A polar satellite resource scheduling method based on genetic algorithm and permutation representation

Abstract

The invention provides a polar satellite resource scheduling method based on genetic algorithm and permutation representation, which represents a resource scheduling scheme as a permutation combination of task IDs, defines task features, converts the task combination of the resource scheduling scheme from priority to a schedule, applies a genetic algorithm to select the best resource scheduling scheme, and uses the best resource scheduling scheme to generate a next-generation resource scheduling scheme until a satisfactory resource scheduling scheme is found or a stopping standard is reached; adopts a genetic algorithm and a permutation-based representation method to iteratively update a resource scheduling scheme set, selects the best resource scheduling scheme from it, and efficiently optimizes the resource scheduling scheme, and explores different genetic operators and selection strategies to improve the performance of the algorithm; solves multi-resource polar satellite scheduling, and realizes effective optimization of polar satellite resource scheduling schemes.

Description

Technical Field

[0001] The invention relates to the field of multi-resource polar satellite scheduling, and in particular to a polar satellite resource scheduling method based on genetic algorithm and permutation representation. Background Art

[0002] Genetic Algorithm (GA) is an optimization algorithm that simulates the natural evolution process and is used to solve complex optimization problems. It is based on Darwin's theory of evolution and genetics theory, and generates the optimal solution by simulating the natural processes of inheritance, mutation, selection, etc. The basic idea of ​​GA is to search for the optimal solution by encoding the candidate solutions (also called individuals) and using genetic operations to simulate the natural evolution process. The advantage of GA is that it can find approximate optimal solutions in large-scale search spaces, and the constraints on the problems are relatively loose, which makes it suitable for a variety of complex optimization problems. However, GA also has some disadvantages, such as slow convergence speed, low requirements on the mathematical properties of the problem, and difficult parameter setting. In practical applications, GA is widely used to solve complex optimization problems such as the traveling salesman problem, feature selection in machine learning, parameter optimization, and design of machine learning models. By simulating the evolutionary process, GA can automatically search the potential solution space and find a better solution.

[0003] Satellite resource scheduling refers to the process of rationally allocating and scheduling resources in a satellite system to maximize system performance and benefits. Satellite refers to satellite resources including bandwidth, power, antenna pointing, time, etc., and the goal of satellite resource scheduling is to achieve optimal utilization of resources while meeting user needs and system constraints. Satellite resource scheduling is a complex optimization problem that requires comprehensive consideration of multiple factors such as mission requirements, resource constraints, and system performance. Effective resource scheduling can improve the performance and benefits of satellite systems, meet user needs, and provide reliable communications and services.

[0004] Polar satellites are satellites that orbit the Earth's poles. The orbits of these satellites cross the Earth's equator perpendicularly, so they pass over the Earth's North and South Poles. The orbits of polar satellites enable them to provide extensive observation and monitoring capabilities of high-latitude regions. They can be used in meteorological observations, earth science research, environmental monitoring, glacier research, ocean observations, communication transmissions and other fields. Due to the special nature of polar satellite orbits, they are able to provide continuous observations of polar regions, unaffected by the Earth's rotation. Summary of the invention

[0005] In order to solve the technical problem that resource scheduling for polar satellites is lacking and highly restricted in the prior art, the present invention provides a polar satellite resource scheduling method based on genetic algorithm and permutation representation. The scheduling solution can be efficiently optimized by using genetic algorithm and permutation and combination-based representation, and the exploration of different genetic operators and selection strategies improves the performance of the algorithm.

[0006] The specific scheme of the present invention is as follows:

[0007] A polar satellite resource scheduling method based on genetic algorithm and permutation representation, comprising:

[0008] S1: Setting resource scheduling scheme: Setting a resource scheduling scheme for a task combination based on the priority order of the task requests; each task request corresponds to a unique task ID, and the task combination is an arrangement of the task IDs;

[0009] S2: define task characteristics: set a satellite resource and visibility window for each task; the visibility window is task attribute information, which stores a list of the earliest start time EST and the latest end time LFT of the satellite resource;

[0010] S3: converting the arrangement order of the task combination of the resource scheduling scheme from priority to a schedule: using a greedy scheduler to convert the task combination, the conversion method is: selecting a task that has not been scheduled from the task combination, and placing it in the first position that meets the requirements in the schedule by default;

[0011] S4: Apply genetic algorithm to select the best resource scheduling solution:

[0012] S41: Evaluate the suitability of the resource scheduling scheme: calculate the total overlapping duration of tasks in the resource scheduling scheme, and evaluate the suitability of the resource scheduling scheme according to the total overlapping duration to obtain a suitability value;

