An autonomous cataloging method and system based on space debris intelligent orbit determination

Abstract

The present application relates to the technical field of space debris cataloging and orbit determination, and particularly relates to an autonomous cataloging method and system based on intelligent orbit determination of space debris; comprising the following steps: collecting space debris astronomical data and preprocessing; using a graph attention network algorithm to generate a node feature vector corresponding to a graph node; calculating the attention weight between different graph nodes through an attention mechanism; solving the optimal orbit parameters of the space debris through a least square smoothing algorithm; clustering the same space debris under different observation batches to obtain complete orbit information of the space debris; calculating the similarity of the space debris with the known space debris right ascension and declination, and determining the relevance; calculating the consistency deviation of the space debris with the known space debris orbit parameters and luminosity value, and determining the consistency; calculating the cataloging confidence of the space debris, and cataloging the space debris; and completing autonomous cataloging and orbit determination of the space debris. The cataloging and orbit determination of the present application can adapt to the real-time management and control requirements of 100,000-level and above density space debris.

Description

Technical Field

[0001] This invention relates to the field of space debris cataloging and orbit determination technology, specifically to an autonomous cataloging method and system based on intelligent orbit determination of space debris. Background Technology

[0002] With the large-scale deployment of global low-Earth orbit satellite constellations and the increasing frequency of space exploration activities, the amount of space debris is growing exponentially, and the space environment is becoming increasingly complex. This is driving the field of space situational awareness into a new stage of collaborative management characterized by high density, high precision, and high timeliness. As a core supporting component of the space situational awareness system, space debris cataloging, in terms of its positioning accuracy, processing efficiency, and cataloging accuracy, directly determines the on-orbit operational safety of spacecraft, the timeliness of space collision early warning, and the level of refined management of space resources. It is a key technological foundation for ensuring the orderly conduct of space activities.

[0003] Existing space debris cataloging and orbit determination technologies have significant limitations: First, these technologies rely heavily on astronomical positioning data from a single observation source, and the core processing procedures require manual intervention to complete error correction and correlation matching operations. This process is cumbersome and prone to human error, resulting in severely insufficient processing timeliness. Second, the spatiotemporal alignment accuracy and correlation accuracy of multi-source astronomical positioning data are low, making it difficult to meet the real-time monitoring and dynamic management needs of space debris with densities of 100,000 or higher. This, in turn, restricts the construction of a space safety early warning system and the efficient utilization of space resources.

[0004] Currently, while some research on space debris cataloging based on artificial intelligence (AI) technology has been conducted internationally—for example, using the Transformer model to achieve the correlation and fusion of multi-source optical trajectories and overcoming the bottleneck of astronomical big data processing through on-board AI preprocessing technology—these technologies largely rely on specific observation systems and have not formed a complete integrated solution encompassing "multi-source data preprocessing - intelligent orbit determination - autonomous cataloging," thus failing to directly meet the actual application needs of my country's space-based and ground-based optical observation networks. Although my country has initially established a network of ground-based optical observation stations and space-based optical observation satellites in multiple locations, shortcomings remain in space debris cataloging and orbit determination technology, including large spatiotemporal alignment errors of multi-source data, low accuracy of debris correlation, and insufficient efficiency in orbit determination and cataloging coordination, resulting in a certain gap compared to international advanced technologies. Summary of the Invention

[0005] The purpose of this invention is to address the technical problem that existing space debris cataloging and orbit determination relies heavily on specific observation systems and has not formed a complete integrated solution of "multi-source data preprocessing - intelligent orbit determination - autonomous cataloging", which cannot directly adapt to the actual application needs of space-based and ground-based optical observation networks. Therefore, this invention provides an autonomous cataloging method and system based on intelligent orbit determination of space debris.

