Method for scheduling hyperspectral satellite multi-satellite regional imaging task

Abstract

The application provides a hypersensitive satellite multi-satellite regional imaging task scheduling method, comprising the following steps: step 1, analyzing hypersensitive satellite multi-satellite regional imaging task requirements, and simplifying the task scheduling process; step 2, taking the shortest total length of strips as the target, performing strip decomposition on the region to be imaged, and obtaining an atomic task set for task scheduling; step 3, according to the task set of step 2, selecting corresponding multi-class decision variables, taking maximizing imaging coverage benefits and minimizing imaging task execution time as the target, and constructing a multi-satellite imaging task scheduling model; step 4, solving the constructed multi-satellite imaging task scheduling model; and step 5, combining the optimal decision variable group obtained in step 4 into a task scheduling scheme. The application effectively adapts to the optimization solving requirement of the multi-class decision variable problem, and can be effectively applied to hypersensitive satellite multi-satellite regional imaging task scheduling.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of satellite remote sensing, and particularly relates to a scheduling method for multi-satellite regional imaging tasks of hypersensitive agile satellites. BACKGROUND

[0002] In recent years, with the development of space technology, agile satellites have begun to develop towards hypersensitive agile satellites with more flexible three-axis attitude maneuvering capabilities (pitch, roll and yaw). Compared with traditional agile satellites, hypersensitive agile satellites greatly increase the data acquisition capability of imaging satellites. At the same time, the networking observation of multiple hypersensitive agile satellites can further improve the data acquisition efficiency and timeliness of remote sensing satellites. However, since hypersensitive agile satellites have the capability of imaging while maneuvering (attitude maneuvering while imaging), their imaging mode is no longer limited to being parallel to the subsatellite track, resulting in a large number of satellite resource combination modes and single-satellite task imaging sequences for multi-satellite imaging task scheduling. Therefore, how to reasonably schedule the task sequences of multiple hypersensitive agile satellites to efficiently complete the coverage of regional targets is a new challenge faced by current multi-satellite joint task scheduling.

[0003] Existing imaging task scheduling modeling methods for hypersensitive agile satellites mainly focus on the task sequence planning of single-satellite imaging tasks. Yang et al. proposed a constraint optimization model and designed a multi-objective differential evolution algorithm based on hybrid coding in the 2018 paper "The Bi-objective Active-Scan Agile Earth Observation Satellite Scheduling Problem: Modeling and Solution Approach". The integrated optimization modeling method of "satellite resource optimization configuration" + "single-satellite imaging task sequence planning" for hypersensitive agile satellite multi-satellite imaging tasks has not been studied by relevant scholars. Existing scheduling model solving methods mainly focus on the research of population evolution algorithms, among which PSO and its improvements are widely used solving algorithms with different particle velocity and position updating strategies for different types of decision variables, such as continuous decision variable updating methods (iwPSO, CLPSO, DNLPSO, SLPSO), binary decision variable updating methods (BPSO) and discrete decision variable updating methods (GPSO, BBPSO). However, the multi-satellite imaging task scheduling optimization model has multiple types of decision variables, and there is no relevant research on solving algorithms for mixed imaging task scheduling models with multiple types of decision variables. Therefore, it is necessary to study the related imaging task scheduling model and solving algorithm for the joint observation requirements of hypersensitive agile satellites. SUMMARY

[0004] The present application aims at the deficiencies of the prior art, and provides a scheduling method for hypersensitive satellite multi-satellite regional imaging tasks.

[0005] To solve the above technical problems, the present application adopts the following technical solutions:

[0006] A scheduling method for hypersensitive satellite multi-satellite regional imaging tasks, comprising the following steps:

[0007] Step 1: analyzing the hypersensitive satellite multi-satellite regional imaging task requirements, and simplifying the task scheduling process;

[0008] Step 2: taking the shortest total length of strips as the target, performing strip decomposition on the imaging region to obtain an atomic task set for task scheduling;

[0009] Step 3: according to the task set of step 2, selecting corresponding multi-class decision variables, and constructing a multi-satellite imaging task scheduling model with the target of maximizing imaging coverage benefits and minimizing imaging task execution time;

[0010] Step 4: solving the constructed multi-satellite imaging task scheduling model;

[0011] Step 5: combining the optimal decision variables obtained in step 4 into a task scheduling scheme.

[0012] Further, the simplification of the multi-satellite imaging task scheduling process in step 1 comprises:

[0013] (1) the hypersensitive satellite only images the generated atomic task set when performing imaging;

[0014] (2) the atomic task can only be imaged by one of the multiple hypersensitive earth observation satellites, and the maximum number of observations required by the atomic task is 1;

[0015] (3) the constraint conditions of the task scheduling model consider the satellite attitude conversion capability constraint and the imaging time window constraint of the atomic task;

[0016] (4) it is assumed that the solar elevation angle satisfies the optical imaging requirement within the satellite imaging time, and the interference of weather conditions is not considered.

