complete the shortest time set required to search for the task location Make adjustments and go to step 4.2.2; otherwise, go directly to step 4.2.2; the search time extension operator refers to an operator used to increase the search time of the i-th task location based on probability.

[0076] In this embodiment, the operator for increasing the search time of the i-th task location based on probability is:

[0077] in,

[0078]

[0079] It indicates the probability of increasing the search time when the i-th task location is selected. When the i-th task location is selected, the increase in the search time at the task location is ,

[0080] represents the time window overlap of the i-th task location, represents the i-th task location and the j The length of overlap of the search time windows for each task location.

[0081] Step 4.2.12: After completing the search, output and the corresponding feasible solution.

[0082] Step 4.3: After each single-objective subproblem is solved using the ant colony algorithm, first compare the solution of the single-objective subproblem with the optimal value stored by the ant colony algorithm. If the solution of the current single-objective subproblem is better than the optimal value stored by the ant colony algorithm, a new local optimal solution is found, and jump to step 4.4, otherwise jump to step 4.8.

[0083] Step 4.4: Compare the newly obtained local optimal solution with each solution in the existing global optimal solution set. If the local optimal solution is the new global optimal solution, jump to step 4.5, otherwise jump to step 4.7.

[0084] Step 4.5: Add the new global optimal solution to the global optimal solution set.

[0085] Step 4.6: Update the global pheromone matrix.

[0086] In this embodiment, the global pheromone update method is: When any ant colony algorithm finds a new global optimal solution, the global pheromone matrix is ​​updated using formula (28);

[0087] in, For the t In the iteration, the i-th task location is j The amount of global pheromone remaining on the search path of the task location, is the residual amount of pheromone, = .

[0088] Step 4.7: Number of iterations Add 1 and update the Pareto front condition.

[0089] In this embodiment, the Pareto front condition is expressed as a triple Description, where , , The sub-problems OM1 , OM2 , OM3 The latest local optimal result of the three sub-goal problems is obtained, and the Pareto front conditions are updated with these three latest local optimal results.

[0090] Step 4.8: Use the neighborhood search algorithm to further optimize the current global optimal solution set.

[0091] In this embodiment, the specific optimization method is: For each solution in the global optimal solution set, suppose there is a solution rn task location search paths, denoted as , the number of task locations contained in each path is , the neighborhood search algorithm is used for each search path to adjust the order of searching for task locations in the search path. If a better path than the current search path is found after the adjustment, the current search path is used to replace it, and the neighborhood search is performed again on the search path until all search paths have been searched in the neighborhood and no better search path can be found. The method of performing neighborhood search on the search path is: for each search path, the two neighborhood search algorithms, 2-opt algorithm and 3-opt algorithm, are used in turn to optimize and adjust the search path.

[0092] The specific operation steps of the 2-opt algorithm are as follows: Figure 5 shown.

[0093] Step 4.8.1: Initialize the algorithm and set the number of iterations im=1 .

[0094] Step 4.8.2: When If yes, go to step 4.8.3; otherwise, go to step 4.8.8.

[0095] Step 4.8.3: When Initialize the exchange starting point , path length rl=1 , and go to step 4.8.4; otherwise, go to step 4.8.7.

[0096] Step 4.8.4: is the starting point and the path length is rl The search order of the task locations in some search paths is reversed and connected into a new path. For specific operations, see Figure 5 (exist Figure 5 In the example, node 3 is , rl=1 ), check constraints (4) to (14) and recalculate the optimization target value of the solution. If the constraints are met and it is a new global optimal solution, update the global optimal solution and return to step 4.8.3, otherwise jump to step 4.8.5.

[0097] Step 4.8.5: If rl< - -1, then add 1 to rl and go to step 4.8.4; otherwise go to step 4.8.6.

[0098] Step 4.8.6: If < -2 ,but Add 1 to initialize the path length rl=1 , and go to step 4.8.4, otherwise go to step 4.8.7.

[0099] Step 4.8.7: im Add 1 and go to step 4.8.2.

