A Multi-objective Cooperative Scheduling Method for Multiple Aircraft in Space

Abstract

The present invention discloses a method for multi-object cooperative scheduling of multiple space vehicles. S1: Initialize the in-orbit satellites of our high-orbit and low-orbit, and establish a mathematical model of the in-orbit satellites of high-orbit and low-orbit by using the principle of six orbital elements. S2: Adopt a particle swarm optimization algorithm based on genetic operations, and perform optimal allocation of multiple targets under the joint observation of multiple equipment according to the target threat, resource constraint, and working period constraint conditions in the mathematical model. S3: Configure resource seizure optimization and perform repair. S4: Judge whether the current state reaches the detection expectation. If so, continuously track the target. If not, return to S2 until the tracking of the target is finally achieved. The present invention introduces high-and low-orbit satellites for target detection and scheduling design, improves the detection utilization rate of in-orbit satellites and the accuracy of target detection, as well as the correctness, rationality of the cooperative scheduling algorithm and the integrity of the full-process scheduling of the target under the condition of known target tracks.

Description

Technical Field

[0001] The present invention relates to the technical field of early warning satellite target detection, and in particular to a space multi-aircraft multi-target coordinated scheduling method. Background Art

[0002] Traditional synchronous orbit satellite detection technology has gradually matured, and the research focus has shifted from synchronous orbit early warning satellite research to Starlink technology. With the complexity and diversification of target detection application mission scenarios, the early warning system needs to achieve rapid tracking and positioning during the flight of missiles. Traditional synchronous orbit satellite technology and Starlink technology consume a lot of orbital satellite resources and cannot achieve effective utilization of space resources.

[0003] For example, patent CN113760506A provides an improved genetic algorithm for multi-satellite coordinated scheduling of earth observation methods. Although this method proposes an algorithm, it is not suitable for engineering practice, and does not provide simulation and data support for the accuracy and efficiency of the method. It only uses a single satellite mathematical model and a single detection method.

[0004] Patent CN112288289A provides a method for rapid planning of multi-satellite collaborative coverage for regional targets. It comprehensively considers the requirements of solution selectivity and rapid calculation, and adopts a method of simplifying complexity. By dividing the problem of "complete regional coverage as early as possible" into several sub-problems such as "each satellite observes the entire range of the region as early as possible", a local optimal solution is obtained for each sub-problem, so as to quickly obtain a global legal solution. On the basis of the legal solution, a global optimal solution is finally found through cyclic iteration. The calculation process of this method is relatively large and is not suitable for complex and changeable battlefield situations. The disadvantage of the above two methods is that they do not effectively utilize the resources of on-orbit satellites. There are certain problems. First, when the system tasks are complex, it is impossible to detect the target in a timely and effective manner, and the target will be lost and the tracking status will be poor. Second, the rapid planning method of multi-satellite collaborative coverage for regional targets can effectively solve the problem of detection efficiency, and the tracking status is also greatly improved, but the number of satellites required is large, and there will be a waste of resources, and the overall satellite utilization rate is not high. The overall efficiency and resource utilization of target tracking are not high. Therefore, how to provide a method for space multi-aircraft multi-target collaborative scheduling is an issue that technical personnel in this field urgently need to solve. Summary of the invention

[0005] An object of the present invention is to propose a multi-object collaborative scheduling method for multiple space vehicles. The present invention introduces high and low orbit satellites for target detection and scheduling design, which improves the detection utilization rate of on-orbit satellites and the accuracy of target detection, as well as the correctness and rationality of the collaborative scheduling algorithm and the integrity of the full-process scheduling of targets in the case of known target trajectories.

[0006] A multi-object collaborative scheduling method for multiple space vehicles according to an embodiment of the present invention is characterized by including the following method steps:

[0007] S1. Initialize the high and low orbit on-orbit satellites of our side, and establish a mathematical model of the high and low orbit on-orbit satellites by using the principle of six orbital elements.

[0008] S2. Adopt a particle swarm optimization algorithm based on genetic operations, and according to the target threat, resource constraints, and working period constraints in the mathematical model, perform optimal allocation of multiple targets under joint observation of multiple equipment, perform dynamic task re-planning when multiple targets change, and analyze the visible range and return frequency of each satellite according to the given observation area and observation duration requirements, and calculate the optimal scheduling plan for each satellite.

[0009] S3. Configure resource seizure optimization and repair, allocate the available satellite resources, and reasonably apply them to the actual scenario.

[0010] S4. Judge whether the current state reaches the detection expectation. If so, continuously track the target. If not, return to S2 until the target is finally tracked.

[0011] Preferably, the mathematical model in S1 includes a ground detection model algorithm and a limb detection model algorithm. The high orbit supports the ground detection model, and the low orbit supports both the ground detection and limb detection modes.

[0012] Preferably, the mathematical model based on six orbital elements in S1 includes establishing a satellite orbit dynamics model and analyzing the space environment perturbation factors, including the variation laws of satellites under the influence of the non-spherical perturbation of the earth, the third body perturbation, the solar radiation pressure perturbation, the atmospheric drag perturbation, and the earth shadow, and completing the orbital dynamics and kinematics modeling of three types of satellites: high orbit, low orbit, and geostationary orbit.

[0013] Among them, the satellite orbit dynamics equation described by classical orbital elements is:

[0014]

[0015] The dynamics equation of the orbital elements of the vernal equinox point is:

[0016]

[0017] Among them, for satellites in geostationary orbit, there is , the above formula can be further simplified as:

[0018]

[0019] In the formula, are the geocentric distance and the speed magnitude of the geostationary orbit satellite respectively, is the magnitude of the earth's angular velocity of rotation.