[0017] Further, the decision variables in step 3 include strip star selection number, strip imaging sequence number, strip imaging direction number, and normalized coefficient of strip endpoint observation time.

[0018] Further, the construction of the multi-satellite imaging task scheduling model in step 3 comprises:

[0019] The primary objective function maximizing the imaging coverage benefit is constructed as:

[0020]

[0021] where M is the number of satellites, n is the number of strips, Y i,j represents whether the jth strip in the strip imaging sequence of the ith satellite determined by the decision variable can complete imaging, Y i,j ∈{0,1}, if Y i,j = 1, the jth strip in the strip imaging sequence of the ith satellite can complete imaging, otherwise, imaging cannot be achieved; G i,j represents the imaging coverage benefit of the jth strip in the strip imaging sequence of the ith satellite determined by the decision variable; when the imaging task is a point target, G i,j represents the ratio of the number of point targets covered by the strip to the corresponding point target, when the imaging task is a surface target, G i,j represents the ratio of the effective coverage area of the region target covered in the strip to the corresponding region target, when the imaging task is a line target, G i,j represents the ratio of the length of the line target covered in the strip to the total length of the corresponding line target.

[0022] and the secondary objective function minimizing the imaging task execution time d is constructed as:

[0023]

[0024] where M is the number of satellites, n is the number of strips, ΔT i,k represents the imaging time interval between adjacent two strip endpoints in the strip endpoint imaging sequence of the ith satellite determined by the decision variable, ΔT i,k = T i,k = T i,k-1 , k represents the kth strip endpoint in the strip endpoint imaging sequence of the ith satellite determined by the decision variable. Y_end i,k represents the attitude conversion time constraint between adjacent two strip endpoints in the strip endpoint imaging sequence of the ith satellite determined by the decision variable.

[0025] Further, the constraint condition of one-to-one correspondence between the decision variable and observation is established for the multi-satellite imaging task scheduling model in step 3, and the constraint condition includes imaging time window constraint and attitude conversion time constraint,

[0026] where the imaging time window constraint is:

[0027]

[0028] The imaging time window constraint represents that the imaging time of the kth strip end point in the strip end point imaging sequence of the ith satellite determined by the decision variable must satisfy the imaging time window constraint, where T i,k is the imaging time of the kth strip end point in the strip end point imaging sequence of the ith satellite determined by the decision variable;[ITW i,k-s , ITW i,k-e is the imaging time window of the kth strip end point in the strip end point imaging sequence of the ith satellite determined by the decision variable, ITW i,k-s is the start time of the imaging time window of the strip end point, ITW i,k-e is the end time of the imaging time window of the strip end point;

[0029] The attitude conversion time constraint is:

[0030]

[0031] The attitude conversion time constraint represents that the difference ΔT i,k between the imaging times of two adjacent strip end points in the strip end point imaging sequence of the ith satellite determined by the decision variable must satisfy the attitude conversion time constraint f_T i,k .

[0032] Further, the multi-satellite imaging task scheduling model is solved by using the improved particle swarm algorithm in step 4.

[0033] Further, the method for solving the multi-satellite imaging task scheduling model by using the improved particle swarm algorithm comprises the following steps:

[0034] Step 3.1, randomly initialize population individuals:

[0035] Set the population size to P, generate P initial populations, each individual represents a multi-satellite imaging task scheduling scheme, and the maximum number of iterations is Iters; wherein the position of the particle represents the decision variable in the multi-satellite imaging task scheduling model constructed in step 3; the initial value of each particle is randomly selected within its respective value range, and the initial speed of the particle is set to 0;

[0036] Step 3.2, calculate the multi-satellite imaging task scheduling scheme of each individual in the current population, and obtain the individual with the optimal fitness value in the current population, wherein the fitness value is the imaging coverage benefit and the imaging task execution time;

[0037] Step 3.3, update the speed and position of the population individuals;

[0038] In this step, a mixed updating strategy of the speed and position of the population individuals is adopted, specifically:

[0039] For continuous decision variables, the particle position is updated after the improved PSO algorithm is used to update the particle velocity;

[0040] For binary decision variables, the updated particle position is obtained after the updated particle velocity is activated by using the Sigmoid activation function;

[0041] For discrete decision variables, the updated particle position is obtained by using the BBPSO algorithm;

[0042] Step 3.4 performs optimization iteration to obtain a multi-satellite imaging task scheduling scheme with maximum imaging coverage revenue and minimum imaging task execution time;

[0043] If the iteration termination condition is not met, steps 3.2 and 3.3 are executed multiple times, and the fitness values of the optimal particles in the current iteration and the last iteration are compared. If the fitness value of the optimal individual in the current iteration is better, the optimal multi-satellite imaging task scheduling scheme is obtained, otherwise, the optimal multi-satellite imaging task scheduling scheme is obtained from the optimal individual in the last iteration.

