Space target correlation positioning method and device based on adaptive particle swarm optimization

Abstract

The invention relates to the technical field of space target tracking, in particular to a space target correlation positioning method and device based on an adaptive particle swarm optimization algorithm, and the method comprises the steps: continuously obtaining angle measurement observation data; on the basis of the orbit state of each space target at the current observation moment relative to the reference observation platform, according to the orbit dynamic model, predicting the forecast state of each space target at the next observation moment and a corresponding prior covariance matrix; through two-stage association processing, angle measurement observation data at the next observation moment is matched with a forecast state; and according to the optimal matching relationship, updating the orbit state of each space target relative to the reference observation platform at the next observation moment and the corresponding covariance matrix. According to the method, through two-stage association processing, primary screening is firstly carried out by utilizing geometric consistency, and then accurate matching is carried out by adopting the global optimization adaptive particle swarm algorithm, so that association ambiguity can be effectively reduced, and the association accuracy and robustness in a complex scene are improved.

Description

Technical Field

[0001] This disclosure relates to the field of space target tracking technology, specifically to a space target association and localization method and apparatus based on an adaptive particle swarm algorithm. Background Technology

[0002] In fields such as space target surveillance and space radar processing, accurate correlation and positioning of space targets are crucial to ensuring spacecraft safety and mission success. The core task is to correctly match angular measurement data from multiple space-based observation platforms with known space target trajectories and to estimate the target's precise orbital state in real time.

[0003] In related technologies, traditional association algorithms, such as the nearest neighbor method and probabilistic data association filtering, typically rely on the statistical distance between observed data and the predicted state of the target for matching. However, in high-orbit space environments, facing complex scenarios such as dense targets, limited field of view of the observation platform, large amounts of angle measurement data, and clutter, traditional methods are prone to association ambiguity, and their association accuracy will significantly decrease, leading to trajectory loss or incorrect association. Therefore, how to design an algorithm that can efficiently process massive amounts of observation data in complex space observation environments, while possessing strong global optimization capabilities and stable association accuracy, thereby significantly improving the accuracy and robustness of multi-target, multi-observation association matching in complex space environments, has become an urgent problem to be solved. Summary of the Invention

[0004] In view of this, this disclosure provides a spatial target association and localization method and apparatus based on the adaptive particle swarm algorithm, in order to solve the problem of how to design an algorithm that can efficiently process massive amounts of observation data and has strong global optimization capabilities and stable association accuracy in complex space observation environments, thereby significantly improving the accuracy and robustness of multi-target and multi-observation association matching in complex space environments.

[0005] This disclosure provides a spatial target association and localization method based on an adaptive particle swarm optimization algorithm. The method includes: continuously acquiring angular measurement observation data collected by at least two observation platforms at each observation time according to a preset sampling period; wherein, one of the at least two observation platforms is designated as a reference observation platform; based on the orbital state of each spatial target relative to the reference observation platform at the current observation time, predicting the predicted state of each spatial target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix according to an orbital dynamics model; wherein, the orbital state at the current observation time is determined by initialization settings or by updating the orbital state at the previous observation time; and employing two-level association processing. The process involves matching the angular measurement observation data at the next observation time with the predicted state. The two-level association process includes: First, based on the geometric consistency between angular measurement observation data from different observation platforms, observation data corresponding to the same space target are paired to form a candidate observation set. Second, the matching relationship between the candidate observation set and the predicted state of each space target is modeled as a multidimensional allocation problem, and an adaptive particle swarm optimization algorithm is used to globally optimize and solve the multidimensional allocation problem to determine the optimal matching relationship between each space target and the angular measurement observation data. Based on the optimal matching relationship, the orbital state of each space target relative to the reference observation platform and the corresponding covariance matrix at the next observation time are updated.

[0006] This disclosure also provides a space target association and localization device based on an adaptive particle swarm optimization algorithm. The device includes: an observation module for continuously acquiring angular measurement observation data collected by at least two observation platforms at each observation time according to a preset sampling period; wherein, one of the at least two observation platforms is designated as a reference observation platform; a prediction module for predicting the predicted state of each space target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix based on the orbital state of each space target relative to the reference observation platform at the current observation time and according to an orbital dynamics model; wherein, the orbital state at the current observation time is determined by initialization settings or by updating the orbital state at the previous observation time; and an association matching module. This system is used to match the angular measurement observation data with the predicted state at the next observation time through two-level correlation processing. The correlation matching module is specifically used to pair observation data corresponding to the same space target based on the geometric consistency between angular measurement observation data from different observation platforms to form a candidate observation set. The correlation matching module is specifically used to model the matching relationship between the candidate observation set and the predicted state of each space target as a multi-dimensional allocation problem, and to use an adaptive particle swarm optimization algorithm to globally optimize and solve the multi-dimensional allocation problem to determine the optimal matching relationship between each space target and the angular measurement observation data. The update module is used to update the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time according to the optimal matching relationship.

[0007] This disclosure also provides an electronic device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the above-described spatial target association and localization method based on the adaptive particle swarm algorithm.

[0008] This disclosure also provides a computer-readable storage medium storing computer instructions for enabling a computer to implement the aforementioned spatial target association and localization method based on the adaptive particle swarm algorithm.

[0009] This disclosure also provides a computer program product, including computer instructions for causing a computer to execute the above-described spatial target association and localization method based on the adaptive particle swarm algorithm.

[0010] The spatial target association and localization method and apparatus based on the adaptive particle swarm algorithm in the above embodiments of this disclosure can effectively reduce association ambiguity and improve the association accuracy and robustness in complex scenarios through two-level association processing. First, geometric consistency is used for preliminary screening, and then a globally optimized adaptive particle swarm algorithm is used for accurate matching.

[0011] Furthermore, by integrating the two-level correlation processing and state update into a complete process, an automated closed loop of accurate observation, optimal correlation, and high-precision state update is achieved, thereby significantly improving the automation level of the positioning process and the final state estimation accuracy. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the specific embodiments or related technologies of this disclosure, the accompanying drawings used in the description of the specific embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart illustrating the spatial target association and localization method based on the adaptive particle swarm algorithm provided in this embodiment of the disclosure; Figure 2 This is a schematic diagram of the fitness iteration process of the spatial target association and localization method based on adaptive particle swarm algorithm provided in the embodiments of this disclosure; Figure 3 This is a schematic diagram of the association result of the PSO algorithm at a single moment in the spatial target association and localization method based on the adaptive particle swarm algorithm provided in the embodiments of this disclosure. Figure 4 This is a schematic diagram of the localization result of the spatial target association localization method based on the adaptive particle swarm algorithm provided in the embodiments of this disclosure; Figure 5 This is a schematic diagram of the spatial target association and localization device based on the adaptive particle swarm algorithm provided in this embodiment of the present disclosure; Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0015] With the development of emerging missions such as on-orbit servicing, space debris removal, and space traffic management, extremely high demands are placed on the real-time, accurate tracking and orbit determination capabilities of non-cooperative space targets. To achieve continuous monitoring of space targets, related technologies often utilize multiple observation satellites equipped with optical cameras to acquire angular information such as the azimuth and elevation angles of the space targets, and combine this information with data correlation and state filtering algorithms to ultimately determine their precise orbits.

[0016] In practical applications, accurate orbit determination highly depends on the accuracy of data association, i.e., how to correctly match angular measurement data from different observation platforms, mixed with noise and clutter, with their respective space targets. Currently, the commonly used association methods are mainly based on the statistical nearest neighbor principle or probabilistic data association filtering. Although these methods have certain advantages in computational efficiency, they often suffer from the following problems when facing complex space observation scenarios: 1. When the number of spatial targets increases and trajectories intersect, association criteria based on local statistical distance (such as nearest neighbor) are prone to association ambiguity, leading to trajectory loss or erroneous intersection between different target trajectories, which seriously restricts the reliability of the system in real complex spatial environments.

[0017] 2. Assigning multiple observations from multiple observation platforms to predict the states of multiple targets is a complex optimization problem. Although intelligent optimization algorithms such as particle swarm optimization have been introduced to seek global solutions, standard algorithms have fixed parameters and an imbalance between exploration and development capabilities. When solving such high-dimensional, nonlinear problems, they are prone to getting trapped in local optima, premature convergence, and cannot guarantee the global optimality of the associated results, thus limiting further improvements in the accuracy of state estimation.

[0018] 3. The modeling methods of related technologies lack an integrated and collaboratively optimized processing flow. Data association and state estimation are usually processed serially as two independent modules. The error of the association module will be directly transmitted and accumulated in the state estimation module. This disconnected process lacks consideration for the overall optimality of association decision-making, making it difficult to adapt to modern space missions with stringent requirements for association accuracy and positioning efficiency, thus restricting the further improvement of the overall system performance and the scope of engineering applications.