[0020] Preferably, the ground detection model in S1 is based on the range of the surface area that the detector can scan and observe, and the coverage range is used to describe the coverage characteristics of the field of view of the missile warning satellite detector on the ground and the constellation networking characteristics, and is measured by indicators such as global coverage rate and multiple coverage rate of key areas.

[0021] Preferably, the static period cooperative scheduling in S2 includes the setting of the objective function and the solution of the multi-variable and multi-constraint optimization problem, and the setting of the objective function includes the following four parts:

[0022] Objective function of the number of observed targets and their threat levels:

[0023] The objective function corresponds to design criteria one and two, and the function expression is as follows:

[0024]

[0025] In the formula, represents the number of targets successfully tracked within a certain allocation sub-period, represents the total number of targets existing in the sub-period system, represents whether the i-th target is successfully observed and tracked within the j-th allocation sub-period. If so, , otherwise , is the threat level value of the i-th target within the j-th allocation sub-period;

[0026] Objective function regarding the number of idle equipment in the system:

[0027] The objective function corresponds to design criterion three, and the function expression is as follows:

[0028]

[0029] In the formula, is the number of idle equipment, is the total number of equipment;

[0030] Objective function regarding the total arc length of equipment combination tracking:

[0031] The objective function corresponds to design criterion four, and the function expression is as follows:

[0032]

[0033] In the formula, is the current moment of the system, is the total duration during which the equipment combination can continuously observe target i starting from the current moment, is the moment when the target lands or the moment when the target moves out of the system;

[0034] Objective function regarding the number of sensor switches:

[0035] The objective function corresponds to Design Criterion Five, and the function expression is as follows:

[0036]

[0037] In the formula, represents the number of switches of all working sensors in two sub - cycles before and after static periodic scheduling, represents the maximum value of the number of sensor switches in all allocation schemes;

[0038] From the above four formulas, it can be comprehensively known that the total objective function of the static periodic scheduling algorithm is:

[0039]

[0040] In the formula, represents the weight of the corresponding objective function in the total objective function, and there is .

[0041] Preferably, for the solution of the multi - variable and multi - constraint optimization problem, a particle in the particle swarm corresponds to a solution of a problem in the solution space, and each particle has its own position, velocity, and cost evaluation index . When performing iterative update, according to the optimal solution of the particle itself and the optimal solution of all particles in the particle swarm to update the position and velocity of each particle;

[0042] Among them, the iterative update equation is:

[0043]

[0044] In the formula, , is the number of particles in the particle swarm; , is the dimension of the solution vector; k is the number of iterations; is the learning factor; is the weight; is a random number between [0, 1]. Preferably, for the guidance problem between high and low orbit satellites, a discrete clustering particle swarm algorithm needs to be adopted, which is described as follows:

[0045] Particle coding mechanism:

[0046] Now, the coding form of the particle is determined as:

[0047]

[0048] In the formula, represents the number of sensors in the current system equipment; the element represents the target number observed by the sensor. For example, if the sensor numbered 1 observes target 1, then , if the sensor numbered 1 is in an idle state, then ;

[0049] Particle swarm initialization mechanism;

[0050] Particle position update mechanism:

[0051] Introduce the relevant operations of the genetic algorithm into the particle swarm algorithm using the genetic algorithm, use the genetic operation as the particle update operator, and adopt the particle swarm optimization algorithm based on genetic operations

[0052] The dynamic mutation operator of the particle itself :

[0053] The inertia weight of the particle swarm optimization algorithm , the mutation probability decreases dynamically to achieve the balance between the random search and local search of the algorithm, representing the particle 's thinking about its own flight speed, and its formal description is as shown in the formula:

[0054]

[0055] In the formula, is a random number uniformly distributed on; is the probability of performing the mutation operation; represents the same mutation operation as in the genetic algorithm;

[0056] represents performing with probability the same mutation operation as represented by, otherwise ;

[0057] Particle and individual extreme value crossover operator :

[0058] The particle crosses with the individual extreme value with a given probability Perform crossover, and select the particle with better fitness after crossover as the updated particle, indicating the particle According to the individual extreme value Adjust the position, and its formal description is as shown in the formula:

[0059]

[0060] In the formula, is a random number uniformly distributed on the acceleration constant; represents the same crossover operation as in the genetic algorithm;

[0061] represents that with probability perform the same crossover operation as in the genetic algorithm represented by, otherwise ;

[0062] Particle and global extreme value crossover operator :

[0063] The particle performs crossover with the global extreme value with a given probability and its formal description is as shown in the formula:

[0064]

[0065] In the formula, is a random number uniformly distributed on the acceleration constant; represents the same crossover operation as in the genetic algorithm;

[0066] represents that with probability perform the same crossover operation as in the genetic algorithm represented by, otherwise ;

[0067] Combining the above formula, the particle position formula can be obtained as:

[0068] .

[0069] Preferably, the particle swarm initialization mechanism is a priority-based particle swarm initialization mechanism, and its specific process is as follows:

[0070] Step1. Calculate the priority value for all subtasks within the scheduling window ;

[0071] Step2. Allocate the priority selection probability according to the priority value of the subtasks. The subtask with a larger priority value has a higher probability of being selected first;

[0072] Step3. For each subtask , calculator value range ;

[0073] Step4. According to the execution effect of the elements in , assign the selection probability to it;

[0074] Step5. Select a subtask from the subtask set for resource allocation according to the probability;

[0075] Step6. Exclude the resources selected by the selected subtask from the available sets of all subtasks that conflict with this subtask. If there are still unprocessed subtasks, return to Step5, otherwise go to Step7;

[0076] Step7. End.