[0044] The iteration process is repeated until the maximum number of iterations Iters is reached, and the optimal fitness value is obtained. The optimal multi-satellite imaging task scheduling scheme is obtained.

[0045] Further, the method for obtaining the individual with the optimal fitness value in the current population in step 3.2 is as follows:

[0046] The fitness values F iter,u of the particles in the current population are sorted to obtain the individual pbest with the optimal fitness value in the improved particle swarm optimization algorithm, i.e., the individual with the maximum imaging coverage revenue and the minimum imaging task execution time. iter,pbest ;

[0047] Wherein, pbest is the individual with the optimal fitness value in the iter iteration of the improved particle swarm optimization algorithm, F iter,u represents the fitness value of the u-th individual in the P population in the iter iteration of the improved particle swarm optimization algorithm, F iter,pbest represents the fitness value of the optimal individual pbest in the iter iteration of the improved particle swarm optimization algorithm.

[0048] The judgment of the fitness value is as follows: two individuals A and B in the population are compared. The imaging coverage revenue of the two individuals is compared first. If the imaging coverage revenue of A is greater than that of B, the fitness value of A is better than that of B. If the imaging coverage revenues of A and B are equal, the imaging task execution time of A and B is compared. If the imaging task execution time of A is less than that of B, the fitness value of A is better than that of B.

[0049] Another object of the present application is to provide a scheduling system for implementing the scheduling method of the above-mentioned hypersensitive satellite multi-satellite regional imaging task, comprising a multi-satellite imaging task scheduling model construction module and a multi-satellite imaging task scheduling model solving module, wherein,

[0050] The multi-satellite imaging task scheduling model construction module is used to construct a multi-satellite imaging task scheduling model comprising multiple types of decision variables, and taking the maximization of imaging coverage benefits and the minimization of imaging task execution time as the objective function, and taking the imaging time window constraint and the attitude conversion time as the constraint condition;

[0051] The multi-satellite imaging task scheduling model solving module is used to solve the constructed multi-satellite imaging task scheduling model by using the improved particle swarm optimization algorithm, and determine the multi-satellite imaging task scheduling scheme.

[0052] Compared with the prior art, the present application has the following beneficial effects:

[0053] The present application realizes one-to-one correspondence with the multi-satellite imaging task scheduling scheme by taking the normalized coefficients of the strip star selection number, the strip imaging sequence number, the strip imaging direction number and the strip endpoint imaging time as the decision variables, and realizes the integrated optimization modeling of "satellite resource optimization configuration" and "single-satellite imaging task sequence planning", and also effectively avoids the problems of complex multi-satellite imaging task scheduling scheme evaluation calculation process and low optimization solving efficiency caused by directly taking the imaging time as the decision variable;

[0054] The present application constructs the objective function taking the maximization of imaging coverage benefits and the minimization of imaging task execution time, and realizes the conversion from user demand to modeling; further, for the three types of decision variables existing in the constructed multi-satellite imaging task scheduling model in the present application, discrete decision variables (p j,iter,u ), continuous decision variables (q i,iter,u , t i,k,iter,u ) and binary decision variables (s i,j,iter,u ), the present application adopts a mixed population individual speed and position updating strategy, effectively overcomes the premature convergence problem of standard particle swarm, and also effectively adapts to the optimization solving demand of the existence of discrete, continuous and binary decision variables, and the present application can be effectively applied to the multi-satellite regional imaging task scheduling of hypersensitive satellites. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 The flowchart of the scheduling method of the hypersensitive satellite multi-satellite regional imaging task of the present application embodiment;

[0056] Figure 2 The regional target schematic diagram of the present application embodiment;

[0057] Figure 3 A strip decomposition result diagram of the embodiment of the present application;

[0058] Figure 4 An imaging coverage yield result diagram of the embodiment of the present application;

[0059] Figure 5 An imaging task execution time result diagram of the embodiment of the present application. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0061] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0062] The present application will be further described in conjunction with specific embodiments, but is not limited to the present application.

[0063] Referring to Figure 1 The embodiment of the present application provides a scheduling method for multi-satellite regional imaging tasks of hypersensitive satellites, comprising the following steps:

[0064] Step 1, analyzing the demand for multi-satellite regional imaging tasks of hypersensitive satellites, and simplifying the task scheduling process;

[0065] Before modeling, the user demand and the task scheduling demand need to be analyzed and decomposed to realize the conversion of the imaging demand of multi-satellite imaging tasks to the elements of the task scheduling model. The input of the model is determined according to the imaging task, the completion time, the sensor type, etc., including the imaging task position and the orbit resource within the specified time; the decision variable of the model is determined according to the satellite resource demand and the input of the model; and the objective function of the model is determined according to the user demand.