[0019] To address the aforementioned problems, various embodiments of this disclosure provide a spatial target association and localization method based on an adaptive particle swarm optimization algorithm. The method includes: continuously acquiring angular measurement observation data collected by at least two observation platforms at each observation time according to a preset sampling period; wherein, one of the at least two observation platforms is designated as a reference observation platform; based on the orbital state of each spatial target relative to the reference observation platform at the current observation time, predicting the predicted state of each spatial target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix according to an orbital dynamics model; wherein, the orbital state at the current observation time is determined by initialization settings or by updating the orbital state at the previous observation time; through two... The two-stage correlation process matches the angular measurement observation data at the next observation time with the predicted state. The first stage of correlation involves pairing observation data corresponding to the same space target based on the geometric consistency between angular measurement observation data from different observation platforms to form a candidate observation set. The second stage of correlation models the matching relationship between the candidate observation set and the predicted state of each space target as a multidimensional allocation problem, and uses an adaptive particle swarm optimization algorithm to globally optimize and solve the multidimensional allocation problem, determining the optimal matching relationship between each space target and the angular measurement observation data. Based on the optimal matching relationship, the orbital state of each space target relative to the reference observation platform and the corresponding covariance matrix at the next observation time are updated.

[0020] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a spatial target association and localization method based on an adaptive particle swarm optimization algorithm provided in this disclosure. The method may include the following steps: Step S101: According to the preset sampling period, continuously acquire angle measurement observation data collected by at least two observation platforms at each observation time for at least one space target.

[0021] In this embodiment, this step aims to construct and operate a space-based network for continuous collaborative observation. This network requires at least two observation platforms distributed in different spatial locations to conduct observations at each discrete observation time (i.e., , , All of them adhere to the same strict time reference and simultaneously detect at least one spatial target within the field of view to obtain a time-series angular measurement observation dataset.

[0022] The preset sampling period defines the time interval between two observation moments of the vector, which determines the temporal density of the observation data and the frequency of system state updates.

[0023] An observation platform can refer to a space or ground-based device that carries observation payloads for detecting space targets and can operate in a specific orbit or location. For example, an observation platform can include, but is not limited to, space detection satellites and in-orbit spacecraft. A unified observation time can refer to a highly synchronized timestamp that is commonly followed by all observation platforms; it can define the common time point corresponding to all angle measurement observation data.

[0024] Space targets can refer to natural or man-made objects located in the space environment that are of interest to and tracked by observation platforms. For example, space targets can include, but are not limited to: artificial satellites that are operating normally in orbit, defunct spacecraft, mission waste, rocket final stages, and space debris.

[0025] Angle measurement data can be angular information about the direction of a space target obtained by an observation platform. For example, angle measurement data can include azimuth and elevation angles. The azimuth angle can be... This means that the pitch angle can be expressed as... express.

[0026] Furthermore, at least one of the two observation platforms is designated as a benchmark observation platform. Here, the benchmark observation platform can be any observation platform. At the algorithm level, any observation platform can be designated as the benchmark; at the hardware level, the basis for designating the benchmark can include, but is not limited to, engineering factors such as the most stable orbital characteristics, the best payload performance, and the most reliable data link.

[0027] When an observation platform is designated as the baseline observation platform, it will serve as the origin of the continuous spatial reference frame and the data processing center throughout the entire observation sequence.

[0028] Furthermore, starting from the initial moment, angle measurement observation data sequences are continuously acquired according to the preset sampling period.

[0029] For example, assuming the initial time The sampling time is 8:00:00 AM Beijing time on October 1, 2025, with a total sampling duration of 10 minutes and a preset sampling period of 1 second; starting from the initial time... In the beginning, it will be , ... This series of consecutive observation moments continuously acquires angular measurement data collected by at least two observation platforms at the same observation moment from at least one space target.

[0030] Step S102: Based on the orbital state of each space target relative to the reference observation platform at the current observation time, predict the predicted state of each space target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix according to the orbital dynamics model.

[0031] In this embodiment, this step constitutes the state prediction environment of this disclosure and is an important component of the Kalman filtering process. Specifically, this step, based on the dynamic laws of the space target, estimates the system's state from the current observation time. Extending forward to the next observation time This provides prior estimates for subsequent data association and state updates.

[0032] Among them, the current observation time The orbital state forms the input basis for state prediction, which is set by initialization or by the previous observation time. The defined state update steps ensure the recursive continuity of the entire estimation process.

[0033] The orbital dynamics model describes the physical laws governing the relative motion of a space target and serves as the theoretical basis for state extrapolation in this step. Preferably, the CW equations of relative motion can be used as a specific implementation of this model.

[0034] The predicted state can refer to the state of a space target at the next observation time, as predicted by a model. The position and velocity vectors of the reference observation platform in the VVLH coordinate system. The prior covariance matrix quantitatively characterizes the uncertainty of the predicted state.

[0035] Furthermore, the aforementioned state prediction stage can be implemented through the prediction phase of Kalman filtering, which mainly includes the following calculation process: The state transition matrix is ​​derived based on the orbital dynamics model; this matrix is ​​then used to determine the current observation time. extrapolate the state vector to the next observation time Simultaneously, based on this matrix and the process noise covariance matrix, the current observation time... The state covariance matrix is ​​propagated forward to obtain the next observation time. The prior covariance matrix.

[0036] Step S103: Through two-level correlation processing, the angle measurement observation data at the next observation time is matched with the forecast status.

[0037] In this embodiment, this step constitutes the data association link of this disclosure, which can be the core of realizing multi-target tracking. It can solve the problem of how to correctly match newly acquired observation data with existing space target trajectories (i.e., prediction status) in complex scenarios with multiple space targets and multiple observation platforms.

[0038] The two-level correlation processing can decompose the complex global correlation problem into two sequential sub-processes: the first-level correlation processing performs preliminary matching between observation data from different observation platforms, and the second-level correlation processing performs final matching between observation data and system forecast status.

[0039] Among them, the angular measurement observation data refers to the set of azimuth and elevation angle measurements actually collected by each observation platform at the next observation time. The predicted state refers to the prior state estimate of each space target relative to the reference observation platform at the next observation time, which is predicted through step S102.

[0040] Specifically, the two-level association processing includes: Step S103a, first-level association processing, based on the geometric consistency between angular measurement observation data from different observation platforms, pairs observation data corresponding to the same spatial target to form a candidate observation set.

[0041] In this embodiment, the first-level correlation processing focuses on the lateral geometric correlation between observation data from different observation platforms, and achieves preliminary screening through the principle of convergence of spatial target lines of sight.

[0042] The first-level correlation processing takes as input the raw angular measurement data acquired by all observation platforms at the next observation time and outputs a candidate observation set. The candidate observation set can be a collection of pre-selected combinations of observation data, which may contain a set of angular measurement data from at least two different observation platforms that are considered to originate from the same space target. Each such combination of data in this set is called a candidate observation set element.

[0043] As an example, suppose there are 2 observation platforms ( , ) and 3 space targets. At a certain moment, Three space targets were observed, and three angle measurement observation data were output. Three space targets were also observed, and three angle measurement observation data were output.

[0044] The goal of the first-level correlation processing is to identify, from these six angle measurement observation data, which data originate from... Angle measurement observation data and which sources The angle measurement observations may point to the same spatial target. For example, the first-level correlation processing can output a candidate observation set containing 2 elements: {( The observation m, (observation n), ( The observation p, The observations are q). That is, it can be preliminarily determined that (m,n) is one pair and (p,q) is another pair, and each pair may correspond to a real space target. In this example, the unpaired data may come from clutter that was not simultaneously captured by both stations.

[0045] Here, the first-level correlation processing can take advantage of the spatial distribution characteristics of multiple observation platforms and perform preliminary pairing of angle measurement observation data from different observation platforms based on the triangulation principle, which can effectively reduce correlation ambiguity in subsequent processing.

[0046] Step S103b, the second-level association processing, models the matching relationship between the candidate observation set and the predicted state of each space target as a multi-dimensional allocation problem, and uses an adaptive particle swarm optimization algorithm to globally optimize and solve the multi-dimensional allocation problem to determine the optimal matching relationship between each space target and the angle measurement observation data.

[0047] In this embodiment, the second-level association processing focuses on the vertical probabilistic association between the observed data and the system state, transforming the matching problem into a global optimization problem for solution.

[0048] The second-level association processing takes as input the candidate observation set output from the first-level processing and the predicted states of each space target obtained through step S102. This processing determines which space target's predicted state should be associated with each element in the candidate observation set (i.e., each pair of observation data).

[0049] Here, the elements in the candidate observation set, i.e., each combination of observation data as defined above, are the basic units for allocation decisions in the second-level correlation processing. The core task of the second-level processing is to optimally allocate these elements to the known space targets (or clutter labeled as not belonging to any target).

[0050] Here, the second-level association processing can model the matching relationship as a multi-dimensional allocation problem, and by introducing the adaptive particle swarm optimization (PSO) algorithm for global optimization, it can stably find the optimal matching relationship in complex scenarios with dense targets and noisy observations.

[0051] Step S104: Based on the optimal matching relationship, update the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time.