[0013] S42: applying a genetic operator: using sequential crossover as a genetic operator to apply to the resource scheduling solution set to generate a new resource scheduling solution set;

[0014] S43: Selecting the best resource scheduling solution: selecting the best resource scheduling solution from the resource scheduling solution set in S42 according to the fitness value;

[0015] S5: Generate a next-generation resource scheduling solution: Generate a next-generation resource scheduling solution using the optimal resource scheduling solution obtained in S43;

[0016] S6: Repeating process: Repeating steps S4-S5 until a stopping criterion is reached, wherein the stopping criterion is that the total overlap duration reaches a duration threshold or the number of iterations reaches an iteration threshold.

[0017] Preferably, the resource scheduling scheme for setting the task combination in S1 includes:

[0018] S11: assign a unique ID to each task request and represent the resource scheduling scheme as an arrangement of task IDs;

[0019] S12: Arrange the task IDs in a specific order to form a priority queue.

[0020] Preferably, the task characteristics are defined in S2, and each task request includes the following task characteristics: satellite resources required for the task, task duration, earliest start time EST, and latest end time LFT.

[0021] Preferably, the earliest start time EST and the latest end time LFT of the satellite are set according to the polar satellite mission; the visibility window for executing the mission request is set using the earliest start time and the latest end time, and is stored in the visibility window described in S2.

[0022] Preferably, the step of converting the arrangement order of the task combinations from priorities to a schedule in S3 includes:

[0023] A greedy scheduler is used to convert a priority queue based on an arrangement into a schedule; the greedy scheduler takes the arrangement of tasks as input and schedules them in a way that minimizes the total overlap time; when the greedy scheduler encounters a task conflict, the task is immediately added to the first position in the schedule that meets the requirements.

[0024] Preferably, the total overlapping duration of tasks in the computing resource scheduling scheme in S4 is calculated as follows:

[0025] Fit(x)=∑overlap(x)

[0026] Among them, x represents the scheduling scheme, x i represents the i-th task in the solution, Fit(x) represents the total overlap duration, overlap(x i ) represents the overlapping duration of the i-th task.

[0027] Preferably, in S41, the sequential crossover is used as a genetic operator set resource scheduling scheme to generate a new resource scheduling scheme set, including:

[0028] A41: Randomly select Z tasks from the permutation corresponding to parent 2;

[0029] A42: Reorder the selected Z tasks in parent 1 so that they have the same relative order as they appear in parent 2;

[0030] A43: Copy the reordered Z tasks to the child, with the positions coming from parent 1 and the order coming from parent 2; then pass the inconsistent tasks randomly selected from parent 1 and parent 2 directly to the child, generate a new resource scheduling plan and add it to the resource scheduling plan set, and generate a new resource scheduling plan set.

[0031] Preferably, the method for generating the next generation resource scheduling solution in S5 is:

[0032] A method combining an elite system and a tournament selection is used to generate a next-generation resource scheduling solution; the elite system is used to directly add the best of the best resource scheduling solutions to the next-generation resource scheduling solution without mutation; the tournament selection is used to select the best solution from a random subset of the best resource scheduling solutions and add it to the next-generation resource scheduling solution. Beneficial effects of the present invention:

[0033] The invention provides a polar satellite resource scheduling method based on genetic algorithm and permutation representation, S1: setting a resource scheduling scheme: setting a resource scheduling scheme of a task combination based on the priority order of task requests; each task request corresponds to a unique task ID, and the task combination is the arrangement of the task IDs; S2: defining task characteristics: setting a satellite resource and a visibility window for each task; the visibility window is task attribute information, and saves a list of the earliest start time EST and the latest end time LFT of the satellite resource; S3: converting the arrangement order of the task combination of the resource scheduling scheme from priority to a schedule: using a greedy scheduler to implement conversion on the task combination, the conversion method is: selecting a task that has not been scheduled from the task combination, and putting it in the first position that meets the requirements in the schedule by default; S4: applying a genetic algorithm to select the best resource scheduling scheme. Aiming at a multi-resource polar satellite scheduling scheme, the present invention adopts a task ID arrangement and combination method to represent and define task characteristics, uses a greedy scheduler to convert the arranged priority queue into a schedule, and evaluates the suitability of the schedule by calculating the total overlapping duration of tasks in the schedule; then uses a genetic algorithm to update the population to efficiently optimize the resource scheduling scheme, and the exploration of different genetic operators and selection strategies improves the performance of the algorithm; solves the multi-resource polar satellite scheduling and realizes the effective optimization of the polar satellite resource scheduling scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A flow chart of a polar satellite resource scheduling method based on genetic algorithm and permutation representation provided by the present invention.