[0077] Preferably, the dynamic periodic cooperative scheduling in S2 includes establishing a preemption algorithm. When a missile target is detected and confirmed, the periodic scheduling sequence needs to be adjusted to respond to new tasks. The specific steps are as follows:

[0078] Step1. Task decomposition:

[0079] Decompose the tracking and detection task of the target into corresponding detection atomic tasks;

[0080] Step2. Calculate the parameters of each atomic task:

[0081] According to the predicted missile trajectory, calculate the visible time window between the satellite and the missile, the priority of each atomic task, and set the alternative resource set of each atomic task;

[0082] Step3. Set the task queue:

[0083] Put the atomic tasks whose execution time is within the current scheduling period into the task queue in sequence, and the atomic tasks with smaller task sequences are arranged in the front of the queue;

[0084] Step4. Judge whether the task queue is empty. If the queue is empty, the adjustment ends and exits; otherwise, go to Step5;

[0085] Step5. Take out the first atomic task from the task sequence (denoted as ), and judge whether the alternative resource set of is empty. If it is empty, return to Step4, otherwise go to Step6;

[0086] Step6. Judge ​Whether there is idle resource in the alternative resource set. If so, select the idle resource from factors such as successful detection probability, target tracking effect, resource utilization and handover, and pick out an optimal resource to allocate to , then return to Step4; otherwise, enter Step7;

[0087] Step7. Find out the resource that can be snatched. Select a resource from the alternative resource set of , denoted as , the atomic task executed by, denoted as , with the lowest priority among the tasks executed by all resources in the alternative resource set, which is the optimal resource that can be snatched for the atomic task ;

[0088] Step8. Judge whether the optimal resource that can be snatched can really be snatched:

[0089] If the priority of the atomic task executed by is higher than the task priority of, then the resource cannot be snatched , and the task will not be executed, and the whole adjustment process ends and exits. Since the task cannot snatch the resource , its subsequent atomic tasks will also be cancelled. Otherwise, allocate to the task , and release the resources allocated to the subsequent tasks of the atomic task . Since the task cannot be executed, its subsequent atomic tasks will also be cancelled, and return to Step4.

[0090] Preferably, after the preemptive algorithm is solved, the preempted target needs to be repaired. The repair process is actually a preemptive process. When the target to be repaired preempts the observation equipment of other targets, the observation equipment of this target is redundant. Otherwise, the target to be repaired fails to be repaired in this sub-cycle and needs to wait until the next sub-cycle for event-based repair or periodic scheduling.

[0091] The beneficial effects of the present invention are:

[0092] The present invention uses the multi-satellite cooperation technology to propose a new target detection system architecture, introduces high and low orbit satellites for target detection and scheduling design, improves the detection utilization rate of on-orbit satellites and the accuracy of target detection, the correctness and rationality of the cooperative scheduling algorithm, and the integrity of the full-process scheduling of the target in the case of known target tracks;

[0093] The present invention realizes using multi-satellite cooperation technology as the main carrier for target detection, and can schedule the number of on-orbit satellites and the number of tracking satellites in real time according to the change of tasks in a complex task environment. In addition, when the detection task scenario changes, it only needs short-time calculation to adapt to the new detection environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] In the drawings:

[0095] Figure 1 is the flowchart of a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0096] Figure 2 is the schematic diagram of rectangular detection of high-orbit satellites in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0097] Figure 1 is the flowchart of a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0098] Figure 2 is the schematic diagram of rectangular detection of high-orbit satellites in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0099] Figure 3 is the schematic diagram of geometric visibility in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0100] Figure 4 is the geometric configuration diagram of limb detection of low-orbit satellites in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0101] Figure 5 is the coverage diagram of low-orbit early warning satellites in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0102] Figure 6 is the flowchart of a discrete particle swarm optimization algorithm based on genetic operations in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention; Figure 7 is the flowchart of a preemption algorithm when a new target appears in a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0103] Figure 8 is the trajectory diagram of high-orbit satellites in Embodiment 1 of a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0104] Figure 9 is the trajectory diagram of low-orbit satellites in Embodiment 1 of a method for multi-objective cooperative scheduling of multiple space vehicles proposed by the present invention;

[0105] Figure 10 The sub - satellite point track diagram of the low - earth orbit satellite in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0106] Figure 11 The curve graph of the change of the semi - latus rectum p in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0107] Figure 12 The curve graph of the change of the eccentricity e in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0108] Figure 13 The curve graph of the change of the orbital inclination i in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0109] Figure 14 The T / S curve graph of ground exploration in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0110] Figure 15 The sub - cycle curve graph of ground exploration in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0111] Figure 16 The T / S curve graph of limb exploration in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0112] Figure 17 The sub - cycle curve graph of limb exploration in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0113] Figure 18 The detection schematic diagram of the geosynchronous satellite in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0114] Figure 19 The schematic diagram of the full - process target allocation in Embodiment 1 of a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention. Detailed implementation manner

[0115] Now, the present invention will be further described in detail with reference to the accompanying drawings. Refer to Figure 1 , a multi - vehicle multi - target cooperative scheduling method in space, characterized by comprising the following method steps: Figure 7 The flowchart of the preemption algorithm when a new target appears in a multi - vehicle multi - target cooperative scheduling method in space proposed by the present invention;

[0116] S1. Obtain target information; initialize the in - orbit high - and low - earth orbit satellites of our side;

[0117] S2. Establish a mathematical model for multiple satellites to meet the basic requirements for the normal operation of multiple satellites;

[0118] S3. Establish a multi-satellite scheduling and detection capability model, and select a reasonable detection method to detect the target;

[0119] S4. Adopt a particle swarm optimization algorithm based on genetic operations to optimize the calculation of the number of satellites used and the usage period during the detection process to achieve the optimal resource allocation. The principle of allocation is to achieve the longest detection period with the least number of satellites;

[0120] S5. Configure resource seizure optimization and repair, allocate available satellite resources, and reasonably apply them to the actual scenario;

[0121] S6. Determine whether the detection and tracking of the target reach the expected effect: if so, conduct continuous tracking and end; if not, return to S3.