[0066] Meanwhile, since the multi-satellite imaging task scheduling is an NP-hard problem, in order to obtain a multi-satellite imaging task scheduling scheme for hypersensitive satellites, the embodiment defines and simplifies the multi-satellite imaging task scheduling process as follows:

[0067] (1) Hypersensitive satellite refers to a new generation of hypersensitive satellite with the ability of rolling, pitching and yawing three-axis maneuvering, and the ability of imaging while performing attitude maneuvering;

[0068] (2) The satellite only images the generated atomic task set when performing imaging, without considering other tasks;

[0069] (3) The atomic task can only be imaged by one of the multiple hypersensitive agile satellite observation satellites, and the maximum number of times the atomic task needs to be observed is 1;

[0070] (4) The constraint conditions of the task scheduling model mainly consider the satellite attitude conversion capability constraint and the imaging time window constraint of the atomic task;

[0071] (5) It is assumed that the solar elevation angle within the satellite imaging time meets the optical imaging requirement, and the on-board storage and energy are sufficient.

[0072] (6) It is assumed that each satellite has only one payload;

[0073] (7) The interference of weather conditions is not considered.

[0074] Step 2. To obtain the atomic task set of the task scheduling, the imaging area is decomposed into strips with the shortest total length of strips as the target;

[0075] In this embodiment, the multi-satellite imaging task scheduling model needs to be constructed on the basis of the imaging strip as the atomic task set, so the multi-satellite imaging task needs to be preprocessed into imaging strips as the input of the model. In this step, the optimal decomposition direction of each region target is calculated with the shortest total length of strips as the target, and each imaging area is decomposed into a group of parallel strips as the atomic task set of the task scheduling model.

[0076] Step 3. According to the task set of step 2, select the corresponding multi-class decision variables, and construct a multi-satellite imaging task scheduling model with the target of maximizing the imaging coverage benefit and minimizing the imaging task execution time;

[0077] After completing the conversion of the multi-satellite imaging task demand into the elements of the imaging task scheduling model, the reasonable definition and simplification of the model in step 1, and taking the imaging strip of step 2 as the atomic task set for model construction, in this step, the selected multi-class decision variables include:

[0078] The first type of decision variable is the strip selection satellite number:

[0079] p j ∈{1,2,...,M}

[0080] Where M represents the number of satellites, j represents the jth strip in the atomic task set, j ∈ {1, 2, …, n}, n is the number of strips in the atomic task set, p j represents the strip selection satellite number, which is used to determine which satellite of the M satellites images the jth strip in the atomic task set;

[0081] The second type of decision variable is the strip imaging sequence number:

[0082]

[0083] Where M represents the number of satellites, i represents the i-th satellite, and q i The strip imaging sequence is numbered to determine the order of the decision variable p. j Given the imaging sequence of the stripes contained in the i-th satellite, the imaging sequence of the stripes contained in each satellite can be determined using the two decision variables mentioned above.

[0084] Third type of decision variable, strip imaging direction number:

[0085]

[0086] Where j represents the decision variable p j and q i The j-th strip in the determined strip imaging sequence of the i-th satellite, where n is the number of strips, s i,j Indicates the decision variable p j and q i The strip imaging direction number of the j-th strip in the determined strip imaging sequence of the i-th satellite is used to determine the decision variable p. j and q i The imaging direction of the j-th strip in the jointly determined strip imaging sequence can then be used to jointly determine a set of strip endpoint imaging sequences through the above three decision variables;

[0087] The fourth type of decision variable is the normalized coefficient at the observation time of the strip endpoints:

[0088]

[0089] Where k represents the k-th strip endpoint in the strip endpoint imaging sequence of the i-th satellite, determined by the three decision variables mentioned above, n is the number of strips, 2*n represents the number of strip endpoints, and t i,k The normalization coefficient represents the observation time of the k-th strip endpoint in the strip endpoint imaging sequence of the i-th satellite, which is determined by the three decision variables mentioned above. It is used to determine the imaging time of the k-th strip endpoint in the strip endpoint imaging sequence.

[0090] After determining the decision variables, the objective function is constructed to maximize the imaging coverage benefit of the region and minimize the task execution time while maintaining the same coverage benefit:

[0091] Primary objective function:

[0092]

[0093] The primary objective function is to maximize the imaging coverage benefit, ensuring the maximum coverage of the imaging task. Wherein, M is the number of satellites, n is the number of strips, Y i,j represents whether the jth strip in the strip imaging sequence of the ith satellite determined by the decision variable can complete imaging, Y i,j ∈{0,1}, if Y i,j =1, the jth strip in the strip imaging sequence of the ith satellite can complete imaging, otherwise, imaging cannot be achieved. G i,j represents the imaging coverage benefit of the jth strip in the strip imaging sequence of the ith satellite determined by the decision variable. When the imaging task is a point target, G i,j represents the ratio of the number of point targets covered by the strip to the corresponding point target, when the imaging task is a surface target, G i,j represents the ratio of the effective coverage area of the region target covered in the strip to the corresponding region target, when the imaging task is a line target, G i,j represents the ratio of the length of the line target covered in the strip to the total length of the corresponding line target.