[0035] Figure 2 A polar satellite resource request graph in a polar satellite resource scheduling method based on genetic algorithm and permutation representation provided by the present invention.

[0036] Figure 3 The present invention provides a Gantt chart of resource allocation requests in a polar satellite resource scheduling method based on a genetic algorithm and arrangement representation.

[0037] Figure 4 The invention provides a Gantt chart for optimal resource allocation in a polar satellite resource scheduling method based on a genetic algorithm and permutation representation. DETAILED DESCRIPTION

[0038] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0039] Example:

[0040] Figure 1 A flow chart of a polar satellite resource scheduling method based on genetic algorithm and permutation representation provided by the present invention is as follows: Figure 1 As shown, a polar satellite resource scheduling method based on genetic algorithm and permutation representation includes:

[0041] S1: Representing the resource scheduling scheme as a permutation and combination of task IDs: Each task request has a unique ID, and the resource scheduling scheme is a permutation and combination of these IDs, and the permutation and combination method is a priority queue.

[0042] Specifically, the method of expressing the solution as a permutation and combination of task IDs in the above S1 is:

[0043] S11: Assign a unique ID to each task request and represent the solution as a permutation of task IDs;

[0044] S12: Arrange the task IDs in a specific order to form a priority queue.

[0045] It should be noted that, in the polar satellite resource scheduling method based on genetic algorithm and arrangement representation provided in this embodiment, the task arrangement is an initial arrangement at the beginning, and it is not used as a parent generation of the genetic algorithm, and a greedy scheduler is required to rearrange it.

[0046] S2: Define mission characteristics: set available satellite resources and visibility windows for the mission; convert ranking-based priority queues into schedules;

[0047] Specifically, the mission characteristics are defined in S2 above, and each mission request includes the following mission characteristics: satellite resources required for the mission, mission duration, earliest start time (EST), and latest end time (LFT).

[0048] Specifically, the earliest start time (EST) and the latest end time (LFT) of the satellite are set according to the polar satellite mission; the visibility window for executing the mission request is set using the earliest start time and the latest end time, and is stored in the time window list in the above S2.

[0049] Exemplarily, each mission request contains the following features: polar satellite resources required for the mission, mission duration, earliest start time (EST), latest end time (LFT); the earliest start time and the latest end time generally depend on the satellite orbit and the time when the satellite passes a specific location on the earth. The earliest start time and the latest end time determine the visibility window in which the mission request needs to be executed. The mission may fill the entire window, or the window may be larger than the mission duration. A request usually requests multiple resources on different polar satellites, which can execute the request at different times.

[0050] like Figure 2 As shown in Figure 1, three tasks make time requests for the same resource. The time window of request 8 has no flexibility: task 8 must start at the earliest start time and end at the latest end time. On the other hand, the durations of request 3 and request 4 allow task 3 and task 4 to move in time. Task 3 and task 4 can be scheduled on this resource without conflict, but task 8 conflicts with task 3 and task 4.

[0051] like Figure 3 As shown, there are 10 resources to be scheduled, each row is a resource, time is on the x-axis, and all requests are shown in all possible locations. The goal is to schedule each request in a location while minimizing conflicts and overlaps.

[0052] S3: Convert the permutation-based priority queue described in S2 into a schedule using a greedy scheduler; tasks in the permutation queue are set at the first requested position that is still available;

[0053] Specifically, the above S3 converts the permutation-based priority queue into a schedule, including: using a greedy scheduler to convert the permutation-based priority queue into a schedule; the greedy scheduler takes the permutation of tasks as input and attempts to schedule in a manner that minimizes the total overlap time; when the scheduler encounters a task conflict, the task is immediately added to the schedule to minimize overlap.

[0054] For example, in an ideal schedule, there are no conflicting tasks and the polar satellite visibility window can be fully utilized. However, a conflict-free schedule is almost impossible to achieve. In some embodiments, a greedy scheduler is used to convert a finite queue based on an arrangement into a schedule. The greedy scheduler takes the arrangement of tasks as input and attempts to schedule in a way that minimizes the total overlap time. Assume that the task arrangement is [9, 2, 8, 7, 5, 4, 1, 3, 0, 6]. At the beginning, the schedule is empty. Assume that [9, 2, 8, 7, 5] is directly put into the schedule without conflict; but task 4 conflicts with existing tasks. When the scheduler encounters a task conflict, it immediately adds the task to the schedule to minimize overlap. This will produce more conflicts, but less overlap.