[0052] In this embodiment, this step constitutes the state update stage in this disclosure, and can be the final step of the Kalman filtering process. Based on the optimal matching relationship determined in step S103, the next observation time is... The angle measurement observation data is fused with the corresponding forecast status.

[0053] Orbit state update can refer to correcting the predicted state obtained in step S102 based on new observation information to obtain a posterior state estimate that is closer to the true state.

[0054] Covariance matrix update can refer to the synchronous correction of the prior covariance matrix corresponding to the forecast state to obtain the posterior covariance matrix. This matrix represents the uncertainty of the updated state estimate, and its value will be smaller than the prior value, indicating that the confidence of the state estimate is improved by fusing observation data.

[0055] Furthermore, the state update process is implemented through a Kalman filter update stage. This process uses Kalman gain to weight and fuse the predicted state and the actual observations. The magnitude of the Kalman gain is determined by both the prior covariance matrix and the observation noise. When the reliability of the observation data is high, the update result is more biased towards the observations; when the reliability of the predicted state is high, the update result is more biased towards the predicted value.

[0056] Furthermore, the updated orbital state and covariance matrix output in this step will serve as the input for the state prediction step in the next step S102, thus initiating a new round of "prediction-correlation-update" recursive loop.

[0057] The spatial target association and localization method and apparatus based on adaptive particle swarm optimization (PSO) algorithm described in the above embodiments of this disclosure, through two-level association processing—firstly, using geometric consistency for preliminary screening, and then employing a globally optimized adaptive PSO algorithm for precise matching—effectively reduces association ambiguity, thereby improving the association accuracy and robustness in complex scenarios. By integrating the two-level association processing and state update into a complete process, an automated closed loop of precise observation, optimal association, and high-precision state update is achieved, significantly improving the automation level of the localization process and the final state estimation accuracy.

[0058] In one possible implementation of step S101 above, continuously acquiring angular measurement data collected by at least two observation platforms on at least one space target at each observation time according to a preset sampling period is achieved through simulation, including: Based on the preset initial orbital state, the orbital state of each observation platform and each space target in the geocentric inertial coordinate system at the observation time is simulated and generated by integrating the dynamic model. Based on the orbital state, the relative position vector of each space target with respect to each observation platform in the camera coordinate system is calculated through coordinate transformation; Based on the relative position vector, the azimuth and elevation angles are calculated and used as angle measurement observation data.

[0059] In this embodiment, the simulation method is implemented through the following steps: Based on the preset initial orbital state, the orbital state of the space target and the observation platform at each observation time within a preset time period is simulated and generated through dynamic models and numerical integration methods.

[0060] The preset initial orbital state can be determined based on the following steps: setting the initial time. Loading space targets and observation platforms at the initial moment The initial orbital state in the Earth-Centered Inertial (ECI) coordinate system; wherein the initial orbital state includes the initial velocity vector and the initial position vector.

[0061] For example, the dynamic model may include, but is not limited to: Earth's non-spherical gravitational field, solar radiation pressure, the Moon, atmospheric drag, Earth's radiation pressure, Earth's tides, relativistic effects, etc.

[0062] Preferably, the numerical integration method can be performed using the Adams-Cowell numerical integrator.

[0063] As an example, let's set the initial time. The time is 8:00:00 AM Beijing time on October 1, 2025; two observation platforms are set up and named respectively. and There are four space targets named respectively. , , and This creates a 2-to-4 observation scenario. The space target and the observation platform are in the initial time... The initial orbital states in the ECI coordinate system are shown in the table below: Table 1 Initial orbital states of space targets and observation platforms

[0064] Based on the initial orbital states shown in Table 1 above, a dynamic model incorporating various perturbations, including the Earth's non-spherical gravitational field, solar radiation pressure, and lunar gravity, was constructed. An Adams-Cowell numerical integrator was then used to perform orbital integration on the space target and observation platform over a period of 10 minutes with a step size of 1 second, generating a sequence of orbital states at each observation time within a preset time period. , .

[0065] in, Indicates the orbital state of a space target. Indicates the first One space target, Indicates the ECI coordinate system. Indicates the first A space target in The position vector located in the ECI coordinate system at any given moment. Indicates the first A space target in The velocity vector located in the ECI coordinate system at any given moment. Indicates the orbital status of the observation platform. Indicates the first One observation platform, Indicates the first Each observation platform The position vector located in the ECI coordinate system at any given moment. Indicates the first Each observation platform The velocity vector located in the ECI coordinate system at any given moment.

[0066] Here, the two formulas above define a sampling point every 1 second within a 10-minute (600-second) simulation period, for a total of 601 time points (including the initial moment). The complete orbital state of the space target is: 2404 sets of position and velocity data for 4 space targets × 601 time points, and 1202 sets of position and velocity data for 2 observation platforms × 601 time points. These data form the basis for generating angle measurement observation data in subsequent steps. Through coordinate transformation, the azimuth and elevation angles of the space target relative to the observation platform at any time can be calculated.

[0067] Furthermore, based on the above orbital state, the relative position vectors of each space target with respect to each observation platform in the camera coordinate system are calculated through coordinate transformation, and the azimuth and elevation angles are obtained from the relative position vectors as angle measurement observation data.

[0068] The coordinate transformation includes converting the orbital state in the geocentric inertial coordinate system into the relative position vectors in the VVLH coordinate system, the body coordinate system, and the camera coordinate system of the observation platform.

[0069] Specifically, for any observation time Any space target and any observation platform Perform the following calculations: Step a1: Convert the orbital state in the ECI coordinate system to the orbital state in the Vertical Velocity Location Heading (VVLH) coordinate system.

[0070] Here, at the observation time Space targets and observation platform The track turntables are respectively and Using the following formula (1), the orbital state in the ECI coordinate system is converted to the orbital state in the VVLH coordinate system: (1) Formula (1) is used to represent space targets. Relative observation platform Position in the VVLH coordinate system ; This is the position rotation matrix from the ECI coordinate system to the VVLH coordinate system; Indicates the first A space target at the observation time The absolute position vector in the ECI coordinate system; Indicates the first Each observation platform at the observation time The absolute position vector in the ECI coordinate system.

[0071] Step b1: Convert the orbital state in the VVLH coordinate system to the orbital state in the body (BODY) coordinate system.

[0072] Here, at the observation time Space targets Relative observation platform Position in the VVLH coordinate system The orbital state in the VVLH coordinate system is converted to the orbital state in the BODY coordinate system using the following formula (2): (2) Formula (2) is used to represent space targets. Relative observation platform Position in the BODY coordinate system ; , and These are rotation matrices about the x, y, and z axes, respectively; , and These represent the installation rotation angles of the BODY coordinate system of the observation platform relative to the VVLH coordinate system around the x, y, and z axes, respectively.

[0073] Step c1: Convert the orbital state in the BODY coordinate system to the orbital state in the camera (CAM) coordinate system.

[0074] Here, at the observation time Space targets Relative observation platform Its position in the BODY coordinate system is The orbital state in the BODY coordinate system is converted to the orbital state in the CAM coordinate system using the following formula (3): (3) Here, formula (3) is used to represent the space target. Relative observation platform Position in CAM coordinate system ; , and These are rotation matrices about the x, y, and z axes, respectively; , , These represent the installation rotation angles of the CAM coordinate system relative to the BODY coordinate system of the observation platform around the x-axis, y-axis, and z-axis, respectively.

[0075] Step d1: Calculate the azimuth and elevation angles based on the relative position vectors in the CAM coordinate system.

[0076] Here, at the observation time Space targets Relative observation platform The position in the CAM coordinate system is The azimuth angle is calculated using the following formulas (4) and (5). and pitch angle : (4) (5) here, Represents the components of the vector along the x-axis in the CAM coordinate system; This represents the y-component of the vector in the CAM coordinate system. This represents the component of the vector along the z-axis in the CAM coordinate system.

[0077] Among them, the azimuth angles obtained by formulas (4) and (5) and pitch angle These are ideal angle measurement data.

[0078] Step e1: Add observation errors to generate the final angle measurement observation data.

[0079] Here, in order to simulate the measurement noise of the sensor in the real observation environment, random errors that conform to specific statistical characteristics can be added to the ideal angle measurement data obtained by formulas (4) and (5).

[0080] For example, random error can have a mean of zero and a standard deviation of . Gaussian white noise. Preferably, the standard deviation is... A value of 0.1 milliradians (mrad) can be used.