[0094] Secondary objective function:

[0095]

[0096] The secondary objective function is to minimize the imaging task execution time, ensuring that the satellite can complete the imaging of the task in the least time. Wherein, M is the number of satellites, n is the number of strips, ΔT i,k represents the imaging time interval between adjacent two strip endpoints in the strip endpoint imaging sequence of the ith satellite determined by the decision variable, ΔT i,k =T i,k =T i,k-1 , k represents the kth strip endpoint in the strip endpoint imaging sequence of the ith satellite determined by the decision variable. Y_end i,k represents the attitude conversion time constraint between adjacent two strip endpoints in the strip endpoint imaging sequence of the ith satellite determined by the decision variable.

[0097] The constraint conditions corresponding one-to-one to the decision variables are established, including:

[0098] Imaging time window constraint:

[0099]

[0100] The constraint condition is the imaging time window constraint, which represents that the imaging time of the kth strip endpoint on the strip endpoint imaging sequence of the ith satellite determined by the decision variable must satisfy the imaging time window constraint, wherein, T i,kis the imaging time of the kth strip end point in the strip end point imaging sequence of the ith satellite determined by the decision variable. i,k-s ,ITW i,k-e is the imaging time window of the kth strip end point in the strip end point imaging sequence of the ith satellite determined by the decision variable, ITW i,k-s is the start time of the imaging time window of the strip end point. i,k-e is the end time of the imaging time window of the strip end point.

[0101] Attitude transition time constraint:

[0102]

[0103] The constraint is the attitude transition time constraint, which indicates that the difference (ΔT i,k ) between the imaging times of the adjacent two strip end points in the strip end point imaging sequence of the ith satellite determined by the decision variable must satisfy the attitude transition time constraint (f_T i,k ) between the adjacent two strip end points.

[0104] Step 4, solving the constructed multi-satellite imaging task scheduling model;

[0105] After reasonable assumptions and definitions are made and the corresponding multi-satellite regional imaging task scheduling model for hypersensitive satellites is constructed through steps 1, 2 and 3, the next step is to solve the model. Since there are different types of decision variables in the imaging task scheduling model constructed in this embodiment, such as discrete decision variables (p j ), continuous decision variables (q i , t i,k ), binary decision variables (s i,j ), the particle swarm optimization algorithm for solving real number problems cannot be simply used to solve the scheduling model in this embodiment. In this embodiment, the particle swarm optimization algorithm is improved, and the improved particle swarm optimization algorithm is used to solve the scheduling model. The specific implementation is as follows:

[0106] Step 3.1, randomly initialize population individuals:

[0107] Set the population size to P, generate P initial populations, each individual represents a multi-satellite imaging task scheduling scheme, and the maximum number of iterations is Iters;

[0108] wherein the position of the particle represents the decision variable in the multi-satellite imaging task scheduling model constructed in step 2; that is, p j,iter,u , q i,iter,u , s i,j,iter,u , t i,k,iter,uThe initial value of each particle is randomly selected within its respective value range, and the initial velocity of the particle is set to 0;

[0109] where iter represents the ith iteration of the improved particle swarm algorithm, iter ∈ {1, 2, …, iter, …, Iters}, u represents the u th individual in the P populations in the ith iteration, u ∈ {1, 2, …, u, …, P}; p j,iter,u represents the band selection number of the j th band of the u th individual in the P populations in the ith iteration; q i,iter,u represents the band imaging sequence number of the band contained in the i th satellite of the u th individual in the P populations in the ith iteration; s i,j,iter,u represents the band imaging direction number of the j th band in the i th satellite of the u th individual in the P populations in the ith iteration; t i,k,iter,u represents the normalized time coefficient of the k th band end point in the i th satellite of the u th individual in the P populations in the ith iteration;

[0110] Step 3.2, calculate the multi-satellite imaging task scheduling scheme of each individual in the current population, and obtain the individual with the optimal fitness value in the current population, and the fitness value is the imaging coverage benefit and the imaging task execution time.

[0111] The multi-satellite imaging task scheduling scheme of each individual in the current population is calculated as follows:

[0112] According to the decision variables p j,iter,u , q i,iter,u , s i,j,iter,u , t i,k,iter,u , the multi-satellite imaging task scheduling scheme in the ith iteration process is determined by using the model in step 2.

[0113] The individual with the optimal fitness value in the current population is obtained as follows:

[0114] The fitness values F iter,u of the particles in the current population are sorted to obtain the individual pbest with the optimal fitness in the solving process of the improved particle swarm algorithm, i.e., the fitness value F iter,pbest of the individual corresponding to the maximum imaging coverage benefit and the minimum imaging task execution time.

[0115] where pbest is the individual with the optimal fitness in the ith iteration of the improved particle swarm algorithm, F iter,u represents the fitness value of the u th individual in the P populations in the ith iteration of the improved particle swarm algorithm, F iter,pbest represents the fitness value corresponding to the optimal individual pbest in the ith iteration of the improved particle swarm algorithm.

