A Cross-Domain Sensor Resource Scheduling Method and System Based on Adaptive Particle Swarm Optimization

Abstract

This invention belongs to the field of sensor scheduling and discloses a cross-domain sensor resource scheduling method and system based on adaptive particle swarm optimization. The method includes initializing the particle swarm; updating the initialized particle swarm using a combination of swarm optimization and particle update algorithms; in the leader bee neighborhood search phase, updating the initialized particle swarm using the particle update algorithm to generate updated particles and calculating the target fitness corresponding to each updated particle; in the follower bee follow search phase, updating the particle swarm again using the roulette wheel algorithm based on the target fitness corresponding to each updated particle to obtain the update result; repeating the processes corresponding to the leader bee neighborhood search phase and the follower bee follow search phase until the maximum number of iterations is reached, and outputting the current update result as the optimal sensor scheduling scheme. This method improves detection accuracy and reduces radiation exposure risk and the number of sensor task switching operations.

Description

Technical Field

[0001] This invention belongs to the field of sensor scheduling technology, specifically relating to a cross-domain sensor resource scheduling method and system based on adaptive particle swarm. Background Technology

[0002] In modern information warfare, securing the "information high ground" and effectively leveraging information superiority are crucial for victory. Sensors act as the "eyes and ears" in the adversarial environment, playing a vital role in acquiring information. However, in adversarial environments, the presence of numerous targets, environmental influences, and potential interference can cause sensors to malfunction or be damaged, rendering them unable to continue their missions. Single-domain sensors, with their limited energy and performance, cannot meet the surveillance and detection demands of such complex adversarial environments. Therefore, multi-domain sensors working in tandem are necessary to fulfill these requirements. However, applying multi-sensor systems for detection in complex adversarial environments necessitates careful resource allocation within these systems, which still have limited resources.

[0003] Existing research on cross-domain sensor resource scheduling mainly includes linear programming methods, information theory-based methods, artificial intelligence-based methods, and heuristic algorithm-based methods. However, most existing research focuses on sensor scheduling problems in a single domain, considering only target detection, target tracking, or both separately, with almost no consideration for multi-domain collaborative problems. While information theory methods can achieve good technical performance, they cannot be well integrated with actual combat requirements and often suffer from high computational complexity and low timeliness. Heuristic algorithms themselves may have problems such as slow convergence speed and low accuracy, and there is a lack of research using heuristic algorithms to solve multi-domain collaborative sensor resource scheduling problems.

[0004] It is evident that most existing sensor scheduling research focuses on sensor scheduling within a single domain. When it comes to cross-domain sensor resource scheduling, existing scheduling methods suffer from insufficient global optimization capabilities, high switching costs, and high radiation exposure risks. Summary of the Invention

[0005] The purpose of this invention is to provide a cross-domain sensor resource scheduling method and system based on adaptive particle swarm optimization, in order to solve the problems of insufficient global optimization capability, high switching cost, and high radiation exposure risk of existing scheduling methods.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a cross-domain sensor resource scheduling method based on adaptive particle swarm optimization, comprising:

[0008] The particle swarm is initialized by using an effective multi-domain sensor scheduling sequence as the initial particles of the swarm.

[0009] The initial particle swarm is updated by combining the bee colony algorithm and the particle update algorithm. The specific steps are as follows:

[0010] In the leading bee neighborhood search phase, the particle update algorithm is used to update the initialized particle swarm, generate updated particles, and calculate the target fitness corresponding to each updated particle.

[0011] During the follower bee search phase, based on the target fitness corresponding to each updated particle, the roulette wheel algorithm is used to update the updated particle swarm again, obtain the update result, and recalculate the target fitness corresponding to the updated particle.

[0012] Repeat the process corresponding to the leading bee neighborhood search phase and the following bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal scheduling scheme for the sensor.

[0013] Furthermore, the initialization process for the particle swarm specifically includes:

[0014] Initialize the population size, number of targets, maximum number of iterations, and number of sensors corresponding to the particle swarm;

[0015] For each target visual sensor set, a generative algorithm is used to generate an effective multi-domain sensor scheduling sequence as the initial particles of the particle swarm, where each dimension is different.

[0016] For the initial particles of the particle swarm, a greedy strategy is used to calculate the initial particle fitness, initial individual optimal solution, initial global optimal solution, and the selection probability of each dimension of the particle for the current solution, individual optimal solution, global optimal solution, and other optional values.

[0017] The multi-domain sensors include space-based sensors, air-based sensors, sea-based sensors, and land-based sensors.

[0018] Furthermore, the specific steps for updating the initialized particle swarm using the particle update algorithm during the guided bee neighborhood search phase are as follows:

[0019] Based on the selection probabilities corresponding to the current solution, the individual optimal solution, the global optimal solution, and the remaining alternative solutions, the roulette wheel algorithm is used to update the initialized particle swarm and generate updated particles.

[0020] Calculate the target fitness of the updated particle, as well as the selection probabilities of the current solution, individual optimal solution, global optimal solution, and other alternative solutions for the updated particle.