[0081] The space target association and positioning method and apparatus based on the adaptive particle swarm optimization algorithm in the above embodiments of this disclosure, by introducing a high-precision dynamic model including Earth's non-spherical gravity, third-body perturbation, and solar radiation pressure, along with an Adams-Cowell numerical integrator, generates orbital trajectories that closely approximate the actual motion of space targets. Based on this, the angular measurement data simulation provides a quantifiable, high-confidence benchmark with a known "ground truth" for the subsequent testing and verification of the association and positioning algorithm, overcoming the difficulty in accurately evaluating algorithm performance due to the unknown true state of the target in real-world tasks. Through the processes of orbit prediction, coordinate transformation chain, and angular measurement calculation, a clear and standardized data generation pipeline is constructed, enabling the method to flexibly adapt to different observation platforms and space scenarios, greatly enhancing the efficiency of algorithm testing and verification, as well as its reproducibility and scalability under different task configurations. Artificially adding observation errors conforming to specific statistical characteristics to the calculated ideal angular measurement data can simulate the measurement noise present in real sensors. This design allows the present disclosure not only to verify the upper limit of the algorithm's performance under ideal conditions, but also to fully assess its association success rate and state estimation stability in near-realistic noisy environments, thereby ensuring the robustness of the algorithm in practical engineering applications.

[0082] In one possible implementation of step S102 above, the state prediction stage can be implemented through the prediction stage of Kalman filtering, which may include the following calculation process: Step S102a: Derive the state transition matrix based on the orbital dynamics model.

[0083] In this embodiment, the space target Relative benchmark observation platform The orbital state is The corresponding state transition matrix is ​​generated using the CW equation. As shown in the following formula (6): (6) in, The orbital angular velocity of the observation platform ( , The gravitational constant of Earth, (where is the orbital radius).

[0084] Step S102b: Use this matrix to determine the current observation time. extrapolate the state vector to the next observation time .

[0085] In this embodiment, the space target Relative benchmark observation platform orbital state From the current observation time Predicting the next observation time The corresponding expression can be shown in the following formula (7): (7) in, Let be the state transition matrix of the above formula (6).

[0086] Step S102c: Based on this matrix and the process noise covariance matrix, perform analysis on the current observation time. The state covariance matrix is ​​propagated forward to obtain the next observation time. The prior covariance matrix.

[0087] In this embodiment, the next observation time The prior covariance matrix is ​​shown in the following formula (8): (8) in, The modeling of process noise covariance did not consider disturbance factors. The current observation time The prior covariance matrix.

[0088] The spatial target association and localization method and apparatus based on the adaptive particle swarm optimization algorithm in the above embodiments of this disclosure introduces the relative motion CW equation as a specific orbital dynamics model and provides an accurate state transition matrix. This ensures that the state prediction strictly follows the physical laws of spatial relative motion, which can significantly improve the physical consistency and prediction accuracy of the predicted state. By simultaneously calculating the prior covariance matrix through formula (8), the uncertainty of the state estimation is rigorously propagated mathematically through the dynamic model, which can achieve accurate quantification and transmission of uncertainty.

[0089] In one possible implementation of step S103a above, based on the geometric consistency between angular measurement observation data from different observation platforms, observation data corresponding to the same space target are paired, including: Angle measurement data from different observation platforms were converted into unit relative direction vectors in the geocentric inertial coordinate system. Based on the position and unit relative direction vector of each observation platform in the geocentric inertial coordinate system, the line of sight of the space target originating from each observation platform is obtained by parameterization; Calculate the degree of convergence of lines of sight to space targets from different observation platforms; Based on the comparison results of the degree of intersection and the preset threshold, it is determined whether the angle measurement observation data of different observation platforms correspond to the same space target.

[0090] In this embodiment, the core idea of ​​this step is to ensure that the lines of sight from different observation platforms pointing to the same spatial target converge at the target's location in space. By systematically performing coordinate transformation, parameterizing the spatial target lines of sight, calculating the degree of convergence, and judging based on preset thresholds, efficient and reliable preliminary screening of observation data can be achieved.

[0091] The unit relative direction vector refers to a unit vector with a magnitude of 1 in the ECI coordinate system, pointing from the observation platform position to the location of the space target. The space target line of sight refers to a straight line extending from the observation platform position along the direction vector, mathematically expressed as a parametric equation between the observation platform position and the unit relative direction vector. The degree of intersection quantifies the proximity of two space target lines of sight in three-dimensional space and is a key indicator for determining whether they point to the same target. The preset threshold is a pre-set threshold value used for decision-making based on factors such as system tolerance error, platform positioning accuracy, and observation noise level.

[0092] Specifically, converting angle measurement observation data into unit relative direction vectors in the ECI coordinate system can be achieved by sequentially performing inverse transformations from the CAM coordinate system, BODY coordinate system, VVLH coordinate system to the ECI coordinate system. This process is the reverse of the coordinate transformation chain in step S101, utilizing the observation platform... The attitude and orbital information at any given time will be used to measure the angle observation data (azimuth angle). and pitch angle Reconstruct the unit relative direction vector in the global coordinate system.

[0093] Here, taking an example where the number of observation platforms equals 2, for the benchmark observation platform... The observed first Angle measurement observation data of a space target ( , The unit relative direction vector is obtained by sequentially performing the above inverse coordinate transformations. It can be represented by the following formula (9): (9) in, This represents the position rotation matrix from the CAM coordinate system to the ECI coordinate system.

[0094] For another observation platform The observed first Angle measurement observation data of a space target ( , The unit relative direction vector is obtained by sequentially performing the above inverse coordinate transformations. It can be represented by the following formula (10): (10) Furthermore, known observation platforms and The position and velocity in the ECI coordinate system are respectively and , can be parameterized separately from and The line of sight to the spatial target is shown in the following formulas (11) and (12): (11) (12) in, Indicates from the benchmark observation platform The starting parameterized spatial target line of sight, Indicates from other observation platforms The starting parameterized spatial target line of sight. Indicates the benchmark observation platform exist The absolute position vector in the ECI coordinate system at a given time. Other observation platforms exist The absolute position vector in the ECI coordinate system at any given moment; Indicated by the benchmark observation platform Observing space targets The unit relative direction vector (expressed in the ECI coordinate system) is obtained by converting angle measurement observation data. Indicated by the benchmark observation platform Observing space targets The unit relative direction vector (expressed in the ECI coordinate system) is obtained by converting angle measurement observation data. Indicates the benchmark observation platform The scalar parameter in the spatial target line of sight formula (11) is used to determine the distance along the line of sight direction; Other observation platforms The scalar parameter in the spatial target line of sight formula (12) is used to determine the distance along the line of sight direction.

[0095] Furthermore, scalar parameters and The solutions can be obtained using the following formulas (13) and (14) respectively: (13) (14) Here, the endpoint parameters of the shortest connection between two spatial target lines of sight can be obtained by using formulas (13) and (14). and This allows us to determine the closest point between the two lines of sight. and .

[0096] Furthermore, we can calculate separately ( )and The angle between 、( )and The angle between As shown in formulas (15) and (16) below: (15) (16) Here, formula (15) calculates the value from the reference observation platform. Location To the shortest endpoint The vector, and the original observation direction of the observation platform. The angle between Similarly, formula (16) calculates the results from other observation platforms. Location To the shortest endpoint The vector, and the original observation direction of the observation platform. The angle between Among them, the included angle and Characterizing the benchmark observation platform Other observation platforms The angular deviation between the direction of the observation line of sight and the direction of their respective "ideal intersection line of sight".

[0097] It is understandable that by using the above formulas (13) and (14), we can find the two points that are "closest to each other" on the line of sight of two space targets, and then use the above formulas (15) and (16) to check how much the actual line of sight of each observation platform needs to "deflect" in order to reach this "closest point".

[0098] The design consideration of the above calculation formula is that judging solely based on the shortest distance between the lines of sight of space targets is insufficient; it is also necessary to assess the adjustment angles required in the line of sight of each space target to achieve this shortest distance. If achieving the shortest distance requires a significant deflection of the line of sight of an observation platform beyond a reasonable range, then even if the distance itself is small, the probability that the two sets of observation data originate from the same target is low.

[0099] Ideally, two observation platforms would be perfectly pointed at the same space target. In this case, the lines of sight between the two space targets would perfectly intersect at the point where the target is located. This point of intersection is itself the endpoint of the shortest connecting line. and They all point precisely to the target in space. At this moment, ( )and The directions are exactly the same, and the included angle is exactly the same. Similarly, .

[0100] In reality, due to observation noise or differences between space targets, the lines of sight between two space targets cannot perfectly intersect. Therefore, it is necessary to find the endpoints of the shortest line connecting the two space targets' lines of sight. and .like Very close to the benchmark observation platform The direction in which the original line of sight points, then the included angle Very small, characterizing observational data It is largely compatible with the intersection hypothesis; conversely, if Significant deviation from the benchmark observation platform The direction in which the original line of sight points requires a large angle. Only then can it be reached, indicating the observation data. This contradicts the intersection hypothesis.