[0116] Wherein, the fitness value is judged as follows: comparing two individuals A and B in the population, the imaging coverage benefits of the two individuals are compared first, if the imaging coverage benefit of A is greater than that of B, the fitness value of A is better than that of B; if the imaging coverage benefits of A and B are equal, the imaging task execution time of A and B is compared, if the imaging task execution time of A is less than that of B, the fitness value of A is better than that of B;

[0117] Step 3.3, updating the speed and position of the population individual;

[0118] Since the present embodiment involves three types of decision variables, discrete variables (p j,iter,u ), continuous variables (q i,iter,u , t i,k,iter,u ), binary variables (s i,j,iter,u ), the present embodiment adopts a mixed updating strategy for the speed and position of the population individual, specifically:

[0119] For continuous decision variables (q i,iter,u , t i,k,iter,u ), after updating the particle speed by using the improved PSO algorithm, the particle position is updated;

[0120] v (iter+1) = c1*v (iter) + c2*(P win x (iter) - P lose x (iter)) + c3*(P center x (iter) - P lose x (iter))

[0121] P lose x (iter+1) = P lose x (iter) + w*v (iter+1)

[0122] Wherein, w represents the inertia weight, the value range is [0, 1], c1, c2, c3 represent the acceleration factor, the value range is [0, 1], v (iter), v (iter+1) are the last iteration speed and the current iteration speed of the particle respectively, P win x (iter), P lose x (iter) represent the last iteration position of the better particle and the last iteration position of the worse particle respectively, P lose x (iter+1) represents the current iteration position of the worse particle, P center x (iter) represents the current iteration average position of all particles.

[0123] For binary decision variables (s i,j,iter,u), the updated particle position is obtained after the updated particle velocity is activated by using a Sigmoid activation function;

[0124] v (iter+1) = c1*v (iter) + c2*(P win x(iter) - P lose x(iter)) + c3*(P center x(iter) - P lose x(iter))

[0125]

[0126]

[0127] wherein w represents an inertia weight, and the value range is [0, 1], c1, c2, c3 represent acceleration factors, and the value range is [0, 1], v (iter) and v (iter+1) represent the last iteration speed and the current iteration speed of the particle respectively, P win x(iter), P lose x(iter) represent the last iteration position of the better particle and the last iteration position of the worse particle respectively, P lose x(iter+1) represents the current iteration position of the worse particle, and P center x(iter) represents the last iteration average position of all particles. s (v (iter+1) ) is a Sigmoid activation function.

[0128] For discrete decision variables (p j,iter,u ), the position of the updated particle is updated by using the BBPSO algorithm;

[0129] P lose x(iter+1) = N (μ, δ)

[0130] μ = (P lose x(iter) + Gbest (iter) ) / 2

[0131] δ = |P lose x(iter) - Gbest (iter) |

[0132] wherein P lose x(iter+1) represents the current iteration position of the worse particle, the position of the particle satisfies a Gaussian distribution, μ is a mean value, and δ is a standard deviation, P lose x(iter) represents the last iteration position of the worse particle, and Gbest (iter) represents the last iteration optimal position of all particles.

[0133] Step 3.4. Iterative optimization is performed to obtain a multi-satellite imaging task scheduling scheme with maximum imaging coverage benefit and minimum imaging task execution time;

[0134] If the iterative termination condition is not met, steps 3.2 and 3.3 are performed multiple times, and the fitness values of the optimal particles in the current iteration and the last iteration are compared, i.e., F iter-1,pbest and F iter,pbest If F iter,pbest is better than F iter-1,pbest , the fitness value of the optimal individual in the current iteration process is the better multi-satellite imaging task scheduling scheme, otherwise, the fitness value of the optimal individual in the last iteration process is the better multi-satellite imaging task scheduling scheme;

[0135] The iterative process is repeated until the maximum number of iterations Iters is reached, thereby obtaining an individual with the optimal fitness value F Iters,pbest , which corresponds to the optimal multi-satellite imaging task scheduling scheme;

[0136] where pbest represents the individual with the optimal fitness value in the population in the iter-th iteration, F iter,pbest represents the fitness value corresponding to the optimal individual pbest in the iter-th iteration; F iter-1,pbest represents the fitness value corresponding to the optimal individual pbest in the (iter-1)-th iteration; and F Iters,pbest represents the fitness value corresponding to the optimal individual pbest when the maximum number of iterations Iters is reached, i.e., the optimal multi-satellite imaging task scheduling scheme.

[0137] Step 5. The optimal decision variable combination obtained in step 4 is restored to a task scheduling scheme;

[0138] In this step, the process of obtaining the scheduling scheme is as follows:

[0139] (1) Obtain the observation satellite number of all strips from the strip selection satellite number;

[0140] (2) Obtain the strip endpoint imaging sequence in a single satellite from the strip imaging sequence number and the push-broom sequence number;

[0141] (3) Obtain the observation start time and end time of each strip in the strip set from the strip endpoint imaging sequence and the normalized coefficient of the strip endpoint observation time;

[0142] (4) The two objective function values correspond to the imaging coverage benefit and the imaging task execution time of the optimal imaging scheme, respectively.