[0055] S4: Apply genetic algorithm to select the best resource scheduling solution.

[0056] S41: Evaluate the suitability of the resource scheduling scheme: calculate the total overlapping duration of tasks in the resource scheduling scheme, and evaluate the suitability of the resource scheduling scheme according to the total overlapping duration to obtain a suitability value;

[0057] Specifically, the calculation of the total overlapping duration of tasks in the resource scheduling scheme in S41 above includes: a specific formula for calculating the total overlapping duration is as follows:

[0058] Fit(x)=Σ0erlap0xi)

[0059] Among them, x represents the scheduling scheme, x i represents the i-th task in the solution, Fit(x) represents the total overlap duration, overlap(x i ) represents the overlapping duration of the i-th task.

[0060] Exemplarily, the suitability of a schedule is evaluated by considering several factors. First, the schedule must satisfy the time window of each task request. If a task request cannot be completed within its time window, it is considered a failure. Second, the schedule must minimize the total overlap time between conflicting tasks. If two or more tasks overlap in time, the overlap time is added to the total overlap time. The goal is to minimize the total overlap time while maximizing the number of tasks that are completed within their respective time windows. Third, the schedule must make efficient use of available resources. If resources are idle for a long time, it is considered a waste. Our goal is to maximize resource utilization while minimizing the total overlap time and the number of failed task requests. The suitability of the schedule is calculated as a weighted sum of these factors, where the weights are determined by the user based on the specific requirements of the scheduling problem. The genetic algorithm uses this fitness function to evaluate the quality of each candidate solution and generates new solutions using genetic operators such as crossover and mutation.

[0061] S42: applying a genetic operator: using sequential crossover as a genetic operator to apply to the resource scheduling solution set to generate a new resource scheduling solution set;

[0062] Specifically, the application of the genetic operator in S42 above uses sequential crossover as the genetic operator to apply to the resource scheduling solution set, including:

[0063] Randomly select Z randomly selected tasks from the permutation corresponding to parent 2;

[0064] Reorder the selected Z tasks in parent 1 to have the same relative order as they appear in parent 2;

[0065] The reordered Z tasks will be copied to the children, with the positions coming from parent 1 and the order coming from parent 2; tasks in parent 1 that are inconsistent with the randomly selected tasks in parent 2 will be passed directly to the children.

[0066] Exemplarily, using sequential crossover as the genetic operator, sequential crossover randomly selects Z randomly selected tasks in the permutation corresponding to parent 2. The selected Z tasks are then reordered in parent 1 to be in the same relative order as they appear in parent 2. These reordered Z tasks are copied to the offspring, with the positions from parent 1 and the order from parent 2. Tasks in parent 1 that are inconsistent with the randomly selected tasks in parent 2 are passed directly to the offspring. Therefore, tasks passed directly from parent 1 to the offspring are inherited according to absolute position. Assume parent 1: A, B, C, D, E, F, G, H, I, J. Parent 2: C, F, J, E, H, B, A, D, I, G, Assume tasks F, B, A, G are selected from parent 2, then the offspring are: F, B, C, D, E, A, G, H, I, J.

[0067] S43: Select the best resource scheduling solution: select the best resource scheduling solution from the resource scheduling solution set in S42 according to the fitness value.

[0068] S5: Generate a next-generation resource scheduling solution: Generate a next-generation resource scheduling solution using the optimal resource scheduling solution obtained in S43.

[0069] Exemplarily, a method combining meritocracy and tournament selection is used to generate the next generation of resource scheduling solutions. Meritocracy is a selection strategy that retains the best solutions from one generation to the next, and the best solution of the previous generation is always included in the next generation to ensure that the best solution is not lost. Tournament selection is a selection strategy that selects the best solution from a random subset of the population. In some embodiments, tournament selection is used to select the remaining solutions for the next generation, and the tournament size is a parameter that determines the number of solutions participating in each tournament; the larger the tournament size, the greater the selection pressure, and the selected solutions will use genetic operators such as crossover and mutation to generate new offspring solutions.

[0070] S6: Repeating process: Repeating steps S4-S5 until a stopping criterion is reached, wherein the stopping criterion is that the total overlap duration reaches a duration threshold or the number of iterations reaches an iteration threshold.

[0071] For example, Figure 4 As shown, it is an optimized arrangement, where each request is arranged in one position.

[0072] It should be noted that the above-described specific implementations can enable those skilled in the art to more fully understand the invention, but do not limit the invention in any way. Therefore, although this specification has described the invention in detail with reference to the drawings and embodiments, those skilled in the art should understand that the invention can still be modified or replaced by equivalents. In short, all technical solutions and improvements that do not deviate from the spirit and scope of the invention should be included in the protection scope of the patent for the invention.