[0101] Finally, calculate the angle between the projection vectors. As a comprehensive measure of the degree of convergence, it is shown in formula (17): (17) In one possible implementation of the above embodiments, calculating the degree of convergence of spatial target lines of sight from different observation platforms may include: taking the spatial target line of sight from the reference observation platform as the first line of sight and taking the spatial target line of sight from any other observation platform as the second line of sight; Calculate the angles between the midpoint of the shortest line connecting the first and second lines of sight and the lines connecting them to the starting points of the first and second lines of sight, respectively. The starting point of the first line of sight is the reference observation platform, and the starting point of the second line of sight is any other observation platform. The angles connecting the midpoints to the starting points of the first line of sight are defined as the angle between the position vector of the space target relative to the reference observation platform in the geocentric inertial coordinate system and the unit relative direction vector of the reference observation platform. The angles connecting the midpoints to the starting points of the second line of sight are defined as the angle between the position vector of the space target relative to other observation platforms in the geocentric inertial coordinate system and the unit relative direction vector of other observation platforms. The average of the two included angles is determined as the included angle of the projection vector, which is used as a measure of the degree of intersection.

[0102] The angle between the lines connecting to the starting point of the first line of sight is the angle calculated by (15). The angle between the lines connecting the two points to the second starting point is the angle calculated using formula (16). The angle between the projection vectors mentioned above is calculated using formula (17). .

[0103] This implementation method can explicitly concretize the calculation of the degree of intersection as a symmetry measurement process with the midpoint of the shortest connection as the geometric reference point.

[0104] Here, if and only if the angle between the projection vectors When the value is less than a preset threshold, the benchmark observation platform can be determined. The Individual observation angle measurement data and other observation platforms The Each observation angle measurement data point passes the geometric consistency test, corresponds to the same space target, and the pair is included in the candidate observation set.

[0105] Furthermore, it is possible to estimate the space target's relative to the benchmark observation platform. Relative position in VVLH coordinate system .

[0106] The spatial target association and localization method and apparatus based on the adaptive particle swarm optimization algorithm in the above embodiments of this disclosure adopt a joint evaluation method of the shortest connection endpoint and the direction deflection angle, avoiding the limitations of relying solely on distance judgment, effectively suppressing the influence of observation noise and platform positioning errors on the association results, and enhancing robustness to noise and errors. By clarifying the specific calculation of the convergence degree as a symmetry evaluation of the direction deflection angle caused by the midpoint of the shortest connection, a clear conceptual and computationally stable geometric consistency criterion is provided. Through systematic coordinate inverse transformation and line-of-sight parameterization, unified processing of multi-observation platform data is achieved, providing high-quality candidate observations for subsequent Kalman filter updates, thereby improving the accuracy of target tracking and localization. The paired data selected through geometric consistency can be directly used for subsequent association optimization of the adaptive particle swarm optimization algorithm and Kalman filter updates, improving the convergence and accuracy of the overall localization algorithm.

[0107] In one possible implementation of step S103b above, the matching relationship between the candidate observation set and the predicted states of each space target is modeled as a multidimensional allocation problem, including: Each element in the candidate observation set is assigned to the predicted state of a space target, or marked as clutter that does not belong to any known space target; wherein, a complete assignment scheme is defined as an association hypothesis. Construct a fitness function to evaluate the merits of each association hypothesis; The objective of solving the multidimensional assignment problem is defined as: finding the association hypothesis that optimizes the fitness function value.

[0108] In this embodiment, the candidate observation set is determined through the above step S103a, and includes pairings of angle measurement observation data from different observation platforms that have passed geometric consistency verification. Each element is considered to possibly correspond to a real space target.

[0109] The predicted state is obtained through the above step S102 and represents the prior state estimate of each space target relative to the reference observation platform in the VVLH coordinate system at the next observation time.

[0110] Clutter can refer to candidate observation set elements that cannot form a reasonable match with the predicted state of any known space target, and can be generated by sensor noise, false alarms, or new targets that have not been cataloged.

[0111] Furthermore, the core of modeling the matching relationship as a multidimensional assignment problem lies in transforming the complex "multi-observation-multi-target" association matching into a decision problem of assigning a specific target (a specific spatial target or clutter) to each element of the candidate observation set. All assignment decisions collectively constitute a complete association hypothesis. This model mathematically defines the criteria (fitness function) for evaluating the quality of any association hypothesis and clarifies the final decision objective (finding the optimal association hypothesis).

[0112] Specifically, modeling the multidimensional allocation problem can be approached from the following core aspects: Step a3, define the association hypothesis.

[0113] Here, a correlation hypothesis represents a complete allocation scheme. Under this scheme, each element in the candidate observation set is uniquely assigned to the predicted state of a known space target, or labeled as clutter. All possible correlation hypotheses constitute the solution space of this multidimensional allocation problem.

[0114] Step b3: Construct the fitness function.

[0115] The fitness function can be a quantitative indicator to measure the quality of the correlation hypothesis. Its design aims to comprehensively evaluate the rationality of the correlation hypothesis, mainly based on the principle of state consistency: that is, the estimated state of candidate observation set elements assigned to the same spatial target should be as consistent as possible with the predicted state of the target in a statistical sense.

[0116] Preferably, the fitness function is constructed based on the concept of Mahalanobis distance to fully account for the uncertainty in state estimation. Mahalanobis distance is a method for measuring the distance between a point and a distribution. By introducing a covariance matrix, it assigns different weights to state components with different uncertainties, thereby providing a more statistically sound assessment of consistency.

[0117] In one specific embodiment, to balance computational complexity and performance, a simplified cost function can be used as the fitness function. This function embodies the core idea of ​​state consistency. It can be represented by the following formula (18): (18) in, Indicates the first The fitness value of the association hypothesis represented by each particle is used to optimize the system. maximize; The larger the value, the better the agreement between the forecast and the observation under that association hypothesis.

[0118] Indicates the index of spatial objects Summation. Here, we assume there are 4 spatial targets in the scene ( This loop iterates through each spatial target. The summation expression internally calculates the first... The sum of squared errors in the x, y, and z components of the VVLH coordinate system between the predicted position vector and the observed estimated position vector of a space target. Indicates the first A space target in Time relative to the benchmark observation platform Similarly, the x-component of the predicted position vector, To predict the y-component of the position vector, This refers to the z-component of the predicted position vector. This indicates that a space target has been assigned. The estimated elements of the candidate observation set, in Similarly, the x-component of the relative position component at time t, The y-component is the relative position component. The z-component is the relative position component.

[0119] The cost function described above calculates the total squared error between the predicted location and the estimated location of all "target-observation" pairs under a specific association assumption. The larger the value, the better the consistency under that association assumption.

[0120] Step c3: Set the solution objective.

[0121] Here, the objective of solving the multidimensional assignment problem is explicitly defined as: finding the association hypothesis that optimizes the fitness function value from all possible association hypotheses. This optimal association hypothesis represents the final optimal matching relationship between the spatial target and the angular measurement observation data at the next observation time.

[0122] The spatial target association and localization method and apparatus based on the adaptive particle swarm optimization algorithm in the above embodiments of this disclosure provide a clear and objective mathematical framework for association decisions in complex scenarios by rigorously modeling the data association problem as a multidimensional allocation problem. This model elevates fuzzy association matching to a well-defined combinatorial optimization problem. By defining a quantified fitness function and a solution objective, it lays a solid foundation for subsequent application of various optimization algorithms for efficient and global optimal solution search, thereby significantly improving the systematicness and scientific nature of association decisions.

[0123] In one possible implementation of step S103b above, an adaptive particle swarm optimization algorithm is used to globally optimize and solve the multidimensional allocation problem, determining the optimal matching relationship between each spatial target and the angle measurement observation data, including: Initialize the particle swarm, where the position vector of each particle in the swarm encodes an association hypothesis; Based on the fitness function, calculate the fitness value of the association hypothesis represented by each particle; During the iteration process, the velocity and position of each particle are dynamically updated based on the individual's historical best position and the group's global best position, and the algorithm control parameters are adjusted synchronously. When the preset convergence condition is met or the maximum preset number of iterations is reached, the correlation hypothesis represented by the globally optimal particle is determined as the optimal matching relationship.

[0124] In this embodiment, the implementation provides a specific optimization algorithm for solving the multidimensional allocation problem defined in the above embodiments.

[0125] Here, an adaptive particle swarm optimization algorithm is adopted. By simulating swarm intelligence, it performs an efficient and parallel global search in a huge solution space consisting of all possible related hypotheses to find a solution that optimizes the fitness function value.

[0126] Specifically, the execution process of this algorithm may include the following steps: Step a4: Initialize the particle swarm.

[0127] Here, a particle swarm of a certain size is randomly generated or initialized according to a specific strategy, and the position vector of each particle is... (in, For particle indexing, This is the vector dimension index, corresponding to the element number of the candidate observation set. Let be the number of iterations, initially. This represents a specific association hypothesis.

[0128] Each dimension of the position vector of each particle corresponds to a candidate observation set element, the value of which (a target ID, 0 representing clutter) indicates which space target the element is assigned to. Simultaneously, the velocity vector of each particle is initialized. Individual historical best position and the global optimal position of the entire population .

[0129] Step b4: Calculate the fitness value.

[0130] For each particle in the population, its fitness value is calculated according to the association hypothesis encoded by the particle's current position vector, using the formula (18) above. This is used to characterize the quality of the correlation hypothesis corresponding to the particle.