[0021] Furthermore, the specific steps for updating the initialized particle swarm using the roulette wheel algorithm include:

[0022] Targets are grouped based on the distance between them, and each group is sorted from highest to lowest average threat level. The order of two targets within a group is processed by enumeration, resulting in all possible target assignment sequences.

[0023] For each target assignment sequence, sensors are selected in order of probability, where:

[0024] When any sensor is selected by the target in the current dimension, this sensor will be removed from the available sensor sequence for targets in subsequent dimensions.

[0025] When two targets select the same sensor number, the probability sorting principle is adopted to select the particle corresponding to the particle with a higher probability and which does not overlap with the previously generated component for updating.

[0026] When updating to the kth dimension, first compare whether the sensor corresponding to the kth dimension of the current solution, the individual optimal solution, and the global optimal solution has been selected by the first k-1 dimensions of the new particle sequence.

[0027] If the k-th dimension of the current solution already exists in the new particle sequence, then the corresponding P1 flag is 0; if the k-th dimension of the individual optimal solution already exists in the new particle sequence, then the corresponding P2 flag is 0; if the k-th dimension of the global optimal solution already exists in the new particle sequence, then the corresponding P3 flag is 0; where P1, P2, P3, and P4 represent the selection probabilities corresponding to the current solution, the individual optimal solution, the global optimal solution, and other optional values, respectively.

[0028] Compare the maximum values ​​of P1~P3 with the flag bit set to 1, and record the maximum value;

[0029] The roulette wheel algorithm and target update sequence are used to update the particles sequentially. If the value of the newly generated dimension does not repeat the value of the previously generated component, the update proceeds to the next dimension; otherwise, the component corresponding to the maximum value of the flag bit 1 in P1~P3 is selected for update.

[0030] Furthermore, the specific calculation formulas for P1, P2, P3, and P4 are as follows:

[0031]

[0032] In the formula, Cycle represents the current iteration number, MaxCycle represents the maximum set iteration number; p takes a value from 0 to 1; This represents the fitness value of the current solution. This represents the fitness value of an individual's optimal solution. This represents the fitness value of the global optimal solution.

[0033] Furthermore, in the follower bee search phase, the specific steps for updating the particle swarm again using the roulette wheel algorithm based on the target fitness corresponding to each updated particle are as follows:

[0034] Normalize the fitness of each updated particle;

[0035] Based on the normalized fitness, the roulette wheel algorithm is then used to update the particle swarm again, and the fitness of the updated particles is recalculated, as well as the selection probabilities of the current solution, individual optimal solution, global optimal solution and other optional value schemes of the updated particles.

[0036] Based on the updated current solution, individual optimal solution, global optimal solution, and selection probabilities corresponding to other optional values ​​of the particle, the individual optimal solution and global optimal solution are updated using a greedy strategy to obtain the updated result.

[0037] Furthermore, it also includes a low-quality particle elimination stage. After the follower bee search stage, it is determined whether the number of particle updates has reached the set elimination limit. If it has, a new particle is regenerated using a generation algorithm to replace the old particle; otherwise, the next step continues.

[0038] Secondly, the present invention provides a cross-domain sensor resource scheduling system based on adaptive particle swarm optimization, comprising:

[0039] The particle swarm initialization module is used to initialize the particle swarm, where an effective multi-domain sensor scheduling sequence is used as the initial particles of the particle swarm.

[0040] The particle swarm update module combines the bee colony algorithm and the particle update algorithm to update the initialized particle swarm. Specifically, it includes:

[0041] The leading bee neighborhood search module is used to update the initialized particle swarm using a particle update algorithm during the leading bee neighborhood search phase, generate updated particles, and calculate the target fitness corresponding to each updated particle.

[0042] The follower bee follower search module is used to update the updated particle swarm again based on the target fitness of each updated particle during the follower bee follower search phase, using the roulette wheel algorithm to obtain the update result and recalculate the target fitness of the updated particles.

[0043] The scheduling scheme output module is used to repeatedly execute the processes corresponding to the leader bee neighborhood search phase and the follower bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal scheduling scheme for the sensor.

[0044] Thirdly, the present invention provides an apparatus comprising:

[0045] Memory, used to store computer programs;

[0046] A processor is used to implement the steps of the above-described cross-domain sensor resource scheduling method based on adaptive particle swarm when executing the computer program.

[0047] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement the steps of the above-described cross-domain sensor resource scheduling method based on adaptive particle swarm optimization.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] This invention provides a cross-domain sensor resource scheduling method based on adaptive particle swarm optimization (PSO). This method combines the global search capability of PSO with the local optimization capability of bee swarm optimization. It first rapidly converges to a near-optimal solution using PSO, and then uses bee swarm optimization to refine the solution locally, thus overcoming the problem of getting trapped in local optima when using bee swarm optimization alone. This method effectively improves the quality of solutions in large-scale sensor resource scheduling problems. Furthermore, through the information exchange and cooperation mechanism of bee swarm optimization, this method enhances the detail of the local search. Especially in sensor resource scheduling, task allocation and resource configuration often exhibit time-varying and spatial distribution characteristics; information sharing and collaborative search help find optimal or near-optimal resource scheduling schemes. Under multi-domain sensor and multi-target constraints, this method comprehensively considers detection accuracy, radiation exposure risk, and the number of sensor task switching times. The proposed method outperforms both bee swarm optimization and PSO in terms of switching cost and global fitness, effectively solving suboptimal or near-optimal solutions, meeting the detection requirements of multi-sensor systems in complex adversarial environments.