[0122] The particle swarm optimization algorithm based on genetic operations includes a periodic static prediction-based guided scheduling according to the target information and the constraint conditions of low-Earth orbit satellites. That is, based on the predictability of the target and the predictability of the motion characteristics of the satellite platform where the sensor is located, the high-Earth orbit satellite predicts the states of the sensor and the target within the interval, establishes a static prediction-based scheduling model, and then through a specific optimization algorithm, pre-allocates the target and the sensor within the scheduling interval to generate a low-Earth orbit satellite scheduling plan to achieve the goal of optimal observation.

[0123] The configuration of resource seizure optimization and repair includes that when a missile target is detected and confirmed, the periodic scheduling sequence needs to be adjusted to respond to new tasks. The target that has been preempted needs to be repaired. The repair process is actually a preemption process. Its algorithm is basically the same as the preemption algorithm. The difference is that when the target to be repaired preempts the observation equipment of other targets, the observation equipment of this target is redundant; otherwise, the repair of the target to be repaired fails in this sub-period and needs to wait until the next sub-period for event-based repair or periodic scheduling.

[0124] The mathematical model includes a ground detection model algorithm and a limb detection model algorithm. The high-Earth orbit supports the ground detection model, and the low-Earth orbit supports two modes: ground detection and limb detection.

[0125] The mathematical model based on the six orbital elements includes establishing a satellite orbit dynamics model and analyzing the space environment perturbation factors, including the variation laws of satellites under the influence of the Earth's non-spherical perturbation, third-body perturbation, solar radiation pressure perturbation, atmospheric drag perturbation, and the influence of the Earth's shadow, and completing the orbital dynamics and kinematics modeling of three types of satellites: high-Earth orbit, low-Earth orbit, and geostationary orbit;

[0126] Among them, the satellite orbit dynamics equation described by classical orbital elements is as follows:

[0127]

[0128] The dynamics equation of the orbital elements of the vernal equinox point is as follows:

[0129]

[0130] Among them, for a satellite in a geostationary orbit, there is , and the above equation can be further simplified as:

[0131]

[0132] In the formula, are respectively the geocentric distance and the velocity magnitude of the geostationary orbit satellite, is the magnitude of the angular velocity of the Earth's rotation.

[0133] Modeling of the detection capability based on multi-satellite scheduling:

[0134] The modeling algorithms of the detection capability mainly include the ground detection model algorithm and the limb detection model algorithm. Among them, high orbits support the ground detection model, and low orbits support two modes: ground detection and limb detection.

[0135] The ground detection mode is based on the range of the surface area that the detector can scan and observe. Theoretically, targets within the coverage of the detector can be detected.

[0136] The coverage mainly describes the coverage characteristics of the field of view of the missile warning satellite detector to the ground and the constellation networking characteristics, and can be measured by indicators such as the global coverage rate and the multiple coverage rate of key regions.

[0137] Considering the equipment visibility and geometric visibility of high-orbit satellites to targets:

[0138] Equipment visibility: Referring to Figure 2 , the coverage range of the high-orbit satellite to the ground is regarded as a rectangle. In the determined coordinate system, the detector is simplified to a rectangular visual cone (referred to as the visual cone) starting from the origin of the coordinate system. The scalar and vector of the visual cone respectively characterize its size and direction. Among them, is the horizontal half-angle of view, also known as the azimuth angle, which characterizes the left and right swing angles of the detector in the direction perpendicular to the satellite orbit plane; is the vertical half-angle of view, also known as the pitch angle, which characterizes the up and down swing angles of the detector in the satellite orbit plane centered on the optical axis direction. When the target is within the rectangular detection range of the detector, it is considered that the high-orbit satellite is visible to the target equipment.

[0139] Geometric visibility: Referring to Figure 3, Assume the target is at point T and the two satellites are at points S1 and S2. It is easy to know that when , the target is geometrically visible to the observation satellite; when (i.e., corresponding to the case of satellite S2), if is satisfied, it is considered that the target is not blocked by the Earth from the observation satellite and is geometrically visible.

[0140] The low-Earth orbit satellite also supports the limb sounding mode. Refer to Figure 4 , the geometric configuration diagram of the limb sounding of the low-Earth orbit early warning satellite shown, where S is the early warning satellite with a height of , SE and SB are the boundaries of the limb sounding range of the sensor beam respectively, and the dashed line represents the maximum detection distance of the sensor. is the Earth's surface, is a certain height plane covered by the early warning satellite beam (tangent to the sensor beam at point D), is the height plane corresponding to the highest point covered by the early warning satellite beam (tangent to the sensor beam at point E). Points A and B are the two intersections of the tangent line of the sensor beam with the Earth's surface and layer, and F is the intersection of the boundary of the satellite sensor detection distance and layer.

[0141] Refer to Figure 5 , when measuring the coverage ability of the early warning satellite for a certain height layer , the actual coverage range is , and its projection on the Earth's surface is 's projection on the Earth's surface. Considering the actual situation of the satellite sensor scanning, the projection of the coverage range of the early warning satellite for a certain height layer on the ground is an annulus.