[0143] Further, the application also provides a hypersensitive satellite multi-satellite regional imaging task scheduling system for implementing the above method, comprising the following modules:

[0144] A multi-satellite imaging task scheduling model construction module constructs a multi-satellite imaging task scheduling model with the normalized coefficients of the strip selection satellite number, the strip imaging sequence number, the strip imaging direction number and the strip endpoint imaging time as decision variables, the maximum imaging coverage benefit and the minimum imaging task execution time as objective functions, and the imaging time window constraint and the attitude conversion time as constraint conditions;

[0145] A multi-satellite imaging task scheduling model solution module solves the constructed multi-satellite imaging task scheduling model by using an improved particle swarm optimization algorithm, and determines a multi-satellite imaging task scheduling scheme.

[0146] In order to more clearly illustrate the embodiment, the embodiment uses four hypersensitive satellites and a regional target imaging task to illustrate the specific implementation process of the above steps. The orbital parameters of each satellite are shown in Table 1, the latitude and longitude information of the regional target is shown in Table 2, and the imaging time window of each satellite is shown in Table 3. Figure 2

[0147] Table 1 Satellite parameters

[0148]

[0149]

[0150] Table 2 Regional target coordinate information

[0151]

[0152] As shown in Figure 3 , the regional target is converted into an imaging strip as the atomic task set of the model input through step 2. The imaging strip is input into the model, and the final multi-satellite imaging task scheduling scheme is obtained through the multi-satellite imaging task scheduling model construction of step 3 and the algorithm solving of step 4, as shown in Table 3. Figure 4 Figure 5

[0153]

[0154] As can be seen from Figure 4 Figure 5 and Table 3, the embodiment can realize the imaging of all atomic tasks, and the optimal imaging task execution time is 19.766s.

[0155] ​​​​The above merely preferred embodiments of the present application, and not therefore limit the embodiments and protection scope of the present application, for those skilled in the art, it should be realized that the equivalent replacement and obvious changes made by the application description, the resulting scheme should be included in the protection scope of the present application.

Claims

1. A scheduling method for hyperspectral satellite multi-satellite regional imaging task, characterized in that, The method comprises the following steps: Step 1, analyzing the multi-satellite regional imaging task demand of hypersensitive satellites, and simplifying the task scheduling process, comprising: (1) the hypersensitive satellite only images the generated atomic task set when performing imaging; (2) the atomic task can only be imaged by one of the multiple hypersensitive satellites, and the atomic task needs to be observed a maximum of 1 time; (3) the constraint condition of the task scheduling model considers the satellite attitude conversion capability constraint and the imaging time window constraint of the atomic task; (4) it is assumed that the solar elevation angle satisfies the optical imaging requirement within the satellite imaging time, and the interference of weather conditions is not considered; Step 2, taking the shortest total strip length as the target, performing strip decomposition on the region to be imaged to obtain the atomic task set of the task scheduling; Step 3, according to the task set of step 2, selecting corresponding multi-class decision variables, taking maximizing the imaging coverage benefit and minimizing the imaging task execution time as the target, and constructing a multi-satellite imaging task scheduling model; Step 4, solving the constructed multi-satellite imaging task scheduling model; Step 5, combining the optimal decision variables obtained in step 4 to restore the task scheduling scheme; Wherein, the method for solving the multi-satellite imaging task scheduling model comprises: Step 3.1, randomly initializing population individuals: Set the population size to P, generate P initial populations, each individual represents a multi-satellite imaging task scheduling scheme, and the maximum iteration number is Iters; wherein, the position of the particle represents the decision variable in the multi-satellite imaging task scheduling model constructed in step 3; the initial value of each particle is randomly selected within its respective value range, and the initial speed of the particle is set to 0; Step 3.2, calculate the multi-satellite imaging task scheduling scheme of each individual in the current population, and obtain the individual with the optimal fitness value in the current population, wherein the fitness value is the imaging coverage benefit and the imaging task execution time; Step 3.3, update the speed and position of the population individuals; In this step, a hybrid population individual speed and position updating strategy is adopted, specifically: For continuous decision variables, the particle position is updated after updating the particle speed by using the improved PSO algorithm; For binary decision variables, the updated particle position is obtained after activating the updated particle speed by using the Sigmoid activation function; For discrete decision variables, the position of the updated particle is obtained by using the BBPSO algorithm; Step 3.4, optimization iteration is performed to obtain a multi-satellite imaging task scheduling scheme with the maximum imaging coverage benefit and the minimum imaging task execution time; In the case where the iteration termination condition is not met, steps 3.2 and 3.3 are executed multiple times, and the fitness values of the optimal particles in the current iteration and the last iteration are compared; if the fitness value of the optimal individual in the current iteration process is better, the fitness value of the optimal individual in the last iteration process is the better multi-satellite imaging task scheduling scheme; Repeat the iteration process until the maximum iteration number Iters is set, so as to obtain the individual with the optimal fitness value, and the individual corresponds to the optimal multi-satellite imaging task scheduling scheme.