[0050] Furthermore, this scheduling method fully leverages the strengths of both algorithms in multi-objective optimization. The particle swarm optimization algorithm quickly finds solutions that satisfy basic constraints through updates to swarm information, while the bee swarm optimization algorithm refines the objectives based on this, finding a more balanced compromise solution. The combined algorithm provides a better compromise solution in multi-objective sensor resource scheduling problems, satisfying the complex needs of various tasks and resources.

[0051] Preferably, the present invention also includes a stage for eliminating low-quality particles. When the number of times a particle is updated reaches the set elimination limit, a new particle is generated using a generation algorithm to replace the old particle. This serves as a broad detection function for particles to escape local optima and avoids premature convergence of the algorithm to a certain extent. Attached Figure Description

[0052] Figure 1 A flowchart of a cross-domain sensor resource scheduling method based on adaptive particle bee colony provided in an embodiment of the present invention;

[0053] Figure 2 The scheduling method provided in this embodiment of the invention corresponds to the sensor scheduling sequence diagram at times 750-850.

[0054] Figure 3 This is a sensor scheduling sequence diagram for time 750-850 of the standard bee colony algorithm provided in an embodiment of the present invention;

[0055] Figure 4 This is a sensor scheduling sequence diagram for time 750-850 of the standard particle swarm optimization algorithm provided in an embodiment of the present invention;

[0056] Figure 5 Simulation diagram of the fitness of three algorithms at time 100 as a function of the number of iterations, provided for an embodiment of the present invention;

[0057] Figure 6 Simulation diagram of the fitness of three algorithms at time 700 as a function of iteration number, provided for embodiments of the present invention;

[0058] Figure 7 Simulation diagram of the fitness of three algorithms at time 850 as a function of iteration number, provided for an embodiment of the present invention;

[0059] Figure 8 Simulation diagram of the fitness of three algorithms at time 1000 as a function of iteration number, provided for an embodiment of the present invention;

[0060] Figure 9 A flowchart of a cross-domain sensor resource scheduling method based on adaptive particle bee colony provided by the present invention;

[0061] Figure 10This is a schematic diagram of a cross-domain sensor resource scheduling system based on adaptive particle bee colony, provided by the present invention. Detailed Implementation

[0062] Example 1

[0063] As described in the background section, most existing research focuses on sensor scheduling problems in a single domain, or considers only target detection, target tracking, or both separately, with almost no consideration given to multi-domain collaborative problems. While information-theoretic methods can achieve good technical performance, they are not well integrated with actual combat requirements and often suffer from high computational complexity and low timeliness. Heuristic algorithms themselves may have problems such as slow convergence speed and low accuracy, and there is a lack of research in the current field using heuristic algorithms to solve multi-domain collaborative sensor resource scheduling problems.

[0064] To address the aforementioned issues, this invention provides a cross-domain sensor resource scheduling method based on adaptive particle swarm optimization. This scheduling method effectively solves the problems of single-domain sensors, slow convergence speed, and susceptibility to local optima in existing technologies.

[0065] The following explanation and description of this embodiment are provided in conjunction with the accompanying drawings:

[0066] like Figure 1 As shown, this embodiment provides a cross-domain sensor resource scheduling method based on adaptive particle swarm optimization, specifically including:

[0067] S1. Initialize the particle swarm. During the parameter initialization phase, it is necessary to determine parameters such as population size, target number, maximum number of iterations, upper limit of leader bee elimination, and number of sensors.

[0068] S101. In this embodiment, the population size of the particle swarm is set to 60, the number of targets is 10, the maximum number of iterations is 100, the maximum number of eliminations is 40, and the number of sensors is the sum of the number of sensors in the four domains of air, land, sea, and air. Among them, the number of air-based sensors is set to 6, the number of space-based sensors is 18, the number of sea-based sensors and land-based sensors is 6 each, and the total number of cross-domain sensors is 36.

[0069] S2. In the particle initialization phase, based on each target visual sensor set, a generation algorithm is used to generate a sensor sequence that is different in each dimension, i.e., the initial particles. The fitness of the initial particles is calculated according to the objective function, and the individual optimal solution, the global optimal solution, and P1~P4 are initialized.

[0070] S201. In this embodiment, each dimensional component of the particle is set as the sensor number visible to the target. Since the number of targets is 10, the dimension of each particle is 10. This particle coding method directly represents the sensor scheduling scheme, avoiding the calculation of invalid particles and facilitating the faster finding of better particles.

[0071] S202. The following introduces the implementation process of the particle initialization algorithm. First, each dimensional component of the particle is randomly selected within the optional value range. Since the visible sensor numbers of two different targets may be the same, when a certain sensor is selected, the subsequent dimensional optional sensor sequence will exclude this sensor, that is, this sensor will no longer be selected, which ensures that the selected sensor numbers in each dimension are not repeated. If the elements generated in the particle are all non-repeated, it can be considered that the particle is legal.