[0142] In the figure, is the sub-satellite point track of the satellite at a certain moment, are the geocentric angles corresponding to the inner circle and the outer circle respectively, and the calculation formulas are as follows:

[0143] In the formula, is layer's corresponding height. The calculation formulas for the sub-satellite point longitude and latitude of the satellite are as follows:

[0144]

[0145] In the formula, (x, y, z) is the coordinate of the satellite in the inertial system at a certain moment, is the sidereal hour angle of Greenwich at the initial moment, is the Earth's rotation speed, .

[0146] Assume that at a certain moment, the target 's projection on the Earth's surface can be represented by longitude and latitude Indicates a certain satellite The sub-satellite point of which Indicates that if the condition

[0147] is satisfied, then the target is considered to be within the detection coverage of the satellite .

[0148] The ground detection model is based on the surface area range that the detector can scan and observe. The coverage range describes the coverage characteristics of the detector's field of view of the missile warning satellite to the ground and the constellation networking characteristics, and is measured by indicators such as the global coverage rate and the multiple coverage rate of key regions.

[0149] The static periodic cooperative scheduling includes the setting of the objective function and the solution of the multi-variable multi-constraint optimization problem. The setting of the objective function includes the following four parts:

[0150] Objective function for the number of observed targets and their threat levels:

[0151] The objective function corresponds to design criteria one and two, and the function expression is as follows:

[0152]

[0153] In the formula, represents the number of targets successfully tracked within a certain allocation sub-period, represents the total number of targets existing in the sub-period system, represents whether the i-th target is successfully observed and tracked within the j-th allocation sub-period. If so, , otherwise , is the threat level value of the i-th target within the j-th allocation sub-period;

[0154] Objective function regarding the number of idle equipment in the system:

[0155] The objective function corresponds to design criterion three, and the function expression is as follows:

[0156]

[0157] In the formula, is the number of idle equipment, is the total number of equipment;

[0158] Objective function regarding the total arc length of equipment combination tracking:

[0159] The objective function corresponds to design criterion four, and the function expression is as follows:

[0160]

[0161] In the formula, is the current moment of the system, is the total duration during which the equipment combination can continuously observe target i starting from the current moment, is the target landing moment or the moment when the target moves out of the system;

[0162] Objective function regarding the number of sensor switches:

[0163] The objective function corresponds to Design Criterion Five, and the function expression is as follows:

[0164]

[0165] In the formula, represents the number of switches of all working sensors in two sub-periods before and after static periodic scheduling, represents the maximum value of the number of sensor switches in all allocation schemes;

[0166] From the above four formulas, it can be comprehensively known that the total objective function of the static periodic scheduling algorithm is:

[0167]

[0168] In the formula, represents the weight of the corresponding objective function in the total objective function, and there is .

[0169] The solution of the multi-variable and multi-constraint optimization problem is that a particle in the particle swarm corresponds to a solution of a problem in the solution space, and each particle has its own position, velocity, and cost evaluation index . When performing iterative update, according to the optimal solution of the particle itself and the optimal solution of all particles in the particle swarm to update the position and velocity of each particle;

[0170] Among them, the iterative update equation is:

[0171]

[0172] In the formula, , is the number of particles in the particle swarm; , is the dimension of the solution vector; k is the number of iterations; is the learning factor; is the weight; is a random number between [0, 1]. Preferably, for the guidance problem between high and low orbit satellites, a discrete grouped particle swarm algorithm needs to be adopted, which is described as follows:

[0173] Particle coding mechanism:

[0174] Now, the coding form of the particle is determined as:

[0175]

[0176] In the formula, represents the number of sensors equipped in the current system; the element represents the target number observed by the sensor. For example, if the sensor numbered 1 observes target 1, then , if the sensor numbered 1 is in the idle state, then ;

[0177] Particle swarm initialization mechanism;

[0178] Particle position update mechanism:

[0179] Introduce the relevant operations of the genetic algorithm into the particle swarm algorithm by using the genetic algorithm, use the genetic operation as the particle update operator, and adopt the particle swarm optimization algorithm based on genetic operation

[0180] Dynamic mutation operator of the particle itself :

[0181] Inertia weight of the particle swarm optimization algorithm , the mutation probability decreases dynamically to achieve the balance between the random search and local search of the algorithm, representing the particle 's thinking about its own flight speed, and its formal description is as shown in the formula:

[0182]

[0183] In the formula, is a random number uniformly distributed on; is the probability of performing the mutation operation; represents the same mutation operation as in the genetic algorithm;

[0184] represents that with probability perform the same mutation operation as represented by, otherwise ;

[0185] Particle and individual extreme value crossover operator :

[0186] The particle crosses with the individual extreme value with a given probability , and selects the particle with better fitness as the updated particle after the crossover, representing the particle adjusts its position according to the individual extreme value , and its formal description is as shown in the formula:

[0187]

[0188] In the formula, is a uniformly distributed random number on; the acceleration constant; represents the same crossover operation as in the genetic algorithm;

[0189] represents that with probability perform the crossover operation the same as the genetic algorithm represented by, otherwise ;

[0190] Particle and global extreme value crossover operator :

[0191] The particle performs crossover with the global extreme value with a given probability and its formal description is as shown in the formula:

[0192]

[0193] In the formula, is a uniformly distributed random number on; the acceleration constant; represents the same crossover operation as in the genetic algorithm;

[0194] represents that with probability perform the crossover operation the same as the genetic algorithm represented by, otherwise ;

[0195] Combining the above formula, the particle position formula can be obtained as:

[0196] .

