Multi-satellite task scheduling method based on genetic algorithm

Abstract

The invention discloses a multi-satellite task scheduling method based on a genetic algorithm, and belongs to the technical field of satellite task scheduling. Establishing a scheduling model based on satellite and ground station visible window data, and integrating a scheduling period constraint, a task uniqueness constraint, an equipment protection time constraint and a frequency band and orbit type matching constraint by taking task income maximization as a target; a simulated annealing local search mechanism is introduced into the genetic algorithm, local optimum is jumped out through a probability acceptance inferior solution, and a self-adaptive strategy of mutation probability is dynamically adjusted according to population fitness change; setting an early stop mechanism to terminate iteration in advance when the fitness is continuously not improved; and finally generating a scheduling scheme. According to the method, the problems that a traditional genetic algorithm is prone to falling into local optimum and constraint processing is rigid are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of satellite mission scheduling, and in particular relates to a multi-satellite mission scheduling method based on genetic algorithm. Background Art

[0002] With the surge in the number of satellites and the increasing complexity of measurement and control requirements, traditional manual scheduling methods are inefficient. Deterministic algorithms struggle to handle large-scale, complex constraints, and standard genetic algorithms are prone to getting stuck in local optima. Existing technologies suffer from the following shortcomings: insufficient constraint handling capabilities, failing to consider practical limitations such as equipment protection time and minimum measurement and control duration; slow algorithm convergence and a lack of early stopping mechanisms; and a lack of adaptive strategies, resulting in low search efficiency.

[0003] In the prior art, for example, Chinese patent application No. 202311431652.6 discloses a satellite mission scheduling method and device based on an improved quantum genetic algorithm. The shortcomings of this patent are that it adopts a fixed mutation probability and only adjusts the evolution direction through a quantum rotating gate, which cannot improve the robustness of the algorithm. In addition, this patent only uses the maximum number of iterations as the termination condition, and lacks efficiency optimization design. Summary of the Invention

[0004] Based on the above analysis, the present invention provides a multi-satellite task scheduling method based on a genetic algorithm. The specific implementation scheme is as follows: A multi-satellite task scheduling method based on a genetic algorithm includes establishing a task scheduling model based on satellite data, ground station data, satellite task data, and visible window data, and performing resource constraints on the task scheduling model; setting and inputting the population size, number of iterations, crossover probability, and initial mutation probability in the genetic algorithm, including the following steps:

[0005] S1: Initialization operation:

[0006] Check the completeness of input data and remove missing values and outliers;

[0007] Setting a threshold value to eliminate the visible window data whose duration is shorter than the threshold value;

[0008] Set heuristic rules and create initial chromosomes, each chromosome represents a complete satellite mission scheduling solution;

[0009] S2: Population evolution:

[0010] Perform selection operations;

[0011] Perform crossover operations on individuals in the population;

[0012] The visible window data allocated in each chromosome is changed by random replacement to complete the mutation operation;

[0013] S3: Optimize operations:

[0014] Combined with simulated annealing algorithm to realize local search;

[0015] Adaptively adjust the mutation rate;

[0016] S4: Determine whether the evolution process has met the stopping condition. When the set maximum number of iterations is reached, it is determined that the stopping condition is met.

[0017] S5: Output the final fitness value, obtain the correspondence between satellite tasks and visible windows from the chromosome results, and generate a scheduling plan.

[0018] Preferably, the start and end time of the task scheduling cycle are set and represented by T s and T e Indicates that the measurement and control resource task scheduling period is [T s ,T e Clarifying time constraints helps to better manage and allocate time resources during the scheduling process, avoid resource conflicts, and improve the accuracy and feasibility of scheduling.

[0019] Preferably, the satellite data includes: S = {s l |l=1,2,…,M},

[0020]

[0021] in Indicates the satellite code, Indicates the priority of the satellite, Indicates the frequency band required by the satellite, and M indicates the number of satellites;

[0022] The ground station data includes: G = {g j |j=1,2,…,N},

[0023]

[0024] in Indicates the ground station code, Indicates the equipment protection time of the ground station. Indicates the frequency band that the ground station antenna can operate, and N indicates the number of ground stations;

[0025] The satellite mission data includes: T = {t i |i=1,2,…,P}, satellite missions are converted from satellite requirements, P represents the number of missions, where,

[0026]

[0027] in, Indicates the mission code. Indicates the code of the corresponding satellite, Indicates the lift track type of the task. Indicates the minimum duration of the task. Indicates the lower limit of the mission's highest elevation angle, Indicates the date of the task, Indicates the benefits of completing the task;

[0028] The visible window data includes: T represents task t i and ground station g j The number of visible windows between

[0029]

[0030] in The code representing the visible window, Indicates the satellite code corresponding to the visible window. Indicates the ground station code corresponding to the window, Indicates the start time of the visible window, Indicates the end time of the visible window, Represents the maximum pitch angle of the visible window, Indicates the type of rail for visible windows.