[0006] To solve the above-mentioned technical problems, the technical solution provided by the present invention is as follows: An autonomous cataloging method based on intelligent orbit determination of space debris includes the following steps: S1. Collect astronomical data of space debris through multiple batches of observations using space-based optical observation satellites and ground-based optical telescopes, and preprocess the astronomical data to obtain standardized astronomical data; S2. A graph attention network algorithm is used to abstract the right ascension and declination of the same space debris in different observation batches within standardized astronomical data into different graph nodes, and to generate node feature vectors corresponding to the graph nodes based on the right ascension and declination of different observation batches. : ; In the formula, , For the first Right ascension and declination after batch correction For observation time, Photometric value; S3. Calculate the attention weights between different graph nodes using the attention mechanism of the graph attention network algorithm. : ; In the formula, For the first Batch nodes and the first Attention weights between batch nodes, with values ​​ranging from [0,1]. It is a 16×8 weight matrix; For feature concatenation operations, the nodes are... With nodes The feature vectors are concatenated to form an 8-dimensional vector; For nodes The set of neighboring nodes is determined by the K-nearest neighbor algorithm, where K=10; LeakyReLU is the activation function with a slope of 0.01. After the attention weights are calculated, the graph attention network algorithm is used to calculate them according to the first... Batch node, first The correlation between data within a batch node is matched with corresponding correlation weights, and the result is obtained by weighted summation. Attention weights of batch nodes ; S4. Solve for the optimal orbital parameters of the space debris using the least squares smoothing algorithm with attention weights and by minimizing the sum of squared residuals between the observed and fitted orbital parameters: ; In the formula, For the number of observation batches, the number of observation batches for a single fragment is ≥5; To observe orbital parameters, including the right ascension of the ascending node. Perigeal argument , track semi-major axis eccentricity True near point angle ; This is a function for fitting orbital parameters; The orbital parameters to be optimized; S5. The K-means clustering algorithm is used to cluster the right ascension and declination sequences of the same space debris under different observation batches. After clustering is completed, the K-means clustering algorithm is used to fuse the right ascension and declination sequences of the same cluster to obtain the complete orbital information of the space debris; S6. Input the complete orbital information of the space debris into the matching threshold output model, and output the matching threshold between the space debris and the known space debris. ; Calculate the similarity of right ascension and declination between space debris and known space debris. : ; In the formula, , The right ascension and declination of the temporary fragments are in "″; , The right ascension and declination of the known fragments are given in "″". The matching threshold; When similarity This indicates a strong correlation between the right ascension and declination of the space debris and known space debris; S7. Calculate the consistency deviation between the orbital parameters and photometric values ​​of space debris and known space debris. : ; in, , These are the orbital parameters for the space debris and known space debris, respectively, in degrees. , These are the luminosity values ​​of space debris and known space debris, respectively, in cd / m². When consistency deviation This indicates that the space debris has a high degree of consistency with the orbital parameters and luminosity values ​​of known space debris; S8, Combining similarity Consistency Deviation Calculate the cataloging confidence of space debris : ; when At that time, the space fragment was confirmed to be a known space fragment and was classified into the known space fragment catalog. when If necessary, mark the space debris as suspicious space debris and re-execute step S1; when When this happens, the space fragment is identified as a new space fragment, and a new space fragment catalog is created. S9. Organize the optimal orbital parameters, cataloging confidence level, and space debris cataloging information of space debris, output a visual report and structured data, and complete the autonomous cataloging and orbit determination of space debris.

[0007] Furthermore, in step S6, the method for establishing the matching threshold output model is as follows: A1. Collect astronomical data of space debris through multiple batches of observations using space-based optical observation satellites and ground-based optical telescopes, and preprocess the astronomical data to obtain standardized astronomical data; A2. By using Min-Max normalization, the astronomical data in the standardized astronomical dataset after removing outliers are uniformly mapped to the [0,1] interval and used as the training dataset. A3. Use the DQN model as the initial matching threshold output model, and separate the target network of the initial matching threshold output model from the current network; An experience replay mechanism was used to train the model on the initial matching threshold using the training dataset, thereby learning the astronomical data characteristics of space debris at different orbital altitudes. Meanwhile, the mean squared error function is used as the loss training function for the initial matching threshold output model; The learning rate for model training was set to 0.001, the number of training iterations was 10,000, the batch size was 64, and the training termination condition was that the loss function value was ≤0.01. S4. After the training termination condition is met, the matching threshold output model is obtained.

[0008] Furthermore, in steps S1 and A1, the astronomical data of the space debris includes: right ascension (″), declination (″), right ascension variation trend Δα (″ / s), declination variation trend Δδ (″ / s), orbital elements Ω (°), and luminosity characteristic Lcorr (cd / m²).