[0197] The particle swarm initialization mechanism is a priority-based particle swarm initialization mechanism, and its specific process is as follows:

[0198] Step1. Calculate the priority value for all subtasks within the scheduling window ;

[0199] Step2. Allocate a priority selection probability according to the priority value of the subtasks. The subtask with a larger priority value has a higher probability of being selected first;

[0200] Step3. For each subtask , calculate the value range ;

[0201] Step4. According to the elements in The execution effect is to assign it a selection probability;

[0202] Step5. Select a subtask from the subtask set for resource allocation according to the probability;

[0203] Step6. Exclude the resources selected by the selected subtask from the available sets of all subtasks that conflict with this subtask. If there are still unprocessed subtasks, return to Step5; otherwise, go to Step7;

[0204] Step7. End.

[0205] The dynamic periodic cooperative scheduling in S2 includes establishing a preemption algorithm. When a missile target is detected and confirmed, the periodic scheduling sequence needs to be adjusted to respond to new tasks. The specific steps are as follows:

[0206] Step1. Task decomposition:

[0207] Decompose the tracking and detection task of the target into corresponding detection atomic tasks;

[0208] Step2. Calculate the parameters of each atomic task:

[0209] According to the predicted missile ballistic trajectory, calculate the visible time window between the satellite and the missile, the priority of each atomic task, and set the alternative resource set for each atomic task;

[0210] Step3. Set the task queue:

[0211] Put the atomic tasks whose execution time is within the current scheduling period into the task queue in sequence, and the atomic tasks with a smaller task sequence are arranged in the front of the queue;

[0212] Step4. Judge whether the task queue is empty. If the queue is empty, the adjustment ends and exits; otherwise, go to Step5;

[0213] Step5. Take out the first atomic task from the task sequence (denoted as ), and judge Whether the alternative resource set of is empty. If it is empty, return to Step4; otherwise, go to Step6;

[0214] Step6. Judge Whether there is idle resource in the alternative resource set of. If there is, select from the factors such as successful detection probability, target tracking effect, resource utilization and switching for the idle resources, and select an optimal resource from them to allocate to , and then return to Step4; otherwise, go to Step7;

[0215] Step7. Identify the resources that can be snatched. Select a resource from the alternative resource set in and denote it as . The atomic task executed by has the lowest priority among the tasks executed by all resources in the alternative resource set. That is, the atomic task is the best resource that can be snatched;

[0216] Step8. Determine whether the best resource that can be snatched can actually be snatched:

[0217] If the priority of the atomic task executed by is higher than the task priority of then the resource cannot be snatched, and the task will not be executed, and the entire adjustment process ends and exits. Since the task cannot snatch the resource its subsequent atomic tasks will also be cancelled. Otherwise, will be allocated to the task , and the resources allocated to the subsequent tasks of the atomic task will be released. Since the task

[0218] cannot be executed, its subsequent atomic tasks will also be cancelled, and it will return to Step4. Embodiment

[0219] S1. Analyze the entire process of satellite cooperative scheduling in three scenario assumptions in the directions of North America, the Western Pacific, and the Indian Ocean according to technical requirements, and verify the rationality of the model algorithm.

[0220] Simulation scenario settings:

[0221] (1) The satellite deployment information is shown in Table 1-1-1-3, including 24 low Earth orbit (LEO) satellites, 4 geostationary orbit (GEO) satellites, and 4 highly elliptical orbit (HEO) satellites;

[0222] Table 1-1 LEO satellite deployment information

[0223]

[0224] Table 1-2 HEO Satellite Deployment Information

[0225]

[0226] Table 1-3 GEO Satellite Deployment Information

[0227]

[0228] (2) The target information is shown in Table 2. Three ballistic missiles are launched from bases at A, B, and C respectively, intending to attack three important cities;

[0229] Table 2 Target Scenario Table

[0230]

[0231] (3) Geosynchronous satellite's azimuth angle for earth observation: 10°; elevation angle: 8°;

[0232] (4) Low Earth Orbit (LEO) satellite uses limb sounding, and the maximum distance of the detector is 7000 km.

[0233] S2, Reference Figures 8 - 9 , By setting the initial six orbital elements of the large elliptical orbit satellite, the initial longitude of the geostationary orbit, and the working time of the satellite, the position change of the geosynchronous satellite and the change of the sub-satellite point trajectory of the geosynchronous satellite can be obtained.

[0234] Reference Figures 10 - 11 , By setting the initial six orbital elements of the LEO satellite and the working time of the satellite, the position change of the LEO satellite and the change of the sub-satellite point trajectory of the LEO satellite can be obtained.

[0235] Reference Figures 12 - 14 , During the operation of the satellite, it will be affected by perturbing forces, and the six orbital elements will also change accordingly; the changes in the semi-latus rectum p, eccentricity e, and orbital inclination i output by the LEO satellite numbered LEO1 during this operation.

[0236] S3, Reference Figures 15 - 18 , Analyze and verify by using direct earth observation and limb sounding methods respectively:

[0237] Earth observation mode, LEO earth observation angle: azimuth angle: 45°; elevation angle: 45°;

[0238] Limb sounding mode, maximum detection distance of LEO satellite: 8000 km.

[0239] As can be seen from the above, when using the ground detection mode, the number of observable satellites for the target is small, and it is impossible to effectively complete the full-process observation of the target and the task scheduling algorithm allocation. Because the LEO satellite orbit is relatively low, at 1600 km; the highest point of the long-range ballistic missile is relatively high, and it is difficult for the LEO satellite using the ground detection mode to effectively maintain long-time observation and tracking of the target.

[0240] S4. At the same time, a solution is obtained. Adopt the limb detection mode, and at the same time increase the detection angle of the LEO satellite to the ground, and adopt the limb + ground detection mode.

[0241] S5. Taking Target2 as an example, hereinafter referred to as T2 for short, the full process analysis of its cooperative scheduling is described.