[0031] Preferably, a decision variable is used to indicate whether the task scheduling model is executed:

[0032]

[0033]

[0034] Here, maxf(s) represents the total benefit of all executed tasks, and s represents the scheduling solution. Introducing decision variables makes the scheduling model more flexible, enabling dynamic adjustments to task execution based on actual conditions, thus enhancing the adaptability and optimization capabilities of the scheduling solution.

[0035] Preferably, the resource constraints include:

[0036] Scheduling period constraint: All satellite missions and available resources should be within the scheduling period:

[0037]

[0038]

[0039] The duration of continuous measurement and control of satellite missions must meet the minimum mission duration:

[0040]

[0041] Satellite mission uniqueness constraint: each mission can only be executed once:

[0042]

[0043] Due to device uniqueness constraints, the ground station antenna can only communicate with one satellite at a time:

[0044]

[0045] Satellite tracking and control uniqueness constraint: a satellite can only communicate with one ground station antenna at a time:

[0046]

[0047] Due to device characteristics, the ground station needs to maintain its own device protection time when performing adjacent tasks:

[0048]

[0049] Other attribute constraints, among which the track type constraint is a soft constraint:

[0050]

[0051] Track Type Constraints:

[0052]

[0053] Preferably, the selection operation in S2 is performed using a tournament method.

[0054] Preferably, the crossover operation in S2 is performed in a uniform crossover manner.

[0055] Preferably, the simulated annealing algorithm in S3 includes:

[0056] Set the initial temperature T0 and the end temperature T f , cooling rate α;

[0057] Perform a local search:

[0058] a. Initialize the current temperature T = T0;

[0059] b. When T>T f Perform the following steps:

[0060] i. Replace the visible window according to the heuristic rules required by the satellite mission to generate the neighborhood solution S′;

[0061] ii. Calculate the fitness difference ΔE = f(S′) - f(S) between the new solution S′ and the current solution S;

[0062] iii. Acceptance criteria:

[0063] If ΔE>0, directly accept the new solution S=S′;

[0064] If ΔE≤0, accept the new solution with probability P=exp(ΔE / T);

[0065] iv. Update the global optimal solution;

[0066] vi. Lower the temperature: T = T × α.

[0067] Preferably, the adaptive adjustment of the mutation rate in S3 includes:

[0068] Record the best fitness value of the last k generations: f best (tk),f best (t-k+1),...,f best (t); the best fitness value refers to the fitness value of the individual with the highest fitness in each generation in the first k generations;

[0069] Adaptive adjustment formula for mutation rate:

[0070] P m (t+1)={min(P max ,P m (t)×1.2), when IR≤0.01

[0071] max(P min ,P m (t)×(1-0.1×IR)), when IR>0.01};

[0072]

[0073] P min =0.01,P max =0.1;

[0074] Among them, P m refers to the mutation rate, t refers to the tth generation, P m (t+1) indicates the adaptive mutation rate adjustment before entering the next generation, and IR indicates the improvement rate.

[0075] Preferably, an early stopping mechanism is added in S4, wherein the early stopping mechanism terminates the iterative process if the fitness value has not improved for N consecutive generations, where N is set according to the data scale and the test situation.

[0076] The beneficial effects of the present invention are:

[0077] This paper proposes a multi-satellite mission scheduling method based on genetic algorithm. It mainly performs problem modeling and constraint abstraction for actual measurement and control scenarios, combines simulated annealing and adaptive strategies with genetic algorithm to overcome the defect that the algorithm is prone to falling into local optimality, and adds an early stopping mechanism in the iterative process to generate a satellite mission scheduling plan that meets the constraints within an acceptable time range. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0079] Figure 1 This is a flow chart of a multi-satellite mission scheduling method based on genetic algorithm. DETAILED DESCRIPTION

[0080] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the specific implementation methods described here are only used to illustrate and explain the present invention and are not used to limit the present invention.

[0081] like Figure 1 As shown, input: satellite data, ground station data, satellite mission data, visible window data, population size, number of iterations, crossover probability, initial mutation probability;

[0082] Output: Final scheduling plan

[0083] Step 1: Initialization operation:

[0084] Step 1.1: Check the integrity and consistency of the data and delete missing values and outliers to ensure the quality of the data and avoid errors caused by data problems in subsequent calculations.

[0085] Step 1.2: Set a threshold and remove visible window data with a duration shorter than this value. This will filter out short-term tasks that are not worth considering and reduce computational complexity.