[0072] S203. Specifically, the individual optimal solution is the optimal solution encountered by each particle during the iteration process, which is randomly generated at the beginning and continuously updated during the iteration process. For the global optimal solution, it refers to the optimal solution searched by all particles during the update process. At the beginning, the current fitness of each initial particle is calculated, and the greedy strategy is used to obtain the current global optimal solution, and this value will also be continuously updated during the iteration process. P1 to P4 are calculated using the following formula, where Cycle represents the current iteration number, MaxCycle is the set maximum iteration number, P4 is a value that continuously decreases as the iteration number increases, and 0 < p < 1 is set; this is to ensure that the particles still have a certain ability to jump out of the local optimal solution in the later stage of the iteration. P1 to P3 are related to the fitness of the particle, where represents the fitness value of the current solution, represents the fitness value of the individual optimal solution, represents the fitness value of the global optimal solution. Since in this embodiment, the smaller the fitness value, the better the corresponding sensor scheduling scheme, the reciprocal of the fitness value is taken during the calculation, that is, the smaller the fitness value of the sensor scheduling scheme, the larger the corresponding P value.

[0073]

[0074] S3. In the search stage of the leading bee domain, it is necessary to use the particle update algorithm to generate updated particles according to the visible sensor set of each target.

[0075] S301. The implementation process of the particle update algorithm will be described in detail below. During the update process, there are four possible values ​​for the i-th dimension of the particle: the value of the i-th dimension of the current solution, the value of the i-th dimension of the individual optimal solution, the value of the i-th dimension of the global optimal solution, and the remaining possible values ​​for the i-th dimension. These correspond to P1 to P4 respectively. For each dimension, the roulette wheel algorithm is used. The larger the P value, the better the fitness of the solution and the greater the probability of it being selected. This reflects the selection probability of the particle after "thinking" about the relevant solutions.

[0076] S302. In this embodiment, considering the impact of threat level, the targets are grouped and sorted. First, the targets are grouped according to the distance between them. Targets that are relatively close to each other are grouped together, indicating that the targets in the same group are more likely to compete for sensor resources. Different groups are sorted according to the average threat level within the group. Targets with higher threat levels are given priority in sensor resource allocation. The order of two targets within a group is processed by enumeration. Finally, all possible target allocation sequences are obtained.

[0077] S303. For each target allocation sequence, sensors are selected in order of probability. Since sensors selected by targets in earlier dimensions cannot be selected again by targets in later dimensions, when a sensor is selected by a target in a certain dimension, that sensor will be removed from the available sensor sequence for targets in later dimensions, thus ensuring that it is not selected repeatedly. However, when a target selects a dimension of an individual optimal solution or a local optimal solution, it may overlap with a previously selected sensor. Therefore, when two targets select the same sensor number, a probability sorting principle is used to select the particle corresponding to the particle with the higher probability that does not overlap with the previously generated component for updating. The implementation process of this method will be described below.

[0078] S304. Further, when updating to the k-th dimension, it is first necessary to compare whether the sensors corresponding to the k-th dimension of the current solution, the individual optimal solution, and the global optimal solution have been selected by the first k-1 dimensions of the new particle sequence. If the k-th dimension of the current solution already exists in the new particle sequence, the corresponding P1 flag is 0; if the k-th dimension of the individual optimal solution already exists in the new particle sequence, the corresponding P2 flag is 0; similarly, if the k-th dimension of the global optimal solution already exists in the new particle sequence, the corresponding P3 flag is 0. Then, the maximum values ​​of P1~P3 with flags set to 1 are compared, and the probability of the maximum value is recorded. The particles are updated sequentially according to the roulette wheel algorithm and the target update sequence mentioned above. If the value of the newly generated dimension does not repeat the previously generated component, the update proceeds to the next dimension. If it repeats, the scheme component corresponding to the maximum value of flags set to 1 in P1~P3 is selected for update. This ensures that the updated particles are valid, i.e., each dimension is unique.

[0079] S4. Calculate the updated fitness values ​​of the particles, update P1~P4 according to the fitness, and use a greedy strategy to update the individual optimal solution and the global optimal solution.

[0080] S401. Similar to the previous process, the fitness values ​​of the updated particles are recalculated using the objective function. If the fitness is better, the individual optimal solution and the global optimal solution are updated. At the same time, P1~P4 corresponding to each particle are updated to prepare for the particle update operation in the following follow-the-bee phase.

[0081] S5. In the bee-following search phase, the roulette wheel algorithm is used, and particles with better fitness are more likely to be followed, i.e., a particle update operation is performed.

[0082] S501. First, normalize the fitness of each particle. Then, use the roulette wheel algorithm. Particles with better fitness are likely to be closer to the optimal solution. Then, perform an update operation on them to ensure that the particles continuously tend towards the optimal solution.

[0083] S6. Similar to S4, recalculate the fitness of the updated particles, update P1~P4 according to the fitness, and use a greedy strategy to update the individual optimal solution and the global optimal solution.

[0084] S7, the low-quality particle elimination stage, determines whether the number of particle updates has reached the set elimination limit of 40. If it has, a new particle is regenerated using the generation algorithm to replace the old particle. This plays a broad role in detecting particles jumping out of local optima and avoids premature convergence of the algorithm to a certain extent.