[0131] Step c4, iterative update.

[0132] The algorithm enters the iteration loop ( In each iteration, perform the following operations: Compare each particle's current fitness value with its individual historical best value. If it is better (i.e., ...), then the fitness value is higher. If it is larger, then update its individual historical best position. Simultaneously, identify the particle with the best fitness in the entire population and update the global optimal position of the population. .

[0133] The algorithm control parameters are adjusted synchronously according to the preset strategy, and the inertia weight is dynamically adjusted. and learning factors and .

[0134] The velocity and position of each particle are updated based on the following formulas (19) and (20): (19) (20) in, , It can be Random numbers within a certain range can be used to introduce randomness into the search. Velocity updates can give particles a tendency to fly towards their own historical best position and the group's historical best position, while position updates can change the correlation assumption.

[0135] Step d4: Repeat the fitness value calculation and iterative update operation in step c4 above, and record the correlation method between individual optimal and global optimal until the maximum number of iterations is reached or the convergence condition is met (e.g., the global optimal fitness value no longer improves significantly within several consecutive generations).

[0136] Finally, the association scheme represented by the globally optimal particle is output as the final association result of the observation target, that is, the association hypothesis that makes the fitness function value defined by formula (18) optimal.

[0137] The spatial target association and localization method and apparatus based on adaptive particle swarm optimization algorithm described in the above embodiments of this disclosure achieve efficient solutions to multidimensional allocation problems by transforming the abstract association problem into a specific particle swarm optimization process. The particle encoding method cleverly maps discrete combinatorial optimization problems into continuous optimization variables, enabling the algorithm to utilize swarm intelligence for parallel search in a vast solution space. The iterative update mechanism ensures that the algorithm continuously moves closer to better association hypotheses, thereby stably and reliably determining the optimal matching relationship in complex scenarios with dense targets and noise and clutter interference in observations.

[0138] In one possible implementation of step S103b above, the algorithm control parameters are adjusted synchronously, including: dynamically adjusting the inertia weight and the learning factor; The dynamic adjustment of the inertia weight is based on the following formula (21): (twenty one) As the current inertia weight, This is the preset maximum value for the inertia weight. This is the preset minimum value for the inertia weight; The dynamic adjustment of learning factors adopts the following strategy: the individual learning factor decreases with the iteration process, while the social learning factor increases with the iteration process.

[0139] In this embodiment, the implementation significantly improves the performance of the particle swarm optimization algorithm in solving the specific optimization problem of data association by dynamically adjusting the core control parameters. This adaptive mechanism aims to balance the algorithm's global exploration and local exploitation capabilities.

[0140] Specifically, the parameter adjustment strategy may include dynamically adjusting the inertia weight. and dynamically adjust learning factors and .

[0141] Among them, inertia weight The proportion of velocity a particle inherits from the previous generation can be determined to control the algorithm's exploration capabilities.

[0142] Preferably, inertial weight The dynamic adjustment can adopt a linear decreasing strategy, setting a larger value in the early stage of particle iteration. This makes the inertial weight A larger inertia weight allows particles to maintain a higher flight speed, which is beneficial for wide-area search across the entire solution space, thereby enhancing global exploration capabilities and preventing premature entrapment in local optima. In the later stages of particle iteration, the inertia weight... Gradually decrease to The particle speed is slowed down, which facilitates a finer search in the potential optimal solution region, thereby strengthening local development and improving the convergence accuracy and speed of the algorithm.

[0143] Furthermore, the learning factors individually control the particles to move towards their own historical best (individual learning factor). ) and group historical best (social learning factor) ( ) Step length of flight direction.

[0144] Among them, individual learning factors A strategy of decreasing the size as the iteration progresses can be adopted. In the early stages of iteration, a larger... This can encourage particles to conduct independent and diverse explorations; in the later stages of iteration, smaller particles... This can weaken the influence of individual experience and encourage particles to converge toward group consensus.

[0145] Social learning factors A strategy of increasing increments as the iteration progresses can be adopted. In the early stages of iteration, a smaller increment... This can avoid premature group convergence; in the later stages of iteration, larger... This can enhance the guiding role of group knowledge and accelerate convergence towards the global optimal solution.

[0146] The spatial target association and localization method and apparatus based on adaptive particle swarm optimization (PSO) algorithm described in the above embodiments of this disclosure, by introducing dynamically adjusted inertia weights and learning factors, endows the algorithm with adaptive search characteristics. This strategy effectively overcomes the shortcomings of the basic PSO algorithm, such as being prone to getting trapped in local optima and having slow convergence speed in the later stages of iteration. In data association application scenarios, this parameter adjustment strategy enables the algorithm to extensively explore various possible association combinations in the early stages and to quickly and accurately lock in the optimal association scheme in the later stages, thereby improving the convergence, stability, and reliability of the final result of the association and localization process as a whole.

[0147] In one possible implementation of step S104 above, the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time are updated according to the optimal matching relationship, including: Based on the optimal matching relationship, the correct angle measurement observation data used to update the orbital state of each space target is determined; Through the update phase of the Kalman filter algorithm, the predicted state is fused with the corresponding correct angle measurement observation data to calculate the updated posterior state estimate; Update the covariance matrix corresponding to the posterior state estimate.

[0148] In this embodiment, the implementation method constitutes a closed loop for state estimation in this disclosure. Based on the optimal matching result provided by the data association link, the predicted state is corrected using new observation data, thereby obtaining a state estimate that is closer to the true value and has less uncertainty.

[0149] Specifically, the state update process is as follows: Step a5: Determine the valid observation data.

[0150] Here, based on the optimal matching relationship output in step S103b, for each spatial target... Determine the correct correlation between the angle measurement observation data and those from at least two observation platforms. , )and( , These data serve as valid inputs for state updates.

[0151] Step b5: Perform Kalman filter update.

[0152] Here, for each space target Perform the following Kalman filter update calculation process: Step b51, Kalman gain calculation, as shown in the following formula (22).

[0153] (twenty two) in, Here is the Kalman gain matrix. The prior covariance matrix, For the observation matrix, To observe the noise covariance matrix.

[0154] Here, the Kalman gain determines the weights of the predicted state and the new observation data during the state update.

[0155] Step b52, the state vector of the space target relative to the observation platform The update is performed as shown in formula (23) below.

[0156] (twenty three) here, This is the difference between the actual observed value and the predicted observed value based on the forecast state.

[0157] Step b53, the covariance matrix is ​​updated as shown in formula (24) below.

[0158] (twenty four) here, The updated posterior covariance matrix typically has smaller values ​​than the prior covariance matrix. This is to characterize how the uncertainty of state estimation is reduced by fusing new observational information.

[0159] The space target association and localization method and apparatus based on the adaptive particle swarm optimization algorithm described in the above embodiments of this disclosure achieve continuous high-precision estimation of the orbital state of space targets by updating the results using Kalman filtering based on high-confidence association results. This step ultimately transforms the results of the preceding processes (prediction and association) into improved positioning accuracy, forming an automated recursive closed loop of "prediction-association-update". Since the observation data used for updating is correct data that has undergone rigorous screening through two-level association processing, the contamination of state estimation by erroneous associations is effectively avoided, thereby ensuring that the entire system can still output stable and reliable positioning results in complex environments.

[0160] In one possible implementation of the above embodiments, please refer to Figure 2 , Figure 2 This is a schematic diagram illustrating the iterative process results of the fitness of the spatial target association and localization method based on the adaptive particle swarm algorithm provided in this embodiment of the disclosure, as shown below. Figure 2 As shown: Figure 2 This demonstrates the evolution of the fitness value distribution of the entire particle population with the number of iterations during the algorithm optimization process. A higher fitness value (i.e., a higher numerical value, closer to zero) indicates a better quality solution and higher accuracy in localization.

[0161] In the early stages of iteration (e.g., from 0 to approximately 5 iterations), the algorithm exhibits remarkable optimization efficiency, with the fitness value rapidly increasing from an initial level of approximately -180 dB to approximately -65 dB. Figure 2 The result is characterized by an extremely steep upward curve. This verifies that the adaptive mechanism employed in this disclosure can quickly guide the population out of the poor-performing search region. During the iteration process (from 0 to 100 iterations), as the adaptive particle swarm optimization algorithm runs, it effectively guides the entire population to evolve towards a better direction. It can be clearly seen that the distribution representing the population solution shifts upward overall and its range narrows.

[0162] As the iteration process continues (e.g., from approximately 5 to 100 iterations), the fitness value continues to show a robust upward trend, eventually converging to an excellent level of approximately -40 dB. This complete convergence trajectory fully demonstrates that the adaptive particle swarm optimization algorithm of this disclosure not only possesses strong global exploration capabilities, avoiding getting trapped in local optima, but also has refined local exploitation capabilities, thereby ensuring the accuracy and reliability of the final association scheme and providing an efficient optimization method for solving the complex problem of spatial target association and localization.