[0086] Step 1.3: Create initial chromosomes. Each chromosome represents a complete satellite mission scheduling solution. When initializing the population, set heuristic rules to sort satellite missions according to their benefits, prioritizing high-payoff tasks. Construct an initial solution space, improve the quality of the population's initial solutions, and increase the probability of the algorithm converging to a high-quality solution.

[0087] Step 2: Population evolution:

[0088] Step 2.1: Perform a selection operation and select the tournament method as the selection operation strategy. The tournament method selects individuals with higher fitness from the current population as parents, ensuring that high-quality genes are passed to the next generation while maintaining randomness to avoid premature convergence.

[0089] Step 2.2: Perform a crossover operation on the individuals in the population, using a uniform crossover method. By uniform crossover, the parent chromosomes are recombined to enhance population diversity and explore new feasible solution spaces.

[0090] Step 2.3: For the individuals in the current population, randomly replace the visible window assigned in the solution to complete the mutation operation. Randomly replacing the visible window prevents the algorithm from falling into the local optimum and maintains the exploration ability of the population.

[0091] Step 3: Optimize operations:

[0092] Step 3.1: Local search optimization, combined with simulated annealing algorithm to achieve local search:

[0093] Set the initial temperature T0 and the end temperature T f , cooling rate α;

[0094] Perform a local search:

[0095] a. Initialize the current temperature T = T0; set the starting temperature of simulated annealing, determine the width of the initial search and the probability of accepting inferior solutions.

[0096] b. When T>T f Perform the following steps:

[0097] i. Replace the visible window according to the heuristic rules required by the satellite mission to generate the neighborhood solution S′;

[0098] ii. Calculate the fitness difference ΔE = f(S′) - f(S) between the new solution S′ and the current solution S;

[0099] iii. Acceptance criteria:

[0100] If ΔE>0, directly accept the new solution S=S′;

[0101] If ΔE≤0, accept the new solution with probability P=exp(ΔE / T);

[0102] iv. Update the global optimal solution;

[0103] vi. Lower the temperature: T = T × αc. Return the optimal solution obtained by local search.

[0104] Step 3.2: Adaptively adjust the mutation rate, track the changes in the optimal fitness of the population in the last k generations, increase the mutation rate when the fitness stagnates, and reduce the mutation rate when the fitness continues to improve. Set the maximum and minimum values of the adjustment to ensure that the mutation is within a reasonable range.

[0105] Specific operations of adaptive mutation rate:

[0106] Record the best fitness value of the last k generations: f best (tk),f best (t-k+1),...,f best (t); the best fitness value refers to the fitness value of the individual with the highest fitness in each generation in the first k generations.

[0107] Adaptive adjustment formula for mutation rate:

[0108] P m (t+1)={min(P max ,P m (t)×1.2), when IR≤0.01

[0109] max(P min ,P m (t)×(1-0.1×IR)), when IR>0.01};

[0110]

[0111] P min =0.01,P max =0.1.

[0112] Among them, P m refers to the mutation rate, t refers to the tth generation, P m (t+1) indicates the adaptive mutation rate adjustment before entering the next generation, and IR indicates the improvement rate.

[0113] Step 4: Determine whether the evolutionary process has met the stopping criteria. This criteria is considered met when the maximum number of iterations has been reached, ensuring that the algorithm terminates within a reasonable timeframe and avoiding infinite loops or excessive computation. Additionally, an early stopping mechanism is implemented. When fitness has not improved for N consecutive generations, iterations are stopped. This improves algorithm efficiency, avoids wasting computing resources on ineffective searches, and allows for early termination when the search becomes stagnant. N is set based on the data size and test results.

[0114] Step 5: Output the final result. After the execution is completed, save the fitness value generated during the execution process to a file so that the algorithm optimization effect can be viewed. Convert the chromosome results generated by the algorithm execution into a scheduling plan.

[0115] The embodiments of the present invention are described in detail above. The description of the above embodiments is only used to help understand the method of the present invention and its core concept. At the same time, for those skilled in the art, according to the concept of the present invention, there may be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A multi-satellite task scheduling method based on a genetic algorithm, comprising establishing a task scheduling model based on satellite data, ground station data, satellite task data, and visible window data, and applying resource constraints to the task scheduling model; setting and inputting a population size, number of iterations, crossover probability, and initial mutation probability into the genetic algorithm, characterized in that: S1: Initialization operation: Check the completeness of input data and remove missing values and outliers; Setting a threshold value to eliminate the visible window data whose duration is shorter than the threshold value; Set heuristic rules and create initial chromosomes, each chromosome represents a complete satellite mission scheduling solution; S2: Population evolution: Perform selection operations on individuals in the population; Perform crossover operations on individuals in the population; The visible window data allocated in each chromosome is changed by random replacement to complete the mutation operation; S3: Optimize operations: Combined with simulated annealing algorithm to realize local search; Adaptively adjust the mutation rate; S4: Determine whether the evolution process has met the stopping condition. When the set maximum number of iterations is reached, it is determined that the stopping condition is met. S5: Output the final fitness value, obtain the correspondence between satellite tasks and visible windows from the chromosome results, and generate a scheduling plan.