[0009] Furthermore, in steps S1 and A1, the specific method for preprocessing the astronomical data of space debris is as follows: Radiometric correction and adaptive denoising algorithms are used to process the raw heterogeneous astronomical data collected by space-based optical observation satellites and ground-based optical telescopes to remove random noise and eliminate interference factors such as illumination changes, thereby improving the signal-to-noise ratio of the data. The accuracy of right ascension and declination in astronomical data is initially assessed. For data that does not meet the accuracy standard, the least squares fitting algorithm is used to perform fitting correction based on standard star catalog data. After the fitting and correction are completed, the time reference skeleton is used for the redundant time synchronization of the BeiDou-3 and Global Positioning System dual systems, and the space reference skeleton is used for the inertial reference frame. The astronomical data of space debris collected by space-based optical observation satellites and ground-based optical telescopes are synchronized with the time reference and aligned with the space reference. After time synchronization and spatial benchmark alignment, the isolated forest algorithm is used to detect outliers in the astronomical data. After removing outliers, a standardized astronomical dataset is obtained.

[0010] Furthermore, in step A2, the formula for Min-Max normalization is: ; In the formula, These are the normalized eigenvalues. These are the original eigenvalues. , These are the maximum and minimum values ​​of the feature, respectively.

[0011] Further, in step A3, the matching threshold output model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer, and an output layer connected in sequence; The input layer, the first hidden layer, the second hidden layer, the third hidden layer, and the output layer are respectively equipped with 4 neurons, 64 neurons, 32 neurons, 16 neurons, and 1 neuron; The activation function of the neuron is ReLU; The mean square error function is: ; In the formula, Current state ,action The Q value below; For the current network parameters, For target network parameters; This is the discount factor, with a value of 0.95; Indicates the state ,action Fine-grained reward value Next state The expectation; and: ; in, This represents the current matching accuracy, with a value range of [0,1]. The value represents the matching error, expressed in arcseconds, and ranges from [0, 0.1]. , These are the weighting coefficients, and , .

[0012] Further, in step S6, the matching threshold is... for: ; In the formula, The initial threshold is, and ; The adjustment parameter is used to match the threshold output model based on the matching degree between known space debris astronomical data and space debris astronomical data. The value range is [-0.5, 0.5].

[0013] Furthermore, in step S1, different space debris are classified into low orbit, medium orbit, and high orbit according to their orbital altitude; Among them, the low-orbit orbit has an altitude of less than 2,000 km, the medium-orbit orbit has an altitude between 2,000 km and 36,000 km, and the high-orbit orbit has an altitude of more than 36,000 km.

[0014] Furthermore, it also includes step S10: Simulation tools were used to construct a low-orbit high-density scenario with 100,000 fragments, a high-orbit long-distance scenario with an orbital altitude of ≥36,000 km, and a complex lighting interference scenario with light intensity fluctuation of ±20%. Space debris simulation data is generated using simulation tools, and the precise ephemeris data of the standard star is compared with the visualization report and structured data output in step S9. If discrepancies exist, the weight coefficients should be iteratively optimized using the gradient descent algorithm. , Matching threshold .

[0015] An autonomous cataloging system based on intelligent orbit determination of space debris includes a multi-source data preprocessing module, a cataloging and orbit determination module, a result output module, and a verification and optimization module connected in sequence. Space-based optical observation satellites and ground-based optical telescopes transmit observed space debris astronomical data to a multi-source data preprocessing module, which then preprocesses the data to obtain standardized astronomical data. The cataloging and orbit determination module sequentially solves for the optimal orbital parameters, performs clustering, obtains matching thresholds, determines similarity, determines consistency deviation, calculates cataloging confidence, and catalogs space debris within standardized astronomical data. The result output module organizes and outputs the optimal orbital parameters, cataloging confidence, and space debris cataloging information of the space debris into a visual report and structured data, thus completing the autonomous cataloging and orbit determination of space debris. Meanwhile, the verification and optimization module generates space debris simulation data through simulation tools, compares the precise ephemeris data of the standard star with the visualization report and structured data, and iteratively optimizes the data through the gradient descent algorithm.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The autonomous cataloging method and system based on intelligent orbit determination of space debris provided by this invention breaks through the limitations of traditional cataloging and orbit determination technology. It constructs an integrated solution for the entire process of "preprocessing of multi-source data (astronomical data of space debris collected by space-based optical observation satellites and ground-based optical telescopes) - intelligent orbit determination - autonomous cataloging", realizing the fully automated operation of the cataloging and orbit determination process, greatly reducing manual intervention, reducing human error, improving processing efficiency, and adapting to the real-time management and control needs of space debris with a density of 100,000 or higher.