[0085] S8. Determine whether the iteration has reached the preset maximum number of iterations. If it has, output the optimal sensor scheduling scheme and the program ends; otherwise, continue iterating according to the algorithm flow.

[0086] Therefore, the sensor resource scheduling problem typically involves rationally allocating limited sensor resources to satisfy multiple objectives under given time and space constraints. This problem is characterized by high nonlinearity, complex constraints, and a vast solution space, making it difficult for traditional optimization methods to solve effectively. Particle swarm optimization (PSO) excels at global search and can quickly find a good solution, but it may easily get trapped in local optima. Bee swarm optimization, on the other hand, emphasizes local search and refined search, especially when close to the optimal solution, helping to find a more accurate solution. This embodiment proposes a cross-domain sensor resource scheduling method based on adaptive particle and bee swarm optimization, combining the advantages of these two algorithms. PSO is responsible for the global search in the solution space, while bee swarm optimization helps with local refined search, enhancing both the algorithm's global search capability and local optimization capability.

[0087] Combining bee swarm optimization and particle swarm optimization algorithms can improve the efficiency of sensor resource scheduling by integrating global and local searches, enabling the rapid identification of optimal resource allocation schemes. The sensor resource scheduling environment can be dynamic, with factors such as the sudden disappearance of targets or sensor interference. The method presented in this invention exhibits strong adaptability, capable of adjusting the search strategy in real-time based on environmental changes, and responding quickly. Sensor resource scheduling problems are often multi-objective, requiring consideration of optimal sensor resource allocation, power consumption, and interference avoidance while simultaneously meeting task requirements. The scheduling method provided in this embodiment can flexibly handle these multi-objective constraints, offering a better compromise solution.

[0088] like Figure 2 , Figure 3 , Figure 4 As shown, the correspondence between the target and the sensor is illustrated for three algorithms (the scheduling method provided in this embodiment, the bee colony algorithm, and the particle swarm algorithm) at times 750-850. Considering the overhead of sensor switching during the detection process, it can be seen that the scheduling method is the best in terms of switching overhead, followed by the bee colony algorithm, and then the particle swarm algorithm. This demonstrates the advantage of the scheduling method in the sensor scheduling problem, which can find a sensor scheduling scheme with the lowest switching cost.

[0089] Figure 5 , Figure 6 , Figure 7 , Figure 8 The figures show the change in fitness with the number of iterations at times 100, 700, 850, and 1000. Time 100 represents the moment when all targets are detected by satellites; time 700 represents the moment when some targets are detected by satellites and some by land, sea, and air sensors; time 850 represents the moment when all targets are detected by land, sea, and air sensors; and time 1000 represents the moment when the number of detected targets decreases while the remaining targets are all detected by land, sea, and air sensors. It can be seen that the fitness of all three algorithms decreases with the number of iterations. The fitness value of our scheduling method at the final convergence at all the above time points is significantly better than the other two algorithms, demonstrating the advantage of our scheduling method in finding the sensor scheduling scheme with the lowest fitness.

[0090] In summary, the cross-domain sensor resource scheduling method based on adaptive particle swarm proposed in this embodiment can flexibly handle these multi-objective constraints in the sensor scheduling problem and provide a relatively excellent scheduling scheme.

[0091] Example 2

[0092] like Figure 9As shown, this embodiment provides a cross-domain sensor resource scheduling method based on adaptive particle swarm optimization, including the following steps:

[0093] The particle swarm is initialized by using an effective multi-domain sensor scheduling sequence as the initial particles of the swarm.

[0094] The initial particle swarm is updated by combining the bee colony algorithm and the particle update algorithm. The specific steps are as follows:

[0095] In the leading bee neighborhood search phase, the particle update algorithm is used to update the initialized particle swarm, generate updated particles, and calculate the target fitness corresponding to each updated particle.

[0096] During the follower bee search phase, based on the target fitness corresponding to each updated particle, the roulette wheel algorithm is used to update the updated particle swarm again, obtain the update result, and recalculate the target fitness corresponding to the updated particle.

[0097] Repeat the process corresponding to the leading bee neighborhood search phase and the following bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal scheduling scheme for the sensor.

[0098] Specifically, the initialization process for the particle swarm includes:

[0099] Initialize the population size, number of targets, maximum number of iterations, and number of sensors corresponding to the particle swarm;

[0100] For each target visual sensor set, a generative algorithm is used to generate an effective multi-domain sensor scheduling sequence as the initial particles of the particle swarm, where each dimension is different.

[0101] For the initial particles of the particle swarm, a greedy strategy is used to calculate the initial particle fitness, initial individual optimal solution, initial global optimal solution, and the selection probability of each dimension of the particle for the current solution, individual optimal solution, global optimal solution, and other optional values.

[0102] The multi-domain sensors include space-based sensors, air-based sensors, sea-based sensors, and land-based sensors.