Claims

1. A polar satellite resource scheduling method based on genetic algorithm and permutation representation, characterized in that: include: S1: Setting resource scheduling scheme: Setting a resource scheduling scheme for a task combination based on the priority order of the task requests; each task request corresponds to a unique task ID, and the task combination is an arrangement of the task IDs; S2: define task characteristics: set a satellite resource and visibility window for each task; the visibility window is task attribute information, which stores a list of the earliest start time EST and the latest end time LFT of the satellite resources; each task contains the following task characteristics: satellite resources required for the task, task duration, earliest start time EST, and latest end time LFT; S3: converting the arrangement order of the task combination of the resource scheduling scheme from priority to a schedule: using a greedy scheduler to convert the task combination, the conversion method is: selecting a task that has not been scheduled from the task combination, and placing it in the first position that meets the requirements in the schedule by default; S4: Apply genetic algorithm to select the best resource scheduling solution: S41: Evaluate the suitability of the resource scheduling scheme: calculate the total overlapping duration of tasks in the resource scheduling scheme, and evaluate the suitability of the resource scheduling scheme according to the total overlapping duration to obtain a suitability value; S42: applying a genetic operator: using sequential crossover as a genetic operator to apply to the resource scheduling solution set to generate a new resource scheduling solution set; S43: Selecting the best resource scheduling solution: selecting the best resource scheduling solution from the resource scheduling solution set in S42 according to the fitness value; S5: Generate a next-generation resource scheduling solution: Generate a next-generation resource scheduling solution using the optimal resource scheduling solution obtained in S43; S6: Repeating process: Repeating steps S4-S5 until a stopping criterion is reached, wherein the stopping criterion is that the total overlap duration reaches a duration threshold or the number of iterations reaches an iteration threshold.

2. A polar satellite resource scheduling method based on genetic algorithm and permutation representation according to claim 1, characterized in that: The resource scheduling scheme for setting the task combination described in S1 includes: S11: assign a unique ID to each task request and represent the resource scheduling scheme as an arrangement of task IDs; S12: Arrange the task IDs in a specific order to form a priority queue.

3. A polar satellite resource scheduling method based on genetic algorithm and permutation representation according to claim 1, characterized in that: The earliest start time EST and the latest end time LFT of the satellite are set according to the polar satellite mission; the visibility window for executing the mission request is set using the earliest start time and the latest end time, and is stored in the visibility window described in S2.

4. A polar satellite resource scheduling method based on genetic algorithm and permutation representation according to claim 1, characterized in that: In S3, the arrangement order of the task combination is converted from priority to a schedule, including: A greedy scheduler is used to convert a priority queue based on an arrangement into a schedule; the greedy scheduler takes the arrangement of tasks as input and schedules them in a way that minimizes the total overlap time; when the greedy scheduler encounters a task conflict, the task is immediately added to the first position in the schedule that meets the requirements.

5. A polar satellite resource scheduling method based on genetic algorithm and permutation representation according to claim 1, characterized in that: The method for calculating the total overlapping duration of tasks in the resource scheduling scheme in S41 is: Among them, x represents the scheduling scheme, x i represents the i-th task in the solution, Fit(x) represents the total overlap duration, overlap(x i ) represents the overlapping duration of the i-th task.

6. A polar satellite resource scheduling method based on genetic algorithm and permutation representation according to claim 1, characterized in that: In S42, the sequential crossover is used as a genetic operator set resource scheduling solution set to generate a new resource scheduling solution set, including: A41: Randomly select Z tasks from the permutation corresponding to parent 2; A42: Reorder the selected Z tasks in parent 1 so that they have the same relative order as they appear in parent 2; A43: Copy the reordered Z tasks to the child, with the positions coming from parent 1 and the order coming from parent 2; then pass the inconsistent tasks randomly selected from parent 1 and parent 2 directly to the child, generate a new resource scheduling plan and add it to the resource scheduling plan set, and generate a new resource scheduling plan set.

7. A polar satellite resource scheduling method based on genetic algorithm and permutation representation according to claim 1, characterized in that: The method for generating the next generation resource scheduling solution in S5 is: A method combining elite system and tournament selection is used to generate the next generation resource scheduling scheme; the elite system is used to directly add the best of the best resource scheduling schemes to the next generation resource scheduling scheme without mutating them; the tournament selection is used to select the best scheme from a random subset of the best resource scheduling schemes and add it to the next generation resource scheduling scheme.

Images