[0163] In one possible implementation of the above embodiments, please refer to Figure 3 , Figure 3 This is a schematic diagram of the association results of the PSO algorithm at a single moment in the spatial target association and localization method based on the adaptive particle swarm algorithm provided in the embodiments of this disclosure, as shown in the figure. Figure 3 As shown: Figure 3 The three sub-graphs are used to visually compare and verify the processing effect of the PSO algorithm described above in this embodiment at a single time step.

[0164] Among them, such as Figure 3 The left subplot shows the initial data state before association. The "×" points in the figure represent multiple estimated relative positions in the VVLH coordinate system of the reference observation platform C1, calculated from the raw observation data of the two observation cameras (C1 and C2) through steps such as the projection vector angle method. These estimated positions are numerous and widely distributed, including both valid observations of real space targets and spurious points caused by observation errors, clutter, or incorrect matching. The "dots" in the figure represent the true relative positions of the space targets with respect to the reference observation platform C1 at that moment, serving as the baseline truth for evaluating the algorithm's performance.

[0165] like Figure 3 The middle subplot shows the actual correlation between the true location of the space target and the observed data, i.e., the ideal correlation result that should exist. This subplot clarifies which observed data (the "×" points in the left subplot) truly originate from which specific space target (the "circles" in the left subplot), providing a criterion for evaluating the algorithm output in the right subplot.

[0166] like Figure 3As shown in the right subfigure, the final association and localization results obtained after optimization using the PSO algorithm described above are presented. By modeling the data association problem as a multidimensional assignment problem and using a fitness function based on Mahalanobis distance for global optimization, this algorithm successfully achieves the following: precise pairing of observation data from different cameras belonging to the same spatial target; effective filtering out false observation points (clutter) that do not belong to any real target; and associating the correct, data-associated observation set with the corresponding spatial target predicted state vector.

[0167] A comparison of the left and right subgraphs clearly shows that the PSO algorithm described above can accurately recover the association structure that is highly consistent with the actual situation in the middle subgraph from the initial observation data full of ambiguity and clutter, and finally output only the pure and correct association results corresponding to the actual target positions (dots). This fully demonstrates the effectiveness and robustness of this disclosure in solving the data association problem in complex multi-target environments, and lays a reliable foundation for the accurate state update of the subsequent Kalman filter.

[0168] In one possible implementation of the above embodiments, please refer to Figure 4 , Figure 4 This is a schematic diagram of the localization result of the spatial target association and localization method based on the adaptive particle swarm algorithm provided in the embodiments of this disclosure, as shown below. Figure 4 As shown: Based on the correct correlation results obtained from the PSO algorithm described above, this disclosure employs a Kalman filter algorithm to continuously update and estimate the relative orbital state of space targets. Figure 4 The curve showing the change of the position estimation error of the space target relative to the reference observation platform over time is presented.

[0169] It can be observed that the algorithm converges rapidly in the initial filtering phase (within approximately 10 seconds). After 10 seconds, the filtering process enters a steady state, and the position estimation error is significantly reduced and stabilizes within 10 meters. This result demonstrates that by effectively combining the adaptive particle swarm correlation framework proposed in this disclosure with the Kalman filter localization algorithm, the system ultimately achieves meter-level high-precision localization of spatial targets. This localization result fully verifies the correctness and superior performance of the entire technical chain from data correlation to state estimation disclosed in this disclosure.

[0170] In one embodiment, a spatial target association and localization device 500 based on an adaptive particle swarm optimization algorithm is provided. This device 500 corresponds one-to-one with the spatial target association and localization method based on the adaptive particle swarm optimization algorithm described in the above embodiments. Figure 5 As shown, the spatial target association and localization device 500 based on the adaptive particle swarm algorithm includes: The observation module 501 is used to continuously acquire angle measurement observation data collected by at least two observation platforms at each observation time according to a preset sampling period; wherein, one of the at least two observation platforms is designated as the reference observation platform. The prediction module 502 is used to predict the predicted state of each space target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix based on the orbital state of each space target relative to the reference observation platform at the current observation time and according to the orbital dynamics model; wherein, the orbital state at the current observation time is determined by initialization setting or by updating the orbital state at the previous observation time. The association matching module 503 is used to match the angle measurement observation data at the next observation time with the forecast status through two-level association processing; The association matching module 503 is specifically used to pair observation data corresponding to the same spatial target based on the geometric consistency between angular observation data from different observation platforms to form a candidate observation set; The association matching module 503 is specifically used to model the matching relationship between the candidate observation set and the predicted state of each space target as a multidimensional allocation problem, and to use an adaptive particle swarm algorithm to perform global optimization and solution of the multidimensional allocation problem to determine the optimal matching relationship between each space target and the angle measurement observation data. The update module 504 is used to update the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time based on the optimal matching relationship.

[0171] In one embodiment, the observation module 501 is specifically used to simulate and generate the orbital states of each observation platform and each space target in the geocentric inertial coordinate system at the observation time by integrating the dynamic model based on the preset initial orbital state. Based on the orbital state, the relative position vector of each space target with respect to each observation platform in the camera coordinate system is calculated through coordinate transformation; Based on the relative position vector, the azimuth and elevation angles are calculated and used as angle measurement observation data.

[0172] In one embodiment, the coordinate transformation includes: converting the orbital state in the geocentric inertial coordinate system into relative position vectors in the VVLH coordinate system, the body coordinate system, and the camera coordinate system of the observation platform in sequence.

[0173] In one embodiment, the association matching module 503 is specifically used to convert angular measurement observation data from different observation platforms into unit relative direction vectors in the geocentric inertial coordinate system; Based on the position and unit relative direction vector of each observation platform in the geocentric inertial coordinate system, the line of sight of the space target originating from each observation platform is obtained by parameterization; Calculate the degree of convergence of lines of sight to space targets from different observation platforms; Based on the comparison results of the degree of intersection and the preset threshold, it is determined whether the angle measurement observation data of different observation platforms correspond to the same space target.

[0174] In one embodiment, the association matching module 503 is specifically used to take the spatial target line of sight of the reference observation platform as the first line of sight and the spatial target line of sight of any other observation platform as the second line of sight; Calculate the angles between the midpoint of the shortest line connecting the first and second lines of sight and the lines connecting them to the starting points of the first and second lines of sight, respectively. The starting point of the first line of sight is the reference observation platform, and the starting point of the second line of sight is any other observation platform. The angles connecting the midpoints to the starting points of the first line of sight are defined as the angle between the position vector of the space target relative to the reference observation platform in the geocentric inertial coordinate system and the unit relative direction vector of the reference observation platform. The angles connecting the midpoints to the starting points of the second line of sight are defined as the angle between the position vector of the space target relative to other observation platforms in the geocentric inertial coordinate system and the unit relative direction vector of other observation platforms. The average of the two included angles is determined as the included angle of the projection vector, which is used as a measure of the degree of intersection.

[0175] In one embodiment, the association matching module 503 is specifically used to assign each element in the candidate observation set to the predicted state of a space target, or to mark it as clutter that does not belong to any known space target; wherein, a complete assignment scheme is defined as an association hypothesis; Construct a fitness function to evaluate the merits of each association hypothesis; The objective of solving the multidimensional assignment problem is defined as: finding the association hypothesis that optimizes the fitness function value.

[0176] In one embodiment, the association matching module 503 is specifically used to initialize the particle swarm, wherein the position vector of each particle in the particle swarm encodes an association hypothesis; Based on the fitness function, calculate the fitness value of the association hypothesis represented by each particle; During the iteration process, the velocity and position of each particle are dynamically updated based on the individual's historical best position and the group's global best position, and the algorithm control parameters are adjusted synchronously. When the preset convergence condition is met or the maximum preset number of iterations is reached, the correlation hypothesis represented by the globally optimal particle is determined as the optimal matching relationship.

[0177] In one embodiment, the fitness function is constructed based on Mahalanobis distance to measure the difference between each element in the candidate observation set and the forecast state, and the calculation of the fitness function incorporates the covariance information of each forecast state.

[0178] In one embodiment, the association matching module 503 is specifically used to synchronously adjust the algorithm control parameters, including: dynamically adjusting the inertia weight and the learning factor; The dynamic adjustment of the inertia weight is based on the following formula:

[0179] As the current inertia weight, This is the preset maximum value for the inertia weight. This is the preset minimum value for the inertia weight; The dynamic adjustment of learning factors adopts the following strategy: the individual learning factor decreases with the iteration process, while the social learning factor increases with the iteration process.

[0180] In one embodiment, the update module 504 is specifically used to determine the correct angular measurement observation data for updating the orbital state of each space target based on the optimal matching relationship; Through the update phase of the Kalman filter algorithm, the predicted state is fused with the corresponding correct angle measurement observation data to calculate the updated posterior state estimate; Update the covariance matrix corresponding to the posterior state estimate.