[0103] In this embodiment, the specific steps for updating the initialized particle swarm using the particle update algorithm during the leading bee neighborhood search phase are as follows:

[0104] Based on the selection probabilities of the four schemes—the current solution, the individual optimal solution, the global optimal solution, and the remaining optional values—the roulette wheel algorithm is used to update the initialized particle swarm and generate updated particles.

[0105] Calculate the target fitness of the updated particle, as well as the selection probabilities of the current solution, individual optimal solution, global optimal solution, and other alternative solutions for the updated particle.

[0106] In this embodiment, the specific steps for updating the initialized particle swarm using the roulette wheel algorithm include:

[0107] Targets are grouped based on the distance between them, and each group is sorted from highest to lowest average threat level. The order of two targets within a group is processed by enumeration, resulting in all possible target assignment sequences.

[0108] For each target assignment sequence, sensors are selected in order of probability, where:

[0109] When any sensor is selected by the target in the current dimension, this sensor will be removed from the available sensor sequence for targets in subsequent dimensions.

[0110] When two targets select the same sensor number, the probability sorting principle is adopted to select the particle corresponding to the particle with a higher probability and which does not overlap with the previously generated component for updating.

[0111] When updating to the kth dimension, first compare whether the sensor corresponding to the kth dimension of the current solution, the individual optimal solution, and the global optimal solution has been selected by the first k-1 dimensions of the new particle sequence.

[0112] If the k-th dimension of the current solution already exists in the new particle sequence, then the corresponding P1 flag is 0; if the k-th dimension of the individual optimal solution already exists in the new particle sequence, then the corresponding P2 flag is 0; if the k-th dimension of the global optimal solution already exists in the new particle sequence, then the corresponding P3 flag is 0; where P1, P2, P3, and P4 represent the selection probabilities corresponding to the current solution, the individual optimal solution, the global optimal solution, and other optional values, respectively.

[0113] Compare the maximum values ​​of P1~P3 with the flag bit set to 1, and record the maximum value;

[0114] The roulette wheel algorithm and target update sequence are used to update the particles sequentially. If the value of the newly generated dimension does not repeat the value of the previously generated component, the update proceeds to the next dimension; otherwise, the component corresponding to the maximum value of the flag bit 1 in P1~P3 is selected for update.

[0115] In this embodiment, the specific calculation formulas for P1, P2, P3, and P4 are as follows:

[0116]

[0117] In the formula, Cycle represents the current iteration number, MaxCycle represents the maximum set iteration number; p takes a value from 0 to 1; This represents the fitness value of the current solution. This represents the fitness value of an individual's optimal solution. This represents the fitness value of the global optimal solution.

[0118] In this embodiment, the specific steps for updating the particle swarm again using the roulette wheel algorithm based on the target fitness of each updated particle during the follower bee search phase are as follows:

[0119] Normalize the fitness of each updated particle;

[0120] Based on the normalized fitness, the roulette wheel algorithm is then used to update the particle swarm again, and the fitness of the updated particles is recalculated, as well as the selection probabilities corresponding to the current solution, individual optimal solution, global optimal solution and other optional values ​​of the updated particles.

[0121] Based on the updated current solution of the particle, the individual optimal solution, the global optimal solution, and the selection probabilities corresponding to the remaining optional values, the individual optimal solution and the global optimal solution are updated using a greedy strategy.

[0122] In this embodiment, a low-quality particle elimination stage is also included. After the follower bee search stage, it is determined whether the number of particle updates has reached the set elimination limit. If it has, a new particle is regenerated using a generation algorithm to replace the old particle; otherwise, the next step is continued.

[0123] like Figure 10 As shown, this embodiment also provides a cross-domain sensor resource scheduling system based on adaptive particle swarm, including: a particle swarm initialization module, used to initialize the particle swarm, wherein an effective multi-domain sensor scheduling sequence is used as the initial particles of the particle swarm; a particle swarm update module, used to update the initialized particle swarm by combining the swarm algorithm and the particle update algorithm, specifically including: a leading bee neighborhood search module, used to update the initialized particle swarm using the particle update algorithm during the leading bee neighborhood search phase, generating updated particles and calculating the target fitness corresponding to each updated particle; a following bee follow search module, used to update the updated particle swarm again using the roulette wheel algorithm based on the target fitness corresponding to each updated particle during the following bee follow search phase, obtaining the update result and recalculating the target fitness corresponding to the updated particles; and a scheduling scheme output module, used to repeatedly execute the processes corresponding to the leading bee neighborhood search phase and the following bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal sensor scheduling scheme.

[0124] The present invention also provides an apparatus comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the cross-domain sensor resource scheduling method based on adaptive particle swarm.

[0125] When the processor executes the computer program, it implements the steps of the above-mentioned cross-domain sensor resource scheduling based on adaptive particle swarm, for example: initializing the particle swarm, wherein an effective multi-domain sensor scheduling sequence is used as the initial particles of the particle swarm; updating the initialized particle swarm by combining the swarm algorithm and the particle update algorithm, specifically including: in the leader bee neighborhood search phase, using the particle update algorithm to update the initialized particle swarm, generating updated particles, and calculating the target fitness corresponding to each updated particle; in the follower bee follow search phase, based on the target fitness corresponding to each updated particle, using the roulette wheel algorithm to update the updated particle swarm again, obtaining the update result and recalculating the target fitness corresponding to the updated particles; repeating the process corresponding to the leader bee neighborhood search phase and the follower bee follow search phase until the maximum number of iterations is reached, and outputting the current update result as the optimal sensor scheduling scheme.