[0017] 2. The autonomous cataloging method and system based on intelligent orbit determination of space debris provided by this invention adopts the DQN model as the matching threshold output model to achieve adaptive adjustment of the matching threshold, which helps to improve the accuracy of space debris association matching and solve the technical problem of inaccurate association of multi-source data.

[0018] 3. The autonomous cataloging method and system based on intelligent orbit determination of space debris provided by this invention relies on the dual-system redundant timing technology of Beidou-3 and Global Positioning System and the inertial coordinate system to realize the time synchronization and spatial reference alignment of multi-source astronomical data, improve the preprocessing quality of multi-source data, and provide reliable and stable data input for subsequent calculations.

[0019] 4. The autonomous cataloging method and system based on intelligent orbit determination of space debris provided by this invention helps to achieve accurate cataloging of space debris by using similarity judgment, consistency deviation judgment, and confidence calculation and evaluation. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the structure of an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] An autonomous cataloging method based on intelligent orbit determination of space debris includes the following steps: S1. Collect astronomical data of space debris through multiple batches of observations using space-based optical observation satellites and ground-based optical telescopes, and preprocess the astronomical data to obtain standardized astronomical data; Different space debris is classified into low-orbit, medium-orbit, and high-orbit categories according to its orbital altitude; Among them, the low-orbit orbit has an altitude of less than 2,000 km, the medium-orbit orbit has an altitude between 2,000 km and 36,000 km, and the high-orbit orbit has an altitude of more than 36,000 km. The astronomical data of the space debris include: right ascension (″), declination (″), right ascension variation trend Δα (″ / s), declination variation trend Δδ (″ / s), orbital elements Ω (°), and luminosity characteristic Lcorr (cd / m²). The specific methods for preprocessing astronomical data from space debris are as follows: Radiometric correction and adaptive denoising algorithms are used to process the raw heterogeneous astronomical data collected by space-based optical observation satellites and ground-based optical telescopes to remove random noise and eliminate interference factors such as illumination changes, thereby improving the signal-to-noise ratio of the data. The accuracy of right ascension and declination in astronomical data is initially assessed. For data that does not meet the accuracy standard, the least squares fitting algorithm is used to perform fitting correction based on standard star catalog data. After the fitting and correction are completed, the time reference skeleton is used for the redundant time synchronization of the BeiDou-3 and Global Positioning System dual systems, and the space reference skeleton is used for the inertial reference frame. The astronomical data of space debris collected by space-based optical observation satellites and ground-based optical telescopes are synchronized with the time reference and aligned with the space reference. After time synchronization and spatial benchmark alignment, the isolated forest algorithm is used to detect outliers in the astronomical data. After removing outliers, a standardized astronomical dataset is obtained. S2. A graph attention network algorithm is used to abstract the right ascension and declination of the same space debris in different observation batches within standardized astronomical data into different graph nodes, and to generate node feature vectors corresponding to the graph nodes based on the right ascension and declination of different observation batches. : ; In the formula, , For the first Right ascension and declination after batch correction For observation time, Photometric value; S3. Calculate the attention weights between different graph nodes using the attention mechanism of the graph attention network algorithm. : ; In the formula, For the first Batch nodes and the first Attention weights between batch nodes, with values ​​ranging from [0,1]. It is a 16×8 weight matrix; For feature concatenation operations, the nodes are... With nodes The feature vectors are concatenated to form an 8-dimensional vector; For nodes The set of neighboring nodes is determined by the K-nearest neighbor algorithm, where K=10; LeakyReLU is the activation function with a slope of 0.01. After the attention weights are calculated, the graph attention network algorithm is used to calculate them according to the first... Batch node, first The correlation between data within a batch node is matched with corresponding correlation weights, and the result is obtained by weighted summation. Attention weights of batch nodes ; S4. Solve for the optimal orbital parameters of the space debris using the least squares smoothing algorithm with attention weights and by minimizing the sum of squared residuals between the observed and fitted orbital parameters: ; In the formula, For the number of observation batches, the number of observation batches for a single fragment is ≥5; To observe orbital parameters, including the right ascension of the ascending node. Perigeal argument , track semi-major axis eccentricity True near point angle ; This is a function for fitting orbital parameters; The orbital parameters to be optimized; S5. The K-means clustering algorithm is used to cluster the right ascension and declination sequences of the same space debris under different observation batches. After clustering is completed, the K-means clustering algorithm is used to fuse the right ascension and declination sequences of the same cluster to obtain the complete orbital information of the space debris; S6. Input the complete orbital information of the space debris into the matching threshold output model, and output the matching threshold between the space debris and the known space debris. ; Calculate the similarity of right ascension and declination between space debris and known space debris. : ; In the formula, , The right ascension and declination of the temporary fragments are in "″; , The right ascension and declination of the known fragments are given in "″". The matching threshold; ; In the formula, The initial threshold is, and ; The adjustment parameter for the matching threshold output model is based on the matching degree between known space debris astronomical data and space debris astronomical data, and its value ranges from [-0.5, 0.5]. When similarity This indicates a strong correlation between the right ascension and declination of the space debris and known space debris; S7. Calculate the consistency deviation between the orbital parameters and photometric values ​​of space debris and known space debris. : ; in, , These are the orbital parameters for the space debris and known space debris, respectively, in degrees. , These are the luminosity values ​​of space debris and known space debris, respectively, in cd / m². When consistency deviation This indicates that the space debris has a high degree of consistency with the orbital parameters and luminosity values ​​of known space debris; S8, Combining similarity Consistency Deviation Calculate the cataloging confidence of space debris : ; when At that time, the space fragment was confirmed to be a known space fragment and was classified into the known space fragment catalog. when If necessary, mark the space debris as suspicious space debris and re-execute step S1; when When this happens, the space fragment is identified as a new space fragment, and a new space fragment catalog is created. S9. Organize the optimal orbital parameters, cataloging confidence level, and space debris cataloging information of space debris, output a visual report and structured data, and complete the autonomous cataloging and orbit determination of space debris.