[0181] It should be noted that the spatial target association and localization device based on the adaptive particle swarm optimization algorithm provided in the above embodiments is only illustrated by the division of the above-described program modules when implementing the corresponding spatial target association and localization method based on the adaptive particle swarm optimization algorithm. In practical applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the above system can be divided into different program modules to complete all or part of the processing described above. In addition, the system provided in the above embodiments and the corresponding Figure 1 The embodiments of the methods shown belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0182] This disclosure also provides an electronic device having the above-described features. Figure 5 The space target association and localization device shown is based on the adaptive particle swarm algorithm.

[0183] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure.

[0184] The following is a detailed reference. Figure 6This diagram illustrates a suitable structural schematic for implementing an electronic device according to embodiments of the present disclosure. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 601, which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) 602 or a program loaded from memory 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the electronic device. The processor 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0185] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0186] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a memory 608, or installed from a ROM 602. When the computer program is executed by the processor 601, it performs the functions defined in the network data stream hardware offloading method for heterogeneous descriptor unified processing of embodiments of this disclosure.

[0187] Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0188] This disclosure also provides a computer-readable storage medium in which the methods described in this disclosure can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the network data stream hardware offloading method for unified processing of heterogeneous descriptors shown in the above embodiments is implemented.

[0189] A portion of this disclosure can be applied to computer program products, such as computer program instructions, which, when executed by a computer, can invoke or provide methods and / or technical solutions according to this disclosure through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, and installation package files. Accordingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions; the computer compiling the instructions and then executing the corresponding compiled program; the computer reading and executing the instructions; or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0190] Although embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A spatial target association and localization method based on adaptive particle swarm optimization algorithm, characterized in that, The method includes: According to a preset sampling period, angle measurement observation data collected by at least two observation platforms at each observation time on at least one space target are continuously acquired; wherein, the at least two observation platforms include one observation platform designated as the reference. Based on the orbital state of each space target relative to the reference observation platform at the current observation time, the predicted state of each space target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix are predicted according to the orbital dynamics model; wherein, the orbital state at the current observation time is determined by initialization setting or by updating the orbital state at the previous observation time. Through a two-level correlation process, the angular measurement observation data at the next observation time is matched with the predicted state. The two-level correlation process includes: The first-level association processing, based on the geometric consistency between angular measurement observation data from different observation platforms, pairs observation data corresponding to the same spatial target to form a candidate observation set; The second-level association processing models the matching relationship between the candidate observation set and the predicted state of each space target as a multidimensional allocation problem, and uses an adaptive particle swarm optimization algorithm to globally optimize and solve the multidimensional allocation problem to determine the optimal matching relationship between each space target and the angle measurement observation data. Based on the optimal matching relationship, update the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time.

2. The method according to claim 1, characterized in that, The continuous acquisition of angular measurement data of at least one space target collected by at least two observation platforms at each observation time according to a preset sampling period is achieved through simulation, including: Based on the preset initial orbital state, the orbital state of each observation platform and each space target in the geocentric inertial coordinate system at the observation time is simulated and generated by integrating the dynamic model. Based on the orbital state, the relative position vector of each space target with respect to each observation platform in the camera coordinate system is calculated through coordinate transformation; Based on the relative position vector, the azimuth and elevation angles are calculated, and the azimuth and elevation angles are used as the angle measurement observation data.

3. The method according to claim 2, characterized in that, The coordinate transformation includes converting the orbital state in the geocentric inertial coordinate system into relative position vectors in the VVLH coordinate system, the body coordinate system, and the camera coordinate system of the observation platform.

4. The method according to claim 1, characterized in that, The pairing of observation data corresponding to the same space target based on the geometric consistency between angular measurement observation data from different observation platforms includes: Angle measurement data from different observation platforms were converted into unit relative direction vectors in the geocentric inertial coordinate system. Based on the position of each observation platform in the geocentric inertial coordinate system and the unit relative direction vector, the line of sight of the space target originating from each observation platform is parameterized. Calculate the degree of convergence of lines of sight to space targets from different observation platforms; Based on the comparison results between the degree of intersection and the preset threshold, it is determined whether the angle measurement observation data of the different observation platforms correspond to the same space target.

5. The method according to claim 4, characterized in that, The calculation of the degree of convergence of lines of sight to space targets from different observation platforms includes: The first line of sight is the spatial target line of sight of the reference observation platform, and the second line of sight is the spatial target line of sight of any other observation platform. Calculate the angles between the midpoint of the shortest line connecting the first line of sight and the second line of sight, and the angles between the midpoint of the line connecting the midpoint of the first line of sight and the midpoint of the second line of sight; wherein, the starting point of the first line of sight is the reference observation platform, and the starting point of the second line of sight is any other observation platform; the angle between the midpoint of the line connecting the midpoint of the first line of sight is defined as the angle between the position vector of the space target relative to the reference observation platform in the geocentric inertial coordinate system and the unit relative direction vector of the reference observation platform; the angle between the midpoint of the line connecting the midpoint of the second line of sight is defined as the angle between the position vector of the space target relative to the other observation platform in the geocentric inertial coordinate system and the unit relative direction vector of the other observation platform; The average of the two included angles is determined as the included angle of the projection vector, which is used as a measure of the degree of intersection.

6. The method according to claim 1, characterized in that, The matching relationship between the candidate observation set and the predicted states of each space target is modeled as a multidimensional allocation problem, including: Each element in the candidate observation set is assigned to the predicted state of a space target, or marked as clutter that does not belong to any known space target; wherein, a complete assignment scheme is defined as an association hypothesis. Construct a fitness function to evaluate the merits of each association hypothesis; The objective of solving the multidimensional assignment problem is defined as: finding the association hypothesis that optimizes the fitness function value.

7. The method according to claim 6, characterized in that, The adaptive particle swarm optimization algorithm is used to globally optimize and solve the multidimensional allocation problem, determining the optimal matching relationship between each spatial target and the angle measurement observation data, including: Initialize a particle swarm, wherein the position vector of each particle in the particle swarm encodes an association hypothesis; Based on the fitness function, calculate the fitness value of the association hypothesis represented by each particle; During the iteration process, the velocity and position of each particle are dynamically updated based on the individual's historical best position and the group's global best position, and the algorithm control parameters are adjusted synchronously. When the preset convergence condition is met or the maximum preset number of iterations is reached, the correlation hypothesis represented by the globally optimal particle is determined as the optimal matching relationship.

8. The method according to claim 6 or 7, characterized in that, The fitness function is constructed based on Mahalanobis distance and is used to measure the difference between each element in the candidate observation set and the forecast state. The calculation of the fitness function incorporates the covariance information of each forecast state.

9. The method according to claim 7, characterized in that, The control parameters of the synchronization adjustment algorithm include: dynamically adjusting the inertia weight and the learning factor; The dynamic adjustment of the inertia weight is based on the following formula: As the current inertia weight, This is the preset maximum value for the inertia weight. This is the preset minimum value for the inertia weight; The dynamic adjustment of learning factors adopts the following strategy: the individual learning factor decreases with the iteration process, while the social learning factor increases with the iteration process.

10. The method according to claim 1, characterized in that, The step of updating the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time based on the optimal matching relationship includes: Based on the optimal matching relationship, the correct angle measurement observation data used to update the orbital state of each space target is determined; Through the update phase of the Kalman filter algorithm, the predicted state is fused with the corresponding correct angle measurement observation data to calculate the updated posterior state estimate; Update the covariance matrix corresponding to the posterior state estimate.

11. A spatial target association and localization device based on adaptive particle swarm optimization algorithm, characterized in that, The device includes: An observation module is used to continuously acquire angle measurement observation data collected by at least two observation platforms at each observation time according to a preset sampling period; wherein, one of the at least two observation platforms is designated as a reference observation platform; The prediction module is used to predict the predicted state of each space target relative to the reference observation platform at the next observation time and the corresponding prior covariance matrix based on the orbital state of each space target relative to the reference observation platform at the current observation time and according to the orbital dynamics model; wherein, the orbital state at the current observation time is determined by initialization setting or by updating the orbital state at the previous observation time. The association matching module is used to match the angle measurement observation data at the next observation time with the predicted state through two-level association processing; The association matching module is specifically used to pair observation data corresponding to the same spatial target based on the geometric consistency between angular measurement observation data from different observation platforms to form a candidate observation set; The association matching module is specifically used to model the matching relationship between the candidate observation set and the predicted state of each space target as a multidimensional allocation problem, and to use an adaptive particle swarm optimization algorithm to globally optimize and solve the multidimensional allocation problem to determine the optimal matching relationship between each space target and the angle measurement observation data. The update module is used to update the orbital state and corresponding covariance matrix of each space target relative to the reference observation platform at the next observation time based on the optimal matching relationship.

12. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the spatial target association and localization method based on the adaptive particle swarm algorithm as described in any one of claims 1 to 10 when executing the computer program.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, it implements the steps of the spatial target association and localization method based on the adaptive particle swarm algorithm as described in any one of claims 1 to 10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the spatial target association and localization method based on the adaptive particle swarm algorithm as described in any one of claims 1 to 10.

Images