[0126] Alternatively, when the processor executes the computer program, it implements the functions of each module in the above system, such as: a particle swarm initialization module, used to initialize the particle swarm, wherein an effective multi-domain sensor scheduling sequence is used as the initial particles of the particle swarm; a particle swarm update module, used to update the initialized particle swarm by combining the bee colony algorithm and the particle update algorithm, specifically including: a leading bee neighborhood search module, used to update the initialized particle swarm using the particle update algorithm during the leading bee neighborhood search phase, generating updated particles, and calculating the target fitness corresponding to each updated particle; a following bee follow search module, used to update the updated particle swarm again using the roulette wheel algorithm based on the target fitness corresponding to each updated particle during the following bee follow search phase, obtaining the update result and recalculating the target fitness corresponding to the updated particles; and a scheduling scheme output module, used to repeatedly execute the processes corresponding to the leading bee neighborhood search phase and the following bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal sensor scheduling scheme.

[0127] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing preset functions, wherein the instruction segments describe the execution process of the computer program in the cross-domain sensor resource scheduling device based on adaptive particle swarm architecture. For example, the computer program can be divided into a particle swarm initialization module, a particle swarm update module, and a scheduling scheme output module; wherein, the particle swarm update module includes a leading bee neighborhood search module and a following bee follow search module; the specific functions of each module are as follows: the particle swarm initialization module is used to initialize the particle swarm, wherein an effective multi-domain sensor scheduling sequence is used as the initial particles of the particle swarm; the particle swarm update module is used to update the initialized particle swarm by combining the bee swarm algorithm and the particle update algorithm, specifically including: the leading bee neighborhood search module, which is used to update the initialized particle swarm using the particle update algorithm during the leading bee neighborhood search phase, generate updated particles, and calculate the target fitness corresponding to each updated particle; the following bee follow search module, which is used to update the updated particle swarm again using the roulette wheel algorithm based on the target fitness corresponding to each updated particle during the following bee follow search phase, obtain the update result, and recalculate the target fitness corresponding to the updated particles; the scheduling scheme output module is used to repeatedly execute the processes corresponding to the leading bee neighborhood search phase and the following bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal sensor scheduling scheme.

[0128] The cross-domain sensor resource scheduling device based on adaptive particle swarm optimization can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. This device may include, but is not limited to, processors and memory. Those skilled in the art will understand that the above examples of cross-domain sensor resource scheduling devices based on adaptive particle swarm optimization do not constitute a limitation on such devices. The device may include more components than described above, or combine certain components, or use different components. For example, the device may also include input / output devices, network access devices, buses, etc.

[0129] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or any conventional processor. This processor is the control center of the adaptive particle swarm-based cross-domain sensor resource scheduling system, connecting various parts of the system via various interfaces and lines.

[0130] The memory can be used to store the computer program and / or modules. The processor implements various functions of the cross-domain sensor resource scheduling device based on adaptive particle bee colony by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory.

[0131] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function (such as sound playback, image playback, etc.). The data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0132] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the cross-domain sensor resource scheduling method based on adaptive particle bee colony.

[0133] If the modules / units integrated in the cross-domain sensor resource scheduling system based on adaptive particle bee colony are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0134] Based on this understanding, the present invention can implement all or part of the processes in the aforementioned cross-domain sensor resource scheduling method based on adaptive particle swarm optimization, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the aforementioned cross-domain sensor resource scheduling method based on adaptive particle swarm optimization. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or a preset intermediate form, etc.

[0135] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0136] It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0137] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