[0242] When the system runs to 10 s, the target T2 is successively detected by the scanning detectors of the high-orbit GEO2, GEO4, and HEO4. The cooperative scheduling starts. The high-orbit satellites immediately call the staring detectors for staring tracking and multi-satellite positioning, and at the same time trigger the event scheduling algorithm. The ground station calculates the observable equipment according to the target position and the equipment position and finds that only LEO1 can observe T2 at this moment. The event scheduling algorithm assigns LEO1 to track and observe T2. When the system runs to 47 s, the equipment LEO23 becomes visible to T2, so it is directly called by the event scheduling algorithm. At this time, LEO1 and LEO23 are already observing and tracking T2, and double-star positioning can be carried out and the trajectory can be predicted. When the system runs to 60 s, LEO16 also joins the tracking and observation of T2 to maintain the observation accuracy constraint for T2.

[0243] When the system runs to 180 s, at the second cycle scheduling point, the static cycle scheduling algorithm is triggered to perform multi-target allocation for the existing targets in the multi-system. Two observation equipments, LEO1 and LEO16, are assigned to T2 for double-star positioning and continuous observation and tracking. After the end of the 11th sub-cycle, LEO1 becomes invisible to T2, and there is no other equipment in the system that can observe T2 at this time. Therefore, only LEO16 performs single-star observation on it until the start of the 24th sub-cycle. From the 24th sub-cycle to the 30th sub-cycle, three equipments, LEO16, LEO21, and LEO22, perform continuous observation on T2; from the 30th sub-cycle to the 41st cycle, two equipments, LEO21 and LEO22, perform continuous observation on T2; from the 41st sub-cycle to the 51st sub-cycle, two equipments, LEO15 and LEO21, perform continuous observation on T2; from the 51st sub-cycle to the 53rd sub-cycle, only one equipment, LEO15, performs continuous observation on T2.

[0244] After the end of the 53rd sub-cycle, no equipment is visible to T2, and the full process of multi-satellite cooperative scheduling for T2 ends. Next, the target is handed over to the ground radar station for subsequent observation and processing.

[0245] Through the analysis of typical scenarios, the correctness, rationality of the collaborative scheduling algorithm and the integrity of the full-process scheduling of the target in the case of known target tracks are verified.

[0246] The present invention uses multi-satellite collaboration technology to propose a new target detection system architecture, introduces high and low orbit satellites for target detection and scheduling design, improves the detection utilization rate of on-orbit satellites and the accuracy of target detection, the correctness and rationality of the collaborative scheduling algorithm, and the integrity of the full-process scheduling of the target in the case of known target tracks;

[0247] The present invention realizes using multi-satellite collaboration technology as the main carrier of target detection, and can real-time schedule the number of on-orbit satellites and the number of tracking satellites according to the change of tasks in a complex task environment. In addition, when the detection task scenario changes, it only needs short-time calculation to adapt to the new detection environment.

[0248] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A multi-objective cooperative scheduling method for multiple space vehicles, characterized in that, The method comprises the following steps: S1. Initialize our high-orbit and low-orbit satellites, and use the orbit six-element principle to establish mathematical models of high-orbit and low-orbit satellites; S2. Using the particle swarm optimization algorithm based on genetic operations, the optimal allocation of multiple targets under multi-equipment joint observation is carried out in the mathematical model according to target threats, resource constraints, and working period constraints. The tasks are dynamically replanned when multiple targets change. According to the given observation area and observation time requirements, the visible range and re-entry frequency of each satellite are analyzed to calculate the optimal scheduling plan for each satellite. S3, configure resource grabbing optimization and repair, configure available satellite resources, and apply them reasonably in actual scenarios; S4, determine whether the current state meets the detection expectation, if so, continue to track the target, if not, return to S2, and finally achieve tracking of the target; The static period cooperative scheduling in S2 includes setting the objective function and solving the multi-variable multi-constraint optimization problem, wherein the objective function setting includes the following four parts: The objective function of the number of observed targets and their threat level is as follows: Wherein, represents the number of targets successfully tracked within a certain allocation sub - period, represents the total number of targets existing in the system during this sub - period, represents whether the i - th target is successfully observed and tracked within the j - th allocation sub - period. If so, , otherwise , is the threat degree value of the i - th target within the j - th allocation sub - period; Regarding the objective function of the number of idle devices in the system, the function expression is as follows: In the formula, is the number of idle devices, is the total number of devices; Regarding the objective function of the total arc length of equipment combination tracking, the function expression is as follows: Wherein, is the current moment of the system, is the total duration from the current moment when the equipment combination can continuously observe target i, is the moment when the target lands or the moment when the target moves out of the system; Regarding the objective function of the number of sensor switching times, the function expression is as follows: Wherein, represents the number of switching times of all working sensors in two sub - cycles before and after static periodic scheduling; represents the maximum value of the number of sensor switching times among all allocation schemes; From the above four equations, we can see that the overall objective function of the static periodic scheduling algorithm is: In the formula, represents the weight of the corresponding objective function in the total objective function, and ; The solution to the multi-variable and multi-constraint optimization problem is that a particle in the particle swarm corresponds to a solution to a problem in the solution space, and each particle has its own position, velocity, and cost evaluation index , when performing iterative updates, according to the optimal solution of the particle itself and the optimal solutions of all particles in the particle swarm to update the position and velocity of each particle; Among them, the iterative update equation is: In the formula, , is the number of particles in the particle swarm; , is the dimension of the solution vector; k is the number of iterations; , are learning factors; is the weight; , are random numbers between [0, 1]; For the guidance problem between high and low orbit satellites, a discrete cluster particle swarm algorithm is needed, which is described as follows: Particle encoding mechanism: The particle encoding form is now determined as: In the formula, represents the number of sensors equipped in the current system; the element represents the target number observed by the sensor. If the sensor numbered 1 observes target 1, then . If the sensor numbered 1 is in an idle state, then ; Particle swarm initialization mechanism; Particle position update mechanism: The genetic algorithm is used to introduce the related operations of the genetic algorithm into the particle swarm optimization algorithm, and the genetic operation is used as the particle update operator, and the particle swarm optimization algorithm based on the genetic operation is adopted; Dynamic mutation operator of the particle itself : Inertia Weight of Particle Swarm Optimization Algorithm , the mutation probability decreases dynamically to achieve the balance between the random search and local search of the algorithm. The representative particle considers its own flight speed, and its formal description is shown in the formula: In the formula, is a uniformly distributed random number; is the probability of performing the mutation operation; represents the same mutation operation as in the genetic algorithm; Indicates that with probability perform a mutation operation identical to the genetic algorithm, otherwise ; Particle and individual extreme value crossover operator :[[-END]] Particles cross with the individual extreme value with a given probability After crossing, the particle with better fitness is selected as the updated particle, indicating that the particle adjusts its position according to the individual extreme value The formal description is as shown in the formula: In the formula, is a uniformly distributed random number; is the acceleration constant; represents the crossover operation as in the genetic algorithm; Indicates that with probability perform the crossover operation same as the genetic algorithm, otherwise ; Particle and Global Extreme Crossover Operator : The particles cross with the global extreme value with a given probability and its formal description is shown in the formula as follows: wherein, is a uniformly distributed random number; is the acceleration constant; represents the same crossover operation as in the genetic algorithm; Indicates with probability Perform the same crossover operation as the genetic algorithm for the representation, otherwise ; Combining the above formula, the particle position formula can be obtained as follows: 。 2. The multi-objective collaborative scheduling method for multiple space vehicles according to claim 1, characterized in that The mathematical model in S1 includes a ground detection model algorithm and a limb detection model algorithm. The high orbit supports a ground detection model, and the low orbit supports both ground detection and limb detection modes.