[0100] Step 4.8.8: Complete the 2-opt algorithm and output the global optimal solution after 2-opt optimization.

[0101] The specific operation steps of the 3-opt algorithm are as follows: Figure 6 shown.

[0102] Step 4.8.1': Initialize the algorithm and set the number of iterations im=1 .

[0103] Step 4.8.2': When If yes, go to step 4.8.3', otherwise go to step 4.8.10'.

[0104] Step 4.8.3': When Initialize the exchange starting point , and go to step 4.8.4', otherwise go to step 4.8.5'.

[0105] Step 4.8.4': and , and The task location search path between them is reversed and reconnected into a new task location search path. For specific connection methods, see Figure 6 (exist Figure 6 In the figure, points 2, 3, 4, and 5 correspond to , , , ). For the newly generated path, check the constraints (4) to (14) and recalculate the optimization target value of the solution. If the constraints are satisfied and it is a new global optimal solution, update the global optimal solution and return to step 4.8.3', otherwise go to step 4.8.5'.

[0106] Step 4.8.5': If ,but = +1 and go to step 4.8.4', otherwise go to step 4.8.6'.

[0107] Step 4.8.6': If ,but = +1, initialization , and go to step 4.8.4', otherwise, go to step 4.8.7'.

[0108] Step 4.8.7': If ,but = +1, initialization , , and go to step 4.8.4', otherwise, go to step 4.8.8'.

[0109] Step 4.8.8': If ,but = +1, initialization , , , and go to step 4.8.4', otherwise, go to step 4.8.9'.

[0110] Step 4.8.9': im Add 1 and go to step 4.8.2'.

[0111] Step 4.8.10': Complete the 3-opt algorithm and output the final global optimal solution.

[0112] After completing the 2-opt and 3-opt algorithms for each solution in the global optimal solution set, the optimization of the global optimal solution set can be completed.

[0113] In this embodiment, IMAS The algorithm is a hybrid multi-objective heuristic optimization algorithm based on multi-ant colony algorithm and neighborhood search algorithm. IMAS In the algorithm, the multi-ant colony algorithm is used to find a set of optimal initial solutions, and the neighborhood search algorithm performs iterative optimization based on the initial feasible solutions found by the multi-ant colony algorithm. IMAS The algorithm flow chart is as follows Figure 2 shown.

[0114] Step 4.9: For each solution in the global optimal solution set after optimization, use the multi-objective aggregation strategy to aggregate them according to user preferences to obtain feasible solutions that meet user needs.

[0115] In this embodiment, the multi-objective aggregation strategy refers to:

[0116] f is a single solution in the global optimal solution set, , , They are the user's preference for the three optimization objectives, and are adjusted according to user needs. is the maximum value of optimization objective 1 in the global optimal solution set, The maximum value of the optimization objective 2 is the global optimal solution. The maximum value of optimization objective 3 is the global optimal solution. To optimize the minimum value of objective 3, To optimize the minimum value of objective 2, is the minimum value of optimization objective 1.

[0117] Step 4.10: Output a feasible solution that meets user requirements.

[0118] The effects that can be achieved by the present invention are described below through specific experiments.

[0119] The specific implementation scheme of the present invention can be described as follows through a case: (1) Simulation environment: All algorithms and simulation experiments described in this paper are run on a computer using an Intel® Core™ i5-4460 (3.2 GHz) and 4 GB of memory.

[0120] (2) Simulation environment description: This simulation experiment selected three different scales of cases: 50 task locations to be searched, 100 task locations to be searched, and 400 task locations to be searched. and conversion speed v The settings such as the search time window adopt the relevant data from the internationally used Solomon standard case set and the Gehring&Homberger standard case set (the specific settings can be obtained from the website https: / / www.sintef.no / projectweb / top / vrptw / solomon-benchmark / ).

[0121] (3) Parameter setting description: Control parameters of ant colony algorithm Set to , the maximum number of iterations of the ant colony algorithm Set to 400, Set to equal.