Claims

1. A cross-domain sensor resource scheduling method based on adaptive particle swarm optimization, characterized in that, include: The particle swarm is initialized by using an effective multi-domain sensor scheduling sequence as the initial particles of the swarm; wherein the multi-domain sensors include space-based sensors, air-based sensors, sea-based sensors, and land-based sensors. The initial particle swarm is updated by combining the bee colony algorithm and the particle update algorithm. The specific steps are as follows: In the leading bee neighborhood search phase, a particle update algorithm is used to update the initialized particle swarm. An allocation optimization method combining target distance grouping and threat ranking is employed to generate all possible target allocation sequences. Updated particles are then generated based on these sequences, and the target fitness of each updated particle is calculated. The specific steps are as follows: Based on the selection probabilities corresponding to the current solution, the individual optimal solution, the global optimal solution, and the other optional value schemes, the roulette wheel algorithm is used to update the initialized particle swarm in multiple dimensions according to the target allocation sequence, generating updated particles. Calculate the target fitness of the updated particle, as well as the selection probabilities of the current solution, individual optimal solution, global optimal solution, and other possible value schemes of the updated particle; During the follower bee search phase, based on the target fitness corresponding to each updated particle, the roulette wheel algorithm is used to perform multi-dimensional probability updates on the updated particle swarm according to the new target allocation sequence to obtain the update results and recalculate the target fitness corresponding to the updated particles. Repeat the process corresponding to the leader bee neighborhood search phase and the follower bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal sensor scheduling scheme. The specific steps for updating the initialized particle swarm using the roulette wheel algorithm include: Targets are grouped based on the distance between them, and each group is sorted from highest to lowest average threat level. The order of two targets within a group is processed by enumeration, resulting in all possible target assignment sequences. For each target assignment sequence, sensors are selected in order of probability, where: When any sensor is selected by the target in the current dimension, this sensor will be removed from the available sensor sequence for targets in subsequent dimensions. When two targets select the same sensor number, the probability sorting principle is adopted to select the particle corresponding to the particle with a higher probability and which does not overlap with the previously generated component for updating. When updating to the kth dimension, first compare whether the sensor corresponding to the kth dimension of the current solution, the individual optimal solution, and the global optimal solution has been selected by the first k-1 dimensions of the new particle sequence. If the k-th dimension of the current solution already exists in the new particle sequence, then the corresponding P1 flag is 0; if the k-th dimension of the individual optimal solution already exists in the new particle sequence, then the corresponding P2 flag is 0; if the k-th dimension of the global optimal solution already exists in the new particle sequence, then the corresponding P3 flag is 0; where P1, P2, P3, and P4 represent the selection probabilities corresponding to the current solution, the individual optimal solution, the global optimal solution, and other optional values, respectively. Compare the maximum values ​​of P1~P3 with the flag bit set to 1, and record the maximum value; The roulette wheel algorithm and target update sequence are used to update the particles sequentially. If the value of the newly generated dimension does not repeat the value of the previously generated component, the update proceeds to the next dimension; otherwise, the component corresponding to the maximum value of the flag bit 1 in P1~P3 is selected for update. The specific calculation formulas for P1, P2, P3, and P4 are as follows: In the formula, Cycle represents the current iteration number, MaxCycle represents the maximum set iteration number; p takes a value from 0 to 1; This represents the fitness value of the current solution. This represents the fitness value of an individual's optimal solution. This represents the fitness value of the global optimal solution.

2. The cross-domain sensor resource scheduling method based on adaptive particle swarm optimization as described in claim 1, characterized in that, The initialization process for the particle swarm includes: Initialize the population size, number of targets, maximum number of iterations, and number of sensors corresponding to the particle swarm; For each target visual sensor set, a generative algorithm is used to generate an effective multi-domain sensor scheduling sequence as the initial particles of the particle swarm, where each dimension is different. For the initial particles of the particle swarm, a greedy strategy is used to calculate the initial particle fitness, initial individual optimal solution, initial global optimal solution, and the selection probability of each dimension of the particle for the current solution, individual optimal solution, global optimal solution, and other optional values.

3. The cross-domain sensor resource scheduling method based on adaptive particle swarm optimization as described in claim 1, characterized in that, In the follower bee search phase, the specific steps for updating the particle swarm again using the roulette wheel algorithm based on the target fitness of each updated particle are as follows: Normalize the fitness of each updated particle; Based on the normalized fitness, the roulette wheel algorithm is then used to update the particle swarm again, and the fitness of the updated particles is recalculated, as well as the selection probabilities of the current solution, individual optimal solution, global optimal solution and other optional value schemes of the updated particles. Based on the updated current solution, individual optimal solution, global optimal solution, and selection probabilities corresponding to other optional values ​​of the particle, the individual optimal solution and global optimal solution are updated using a greedy strategy to obtain the updated result.

4. The cross-domain sensor resource scheduling method based on adaptive particle swarm optimization as described in claim 1, characterized in that, in, It also includes a low-quality particle elimination stage. After the follower bee search stage, it is determined whether the number of particle updates has reached the set elimination limit. If it has, a new particle is regenerated using a generation algorithm to replace the old particle; otherwise, the next step is continued.

5. A cross-domain sensor resource scheduling system based on adaptive particle swarm optimization, used to implement the cross-domain sensor resource scheduling method based on adaptive particle swarm optimization as described in any one of claims 1-4, characterized in that, include: The particle swarm initialization module is used to initialize the particle swarm, where an effective multi-domain sensor scheduling sequence is used as the initial particles of the particle swarm. The particle swarm update module combines the bee colony algorithm and the particle update algorithm to update the initialized particle swarm. Specifically, it includes: The leading bee neighborhood search module is used to update the initialized particle swarm using a particle update algorithm during the leading bee neighborhood search phase, generate updated particles, and calculate the target fitness corresponding to each updated particle. The follower bee follower search module is used to update the updated particle swarm again based on the target fitness of each updated particle during the follower bee follower search phase, using the roulette wheel algorithm to obtain the update result and recalculate the target fitness of the updated particles. The scheduling scheme output module is used to repeatedly execute the processes corresponding to the leader bee neighborhood search phase and the follower bee follow search phase until the maximum number of iterations is reached, and output the current update result as the optimal scheduling scheme for the sensor.

6. A device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the cross-domain sensor resource scheduling method based on adaptive particle swarm as described in any one of claims 1-4 when executing the computer program.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it is used to implement the steps of the cross-domain sensor resource scheduling method based on adaptive particle bee colony as described in any one of claims 1-4.

Images