3. The multi-objective cooperative scheduling method for multiple space vehicles according to claim 2, characterized in that, The ground detection model in S1 is based on the range of the surface area that the detector can scan and observe. The coverage range describes the coverage characteristics of the missile early warning satellite detector's field of view to the ground and the constellation networking characteristics, and is measured using indicators such as global coverage and multiple coverage of key areas.

4. A multi-objective cooperative scheduling method for multiple space vehicles according to claim 1, characterized in that The particle swarm initialization mechanism is a priority-based particle swarm initialization mechanism, and its specific process is as follows: Step1. Calculate the priority values for all subtasks within the scheduling window ; Step 2: Assign the probability of priority selection to each subtask according to its priority value. The subtask with a larger priority value has a higher probability of being selected. Step 3. For each subtask , calculate the value range of the calculator ; Step 4. According to the execution effect of the element pair in it, assign it a selection probability; Step 5. Select a subtask from the subtask set according to the probability for resource allocation; Step 6: Remove the resources selected by the selected subtask from the available set of all subtasks that conflict with the subtask. If there are still unprocessed subtasks, return to Step 5, otherwise go to Step 7. Step 7. End.

5. A multi-objective collaborative scheduling method for multiple space vehicles according to claim 1, characterized in that The dynamic cycle collaborative scheduling in S2 includes establishing a preemption algorithm. When a missile target is detected and confirmed, the cycle scheduling sequence needs to be adjusted to respond to new tasks. The specific steps are as follows: Step1. Task decomposition: Decompose the tracking and detection task of the target into corresponding detection atomic tasks; Step2. Calculate the parameters of each atomic task: According to the predicted missile ballistic trajectory, calculate the visible time window between the satellite and the missile, the priority of each atomic task, and set the alternative resource set for each atomic task; Step3. Set up the task queue: Put the atomic tasks whose execution time is within the current scheduling cycle into the task queue in sequence, and the atomic tasks with smaller task sequences are arranged in the front of the queue; Step4. Determine whether the task queue is empty. If the queue is empty, the adjustment is completed and exit; otherwise, enter Step5; Step5. Take out the first atomic task from the task sequence (denoted as ), and determine whether the alternative resource set of is empty. If it is empty, go back to Step4; otherwise, go to Step6. Step6. Determine whether there is idle resource in the alternative resource set of. If so, select the idle resource from factors such as successful detection probability, target tracking effect, resource utilization and handover, and pick out an optimal resource to allocate to , then go back to Step4; otherwise, enter Step7; Step 7. Find out the resources that can be snatched, and select a resource from the alternative resource set in and denote it as . Execute the atomic task, denoted as . Its priority is the lowest among the tasks executed by all resources in the alternative resource set. That is the best resource that can be snatched for the atomic task . Step8. Determine whether the best resource that can be snatched can really be snatched: If the atomic task being executed has a higher priority than the task priority, then it cannot preempt resources , and the task will not be able to be executed, the entire adjustment process ends, exits, because the task cannot preempt resources , and its subsequent atomic tasks will also be cancelled, otherwise it will be allocated to the task , and the resources allocated to the subsequent tasks of the atomic task will be released. Because the task cannot be executed, its subsequent atomic tasks will also be cancelled, and return to Step4.

6. A multi-objective cooperative scheduling method for multiple space vehicles according to claim 5, characterized in that After the preemption algorithm is solved, the preempted target needs to be repaired. The repair process is actually a preemption process. When the target to be repaired preempts the observation equipment of other targets, the observation equipment of this target is redundant. Otherwise, the target to be repaired fails to be repaired in this sub-cycle and needs to wait until the next sub-cycle for event-based repair or periodic scheduling.

Images