[0122] (4) Comparative algorithm description: To illustrate the effectiveness of the algorithm described in the present invention, we compare the optimization method described in the present invention with three other widely used multi-objective optimization algorithms (NSGA II, reference (7) “Deb, K., Pratap, A., Agarwal, S. and Meyarivan, T. (2002). ”A fast and elitist multiobjective genetic algorithm: NSGA-II,” in IEEE Transactions on Evolutionary Computation, vol. 6, no. 2, pp. 182-197.”; SPEA2, reference (8) “Zitzler, E., Laumanns, M., and Thiele, L. (2001). SPEA2: Improving the strength Pareto evolutionary algorithm for multiobjective optimization. In Proceedings of evolutionary methods for design, optimisation, and control, Barcelona, ​​Spain (pp. 95–100)”; VNS, reference (9) “ Hansen P, Mladenović N, Moreno P. Variable neighborhood search: methods and applications [J]. Annals of Operations Research. 2010,175 (1): 367–407.”) and the specific results are as follows.

[0123] Table 1 Simulation results of 100 task locations to be searched

[0124] Table 2 Simulation results of 400 task locations to be searched

[0125] The Solomon standard example set and the Gehring & Homberger standard example set used in the present invention divide the examples into three categories (C type, R type, RC type). This experiment conducted comparative tests on the three types of examples, and calculated the effectiveness of the four types of algorithms involved in the test (MAS-LS (algorithm described in the present invention), NSGA II (comparison algorithm 1), SPEA 2 (comparison algorithm 2), VNS (comparison algorithm 3)), and compared the average value and optimal value of the optimal solution obtained by the above four types of algorithms on the three optimization objectives. In Tables 1 and 2, the mark 0.0% represents that the optimization target value of the solution obtained by the algorithm has obtained the optimal value in this type of example, and the other values ​​represent the difference between the optimization target value of the solution obtained by the algorithm and the optimal value in this type of example.

[0126] From the results in Table 1 and Table 2, it can be seen that in all the test examples, the results of the MAS-LS algorithm described in the present invention have achieved the best results in the target values ​​of the three optimization objectives. Therefore, it can be concluded that the algorithm described in the present invention has better performance in solving the large-area marine target search task allocation problem than the currently widely used algorithms.

[0127] The present invention also provides a large-area maritime target search task allocation system, which uses the steps of the large-area maritime target search task allocation method described above.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for allocating tasks for searching targets at sea in a large area, characterized in that: The following steps are involved: Step 1: Get the set of task locations N to be searched and the shortest search time sm to complete each task i , i represents the i-th task location, where For the actual mission location, is the virtual search mission start location of any imaging satellite in the constellation, and n is the number of actual mission locations; Step 2: Calculate the conversion time required for the individual imaging satellite to transfer between two mission locations and form a conversion time matrix C. ,in , Indicates the mission location i Mission location j The satellite switching distance between Indicates the mission location i Mission location j The satellite switching time between the two missions, v represents the switching speed between imaging satellite missions; Step 3: Construct a task allocation model based on the conversion time matrix C, the satellite constellation K, and the search task location set N. , km represents the maximum number of available imaging satellites contained in the constellation, and k represents the kth imaging satellite in the satellite constellation K; Step 4: Solve and output the task allocation model. The solution obtained is a set of task allocation sequences for all individual imaging satellites in the satellite constellation K. The task allocation sequence of each satellite forms a search path according to the observation time.