[0023] S10. Simulation tools were used to construct a low-orbit high-density scenario with 100,000 fragments, a high-orbit long-distance scenario with an orbital altitude ≥36,000km, and a complex lighting interference scenario with light intensity fluctuations of ±20%. Space debris simulation data is generated using simulation tools, and the precise ephemeris data of the standard star is compared with the visualization report and structured data output in step S9. If discrepancies exist, iterative optimization using the gradient descent algorithm should be performed. , Matching threshold .

[0024] In this embodiment, the DQN model is used as the matching threshold output model. The method for establishing the matching threshold output model is as follows: A1. Collect astronomical data of space debris through multiple batches of observations using space-based optical observation satellites and ground-based optical telescopes, and preprocess the astronomical data to obtain standardized astronomical data; A2. By using Min-Max normalization, the astronomical data in the standardized astronomical dataset after removing outliers are uniformly mapped to the [0,1] interval and used as the training dataset. The formula for Min-Max normalization is: ; In the formula, These are the normalized eigenvalues. These are the original eigenvalues. , These are the maximum and minimum values ​​of the feature, respectively. A3. Use the DQN model as the initial matching threshold output model, and separate the target network of the initial matching threshold output model from the current network; An experience replay mechanism was used to train the model on the initial matching threshold using the training dataset, thereby learning the astronomical data characteristics of space debris at different orbital altitudes. Meanwhile, the mean squared error function is used as the loss training function for the initial matching threshold output model; The mean square error function is: ; In the formula, Current state ,action The Q value below; For the current network parameters, For target network parameters; This is the discount factor, with a value of 0.95; Indicates the state ,action Fine-grained reward value Next state The expectation; and: ; in, This represents the current matching accuracy, with a value range of [0,1]. The value represents the matching error, expressed in arcseconds, and ranges from [0, 0.1]. , These are the weighting coefficients, and , ; The learning rate for model training was set to 0.001, the number of training iterations was 10,000, the batch size was 64, and the training termination condition was that the loss function value was ≤0.01. S4. After the training termination condition is met, the matching threshold output model is obtained.

[0025] The matching threshold output model consists of an input layer, a first hidden layer, a second hidden layer, a third hidden layer, and an output layer connected in sequence. The input layer, the first hidden layer, the second hidden layer, the third hidden layer, and the output layer are respectively equipped with 4 neurons, 64 neurons, 32 neurons, 16 neurons, and 1 neuron; The activation function of the neuron is ReLU.