2. The scheduling method of claim 1, wherein, The decision variable in step 3 includes a strip selection star number, a strip imaging sequence number, a strip imaging direction number, and a normalized coefficient of a strip end point observation time.

3. The scheduling method of claim 1, wherein, The step 3 includes constructing a multi-satellite imaging task scheduling model. A primary objective function of maximizing imaging coverage benefits is constructed: ； In the formula, M For the number of satellites, n For the number of stripes, Y i,j The first decision variable is determined by the decision variable. i The first of the strip imaging sequences of the satellite j Can each strip complete an image? Y i,j If ∈{0,1}, Y i,j =1, then the first i The first of the strip imaging sequence of the satellite j Each stripe is required to complete imaging; otherwise, imaging is not possible. G i,j The first decision variable is determined by the decision variable. i The first of the strip imaging sequences of the satellite j The imaging coverage gain per strip; when the imaging task is a point target. G i,j This represents the ratio of the number of point targets covered by the strip to the number of corresponding point targets. When the imaging task is an area target, G i,j This represents the ratio of the effective coverage area of ​​the target area within the strip to the corresponding target area. When the imaging task is a line target, G i,j This represents the ratio of the length of the line targets covered within the strip to the total length of the corresponding line targets; And a secondary objective function of minimizing imaging task execution time d is constructed: ； in, M For the number of satellites, n For the number of stripes, The first decision variable is determined by the decision variable. i In the imaging sequence of the endpoints of the strips from a satellite, the imaging time interval between two adjacent endpoints is... , k The first decision variable is determined by the decision variable. i The first of the strip endpoint imaging sequences of the satellite k Each strip endpoint; Y_end i,k The characterization of the first decision variable is determined by the decision variable. i The attitude transition time constraint between two adjacent strip endpoints in the strip endpoint imaging sequence of a satellite.

4. The scheduling method of claim 1, wherein, In step 3, a constraint condition corresponding to the decision variable and one-to-one observation of the multi-satellite imaging task scheduling model is established, and the constraint condition includes an imaging time window constraint and an attitude conversion time constraint, The imaging time window constraint is: the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable i the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable k the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable T i,k the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable i the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable k the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable ITW i,k-s , ITW i,k-e the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable i the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable k the imaging time window constraint for the imaging time instant of the jth strip end in the strip end imaging sequence of the i th satellite determined by the decision variable ITW i,k-s the start time of the imaging time window for the strip end ITW i,k-e the end time of the imaging time window for the strip end The attitude conversion time constraint is: The attitude transition time constraint represents a difference between imaging instants of two adjacent strip end points on a strip end point imaging sequence of the GEO satellite i The attitude transition time constraint represents a difference between imaging instants of two adjacent strip end points on a strip end point imaging sequence of the GEO satellite The attitude transition time constraint represents a difference between imaging instants of two adjacent strip end points on a strip end point imaging sequence of the GEO satellite .

5. The scheduling method of claim 1, wherein, In step 4, the improved particle swarm algorithm is used to solve the multi-satellite imaging task scheduling model.

6. The method of claim 1, wherein, The specific method for obtaining the individual with the optimal fitness value in the current population in step 3.2 is: fitness value of the particle in the current population F iter,u sorting to obtain the individual with the optimal fitness value in the solving process of the improved particle swarm algorithm pbest i.e. the fitness value corresponding to the individual with the maximum imaging coverage revenue and the minimum imaging task execution time F iter,pbest ; wherein, pbest To improve the particle swarm algorithm in iter the individual with the optimal fitness value in the F iter,u denotes the optimal fitness value of the individual with the optimal fitness value in the iter the individual with the optimal fitness value in the P the individual with the optimal fitness value in the u the individual with the optimal fitness value in the F iter,pbest denotes the optimal fitness value of the individual with the optimal fitness value in the iter the individual with the optimal fitness value in the pbest the individual with the optimal fitness value in the Fitness values ​​are determined as follows: for two individuals in the population... A and B When comparing, prioritize comparing the imaging coverage gains of the two individuals. A The imaging coverage benefit is greater than B The imaging coverage benefit, then A Its fitness value is better than B If A and B If the imaging coverage gains are equal, then compare A and B The imaging task execution time, if A The imaging task execution time is less than B The imaging task execution time, then A Its fitness value is better than B .

7. A scheduling system of a scheduling method of a hyperspectral agile satellite multi-satellite regional imaging task, characterized in that, The method is suitable for implementing the scheduling method of the hypersatellite multi-satellite regional imaging task according to any one of claims 1-6.

Images