2. A method for allocating tasks for searching targets at sea in a large area according to claim 1, characterized in that: The task allocation model is: Optimization goal: Constraints: Formula (1) indicates the shortest total conversion time, formula (2) indicates the least number of imaging satellites, and formula (3) indicates the longest total search time; is the decision variable, , which means: when , representing the imaging satellite k From the mission location i Transfer to the mission location j , and complete the task at the location i and mission location j Target search; otherwise, ; Formula (4) indicates that the total number of imaging satellites in use cannot exceed the upper limit of the number of available imaging satellite constellations; Formulas (5) and (6) indicate that any imaging satellite must return to the starting point after completing its mission, while ensuring the continuity of the search path of any imaging satellite; Formula (7) indicates that any location to be searched can only be searched once by one imaging satellite; Formula (8) indicates that the power resources consumed by the search task assigned to a single imaging satellite cannot exceed the upper limit of its power resources. , indicating the completion of The power required to search for each task location is the amount of power consumed, and E is the power set consumed when searching for each task location. Formulas (9), (10), and (11) indicate that the search time for the task location must be arranged within the expected search time window for the task location. Indicates search The earliest time a mission location can be searched, Indicates The latest time a mission location can be searched, Indicates The actual search start time for each mission location, Indicates The actual search end time of the mission location; Formula (12) indicates that the imaging satellite’s search time at the mission location cannot be less than the minimum time required to complete the search at the location, Indicates completion of The shortest time required to search for a mission location; Formula (13) ensures the continuity of the time schedule of a single imaging satellite. Indicates that from Mission location to The time required to change the task location, Indicated in The waiting time at the mission location is the time required for the imaging satellite to start searching immediately after arriving at the mission location within the time allowed.

3. A method for allocating tasks for searching targets at sea in a large area according to claim 2, characterized in that: The method for solving the task allocation model is a hybrid multi-objective heuristic optimization algorithm based on a multi-ant colony algorithm and a neighborhood search algorithm.

4. A method for allocating large-area marine target search tasks according to claim 3, characterized in that ,The hybrid multi-objective heuristic optimization algorithm refers to: Step 4.1: Initialize the Pareto front condition and the global pheromone matrix, and set the number of iterations , the Pareto front condition refers to decomposing the task allocation model into multiple independent single-objective optimization problems with additional constraints, and the additional constraints refer to additional constraints added to each single-objective sub-problem while inheriting the constraints of the current task allocation model; Step 4.2: Use the Pareto frontier condition to decompose the multiple optimization objectives of the task allocation model into multiple single-objective sub-problems, establish an ant colony algorithm for each single-objective sub-problem and solve it separately. The initial pheromone matrix of each ant colony algorithm is the same as the global pheromone matrix; Step 4.3: After each single-objective subproblem is solved using the ant colony algorithm, the solution of the single-objective subproblem is first compared with the optimal value stored by the ant colony algorithm. If the solution of the current single-objective subproblem is better than the optimal value stored by the ant colony algorithm, a new local optimal solution is found, and then jump to step 4.4, otherwise jump to step 4.8; Step 4.4: Compare the newly obtained local optimal solution with each solution in the existing global optimal solution set. If the local optimal solution is the new global optimal solution, jump to step 4.5, otherwise jump to step 4.7; Step 4.5: Add the new global optimal solution to the global optimal solution set; Step 4.6: Update the global pheromone matrix; Step 4.7: Number of iterations Add 1 and update the Pareto front condition; Step 4.8: Use the neighborhood search algorithm to further optimize the current global optimal solution set; Step 4.9: For each solution in the global optimal solution set after optimization, use a multi-objective aggregation strategy to aggregate according to user preferences to obtain a feasible solution that meets user needs; Step 4.10: Output a feasible solution that meets user requirements.