[0026] like Figure 1 As shown, the present invention also provides an autonomous cataloging system based on intelligent orbit determination of space debris, comprising a multi-source data preprocessing module, a cataloging and orbit determination module, a result output module, and a verification and optimization module connected in sequence. Space-based optical observation satellites and ground-based optical telescopes transmit observed space debris astronomical data (including right ascension data, declination data, orbital elements data, photometric data, etc.) to a multi-source data preprocessing module. The multi-source data preprocessing module performs preprocessing (including data denoising and radiometric correction, initial judgment and recalibration of right ascension and declination accuracy, and spatiotemporal synchronization and alignment of multi-source data) to obtain standardized astronomical data. The cataloging and orbit determination module sequentially solves for the optimal orbital parameters, performs clustering, obtains matching thresholds, determines similarity, determines consistency deviation, calculates cataloging confidence, and catalogs space debris within standardized astronomical data. The result output module organizes and outputs the optimal orbital parameters, cataloging confidence, and space debris cataloging information of the space debris into a visual report and structured data, thus completing the autonomous cataloging and orbit determination of space debris. Meanwhile, the verification and optimization module generates space debris simulation data through simulation tools, compares the precise ephemeris data of the standard star with the visualization report and structured data, and iteratively optimizes the data through the gradient descent algorithm.

[0027] The output module supports integration with external space situational awareness application platforms, providing high-quality data support for space traffic early warning, spacecraft on-orbit safety assurance, and decision support.

[0028] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present invention should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An autonomous cataloging method based on intelligent orbit determination of space debris, characterized in that, Includes the following steps: S1. Collect astronomical data of space debris through multiple batches of observations using space-based optical observation satellites and ground-based optical telescopes, and preprocess the astronomical data to obtain standardized astronomical data; S2. A graph attention network algorithm is used to abstract the right ascension and declination of the same space debris in different observation batches within standardized astronomical data into different graph nodes, and to generate node feature vectors corresponding to the graph nodes based on the right ascension and declination of different observation batches. : ; In the formula, , For the first Right ascension and declination after batch correction For observation time, Photometric value; S3. Calculate the attention weights between different graph nodes using the attention mechanism of the graph attention network algorithm. : ； In the formula, For the first Batch nodes and the first Attention weights between batch nodes, with values ​​ranging from [0,1]. It is a 16×8 weight matrix; For feature concatenation operations, the nodes are... With nodes The feature vectors are concatenated into an 8-dimensional vector; For nodes The set of neighboring nodes is determined by the K-nearest neighbor algorithm, where K=10; LeakyReLU is the activation function with a slope of 0.

01. After the attention weights are calculated, the graph attention network algorithm is used to calculate them according to the first... Batch node, first The correlation between data within a batch node is matched with corresponding correlation weights, and the result is obtained by weighted summation. Attention weights of batch nodes ; S4. Solve for the optimal orbital parameters of the space debris using the least squares smoothing algorithm with attention weights and by minimizing the sum of squared residuals between the observed and fitted orbital parameters: ； In the formula, For the number of observation batches, the number of observation batches for a single fragment is ≥5; To observe orbital parameters, including the right ascension of the ascending node. Perigeal argument , track semi-major axis eccentricity True near point angle ; This is a function for fitting orbital parameters; The orbital parameters to be optimized; S5. The K-means clustering algorithm is used to cluster the right ascension and declination sequences of the same space debris under different observation batches. After clustering is completed, the K-means clustering algorithm is used to fuse the right ascension and declination sequences of the same cluster to obtain the complete orbital information of the space debris; S6. Input the complete orbital information of the space debris into the matching threshold output model, and output the matching threshold between the space debris and the known space debris. ; Calculate the similarity of right ascension and declination between space debris and known space debris. : ； In the formula, , The right ascension and declination of the temporary fragments are in "″; , The right ascension and declination of the known fragments are given in "″". The matching threshold; When similarity This indicates a strong correlation between the right ascension and declination of the space debris and known space debris; S7. Calculate the consistency deviation between the orbital parameters and photometric values ​​of space debris and known space debris. : ； in, , These are the orbital parameters for the space debris and known space debris, respectively, in degrees. , These are the luminosity values ​​of space debris and known space debris, respectively, in cd / m². When consistency deviation This indicates that the space debris has a high degree of consistency with the orbital parameters and luminosity values ​​of known space debris; S8, Combining similarity Consistency Deviation Calculate the cataloging confidence of space debris : ； when At that time, the space fragment was confirmed to be a known space fragment and was classified into the known space fragment catalog. when If necessary, mark the space debris as suspicious space debris and re-execute step S1; when When this happens, the space fragment is identified as a new space fragment, and a new space fragment catalog is created. S9. Organize the optimal orbital parameters, cataloging confidence level, and space debris cataloging information of space debris, output a visual report and structured data, and complete the autonomous cataloging and orbit determination of space debris.

2. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 1, characterized in that, In step S6, the specific method for establishing the matching threshold output model is as follows: A1. Collect astronomical data of space debris through multiple batches of observations using space-based optical observation satellites and ground-based optical telescopes, and preprocess the astronomical data to obtain standardized astronomical data; A2. By using Min-Max normalization, the astronomical data in the standardized astronomical dataset after removing outliers are uniformly mapped to the [0,1] interval and used as the training dataset. A3. Use the DQN model as the initial matching threshold output model, and separate the target network of the initial matching threshold output model from the current network; An experience replay mechanism was used to train the model on the initial matching threshold using the training dataset, thereby learning the astronomical data characteristics of space debris at different orbital altitudes. Meanwhile, the mean squared error function is used as the loss training function for the initial matching threshold output model; The learning rate for model training was set to 0.001, the number of training iterations was 10,000, the batch size was 64, and the training termination condition was that the loss function value was ≤0.

01. A4. After the training termination condition is met, the matching threshold output model is obtained.

3. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 2, characterized in that, In steps S1 and A1, the astronomical data of the space debris includes: right ascension (″), declination (″), right ascension variation trend Δα (″ / s), declination variation trend Δδ (″ / s), orbital elements Ω (°), and luminosity characteristic Lcorr (cd / m²).

4. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 2, characterized in that, In steps S1 and A1, the specific method for preprocessing the astronomical data of space debris is as follows: Radiometric correction and adaptive denoising algorithms are used to process the raw heterogeneous astronomical data collected by space-based optical observation satellites and ground-based optical telescopes to remove random noise and interference factors such as illumination changes, thereby improving the signal-to-noise ratio of the data. The accuracy of right ascension and declination in astronomical data is initially assessed. For data that does not meet the accuracy standard, the least squares fitting algorithm is used to perform fitting correction based on standard star catalog data. After the fitting and correction are completed, the time reference skeleton is used as the redundant time synchronization of the BeiDou-3 and Global Positioning System dual systems, and the space reference skeleton is used as the inertial reference frame. The astronomical data of space debris collected by space-based optical observation satellites and ground-based optical telescopes are synchronized with the time reference and aligned with the space reference. After time synchronization and spatial benchmark alignment, the isolated forest algorithm is used to detect outliers in the astronomical data. After removing outliers, a standardized astronomical dataset is obtained.

5. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 2, characterized in that, In step A2, the formula for Min-Max normalization is: ； In the formula, These are the normalized eigenvalues. These are the original eigenvalues. , These are the maximum and minimum values ​​of the feature, respectively.

6. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 2, characterized in that, In step A3, the matching threshold output model includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer, and an output layer connected in sequence; The input layer, the first hidden layer, the second hidden layer, the third hidden layer, and the output layer are respectively equipped with 4 neurons, 64 neurons, 32 neurons, 16 neurons, and 1 neuron; The activation function of the neuron is ReLU; The mean square error function is: ； In the formula, Current state ,action The Q value below; For the current network parameters, For target network parameters; This is the discount factor, with a value of 0.95; Indicates the state ,action Fine-grained reward value Next state The expectation; and: ; in, This represents the current matching accuracy, with a value range of [0,1]. The value represents the matching error, expressed in arcseconds, and ranges from [0, 0.1]. , These are the weighting coefficients, and , .

7. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 1, characterized in that, In step S6, the matching threshold is... for: ； In the formula, The initial threshold is, and ; The adjustment parameter is used to match the threshold output model based on the matching degree between known space debris astronomical data and space debris astronomical data. The value range is [-0.5, 0.5].

8. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 1, characterized in that, In step S1, different space debris are classified into low orbit, medium orbit, and high orbit according to their orbital altitude; Among them, the low-orbit orbit has an altitude of less than 2,000 km, the medium-orbit orbit has an altitude between 2,000 km and 36,000 km, and the high-orbit orbit has an altitude of more than 36,000 km.

9. The autonomous cataloging method based on intelligent orbit determination of space debris according to claim 1, characterized in that, It also includes step S10: Simulation tools were used to construct a low-orbit high-density scenario with 100