2. The multi-satellite task scheduling method based on genetic algorithm according to claim 1, characterized in that: Set the start and end time of the task scheduling cycle and use T s and T e Indicates that the measurement and control resource task scheduling period is [T s ,T e ]express.

3. The multi-satellite task scheduling method based on genetic algorithm according to claim 1, characterized in that: The satellite data includes: S = {s l |l=1,2,…,M}, in Indicates the satellite code, Indicates the priority of the satellite, Indicates the frequency band required by the satellite, and M indicates the number of satellites; The ground station data includes: G = {g j |j=1,2,…,N}, in Indicates the ground station code, Indicates the equipment protection time of the ground station. Indicates the frequency band that the ground station antenna can operate, and N indicates the number of ground stations; The satellite mission data includes: T = {t i |i=1,2,…,P}, satellite missions are converted from satellite requirements, P represents the number of missions, where, in, Indicates the mission code. Indicates the code of the corresponding satellite, Indicates the lift track type of the task. Indicates the minimum duration of the task. Indicates the lower limit of the mission's highest elevation angle, Indicates the date of the task, Indicates the benefits of completing the task; The visible window data includes: T represents task t i and ground station g j The number of visible windows between in The code representing the visible window, Indicates the satellite code corresponding to the visible window. Indicates the ground station code corresponding to the window, Indicates the start time of the visible window, Indicates the end time of the visible window, Represents the maximum pitch angle of the visible window, Indicates the type of rail for visible windows.

4. The multi-satellite task scheduling method based on genetic algorithm according to claim 1 is characterized in that , use decision variables to indicate whether the task scheduling model is executed: Among them, maxf(s) represents the benefits of all executed tasks, and s represents the scheduling plan.

5. The multi-satellite task scheduling method based on genetic algorithm according to claim 1, characterized in that: The resource constraints include: Scheduling period constraint: All satellite missions and available resources should be within the scheduling period: The duration of continuous measurement and control of satellite missions must meet the minimum mission duration: Satellite mission uniqueness constraint: each mission can only be executed once: Due to device uniqueness constraints, the ground station antenna can only communicate with one satellite at a time: Satellite tracking and control uniqueness constraint: a satellite can only communicate with one ground station antenna at a time: Due to device characteristics, the ground station needs to maintain its own device protection time when performing adjacent tasks: Other attribute constraints: Angle constraint: Frequency band constraints: Track Type Constraints:

6. The multi-satellite task scheduling method based on genetic algorithm according to claim 1, characterized in that ,The selection operation described in S2 adopts the tournament method to select.

7. The multi-satellite task scheduling method based on genetic algorithm according to claim 1 is characterized in that ,The crossover operation described in S2 adopts a uniform crossover ,method to perform the crossover operation.

8. The multi-satellite task scheduling method based on genetic algorithm according to claim 1 is characterized in that ,The simulated annealing algorithm described in S3 includes: Set the initial temperature T0 and the end temperature T f , cooling rate α; Perform a local search: a. Initialize the current temperature T = T0; b. When T>T f Perform the following steps: i. Replace the visible window according to the heuristic rules required by the satellite mission to generate the neighborhood solution S′; ii. Calculate the fitness difference ΔE = f(S′) - f(S) between the new solution S′ and the current solution S; iii. Acceptance criteria: If ΔE>0, directly accept the new solution S=S′; If ΔE≤0, accept the new solution with probability P=exp(ΔE / T); iv. Update the global optimal solution; vi. Lower the temperature: T = T × α.

9. The multi-satellite task scheduling method based on genetic algorithm according to claim 1, characterized in that: The adaptive adjustment of the mutation rate described in S3 includes: Record the best fitness value of the last k generations: f best (tk),f best (t-k+1),...,f best (t); the best fitness value refers to the fitness value of the individual with the highest fitness in each generation in the first k generations; Adaptive adjustment formula for mutation rate: P m (t+1)={min(P max ,P m (t)×1.2), when IR≤0.01 max(P min ,P m (t)×(1 - 0.1×IR)), when IR > 0.01}; P min =0.01,P max =0.1; Among them, P m refers to the mutation rate, t refers to the tth generation, P m (t+1) indicates the adaptive mutation rate adjustment before entering the next generation, and IR indicates the improvement rate.

10. The multi-satellite task scheduling method based on genetic algorithm according to claim 1, characterized in that: An early stopping mechanism is added to S4. The early stopping mechanism terminates the iteration process if the fitness value does not improve for N consecutive generations, where N is set according to the data scale and test situation.

Images