5. A method for allocating tasks for searching targets at sea in a large area according to claim 4, characterized in that: The ant colony algorithm refers to: Step 4.2.1: Define the single-objective sub-problem to be solved by the ant colony algorithm OMa , a=1, 2, 3, is the serial number of the single-objective subproblem, initializes the local pheromone matrix and the optimization target value, and sets the number of iterations t The local pheromone matrix and the optimization target value are initialized as follows: in, is the initial value of the local pheromone matrix, is the current value of the global pheromone matrix, is the ath single-objective subproblem, is the latest local optimal result of the a-th single-objective subproblem; Step 4.2.2: Place the artificial ants at the departure point of the imaging satellite cluster to start a round of search; Step 4.2.3: The artificial ant selects the task location to be searched next based on the probability matrix calculated by formula (20); in, is the heuristic factor, is the probability of selecting the jth task location as the next search location; in, For the t The remaining path at the iteration ij The amount of pheromones on , Respectively indicated by i The conversion time required to turn from the jth task location to the jth task location and the search path ij The degree of importance of the pheromone concentration on the deposit; Step 4.2.4: After the artificial ant selects the task location for the next search, it compares the current time with the latest searchable time of all remaining task locations. If there are still task locations that can be searched, it returns to step 4.2.3 to start the next round of selection; otherwise, it goes to step 4.2.5; Step 4.2.5: The artificial ant returns to the initial task location and completes the construction of a task location search path. At this time, if there are still task locations that have not been searched, it returns to step 4.2.2 and starts a new round of search; otherwise, it completes the construction of a new solution, and the number of iterations is t Increase 1 ， Go to step 4.2.6; Step 4.2.6: For a new solution, first determine whether it is a feasible solution that satisfies all constraints. If so, go to step 4.2.7; otherwise, go to step 4.2.9; Step 4.2.7: For a new feasible solution, update the pheromone according to formula (22); in, is the pheromone evaporation coefficient, is the pheromone residue constant; Step 4.2.8: If the optimization objective value of the feasible solution is better than , then update , and go to step 4.2.10; otherwise, go directly to step 4.2.10; Step 4.2.9: For a new infeasible solution, update the pheromone according to formula (23). （23） Step 4.2.10: After completing the pheromone update, make a judgment. If the number of iterations is t Less than the maximum number of iterations of the ant colony algorithm , then go to step 4.2.11, otherwise go to step 4.2.12; Step 4.2.11: If it is a subproblem OM3 , then the search time extension operator is called to complete the shortest time set required to search for the task location Make adjustments and go to step 4.2.2; otherwise, go directly to step 4.2.2; the search time extension operator refers to an operator used to increase the search time of the i-th task location based on probability: Step 4.2.12: After the search is completed, output and the corresponding feasible solution.

6. A method for allocating tasks for searching targets at sea in a large area according to claim 5, characterized in that: The operator that increases the search time of the i-th task location based on probability is: in, It indicates the probability of increasing the search time when the i-th task location is selected. When the i-th task location is selected, the increase in the search time at the task location is , represents the congestion of the search time window of the i-th task location, represents the i-th task location and the j The length of overlap of the search time windows for each task location.

7. A method for allocating tasks for searching targets at sea in a large area according to claim 6, characterized in that: The global pheromone update method is: When any ant colony algorithm finds a new global optimal solution, the global pheromone matrix is ​​updated using formula (28); in, For the t In the iteration, the i-th task location is j The amount of global pheromone remaining on the search path of the task location, is the residual amount of pheromone, = .

8. A method for allocating tasks for searching targets at sea in a large area according to claim 7, characterized in that: The method of using the neighborhood search algorithm in step 4.8 to further optimize the current global optimal solution set is: For each solution in the global optimal solution set, suppose there is a solution rn task location search paths, denoted as , the number of task locations contained in each path is , the neighborhood search algorithm is used for each search path to adjust the order of searching for task locations in the search path. If a better path than the current search path is found after the adjustment, the current search path is used to replace it, and the neighborhood search is performed again on the search path until all search paths have been searched through the neighborhood search and no better search path can be found.

9. A method for allocating tasks for searching targets at sea in a large area according to claim 8, characterized in that: The multi-objective aggregation strategy refers to: f is a single solution in the global optimal solution set, , , are the user's preferences for the three optimization objectives, and are adjusted according to user needs. is the maximum value of optimization objective 1 in the global optimal solution set, The maximum value of the optimization objective 2 is the global optimal solution. The maximum value of optimization objective 3 is the global optimal solution. To optimize the minimum value of objective 3, To optimize the minimum value of objective 2, is the minimum value of optimization objective 1.

10. A large-area marine target search task allocation system, characterized in that: Use the steps of a large-area marine target search task allocation method as described in any one of claims 1 to 9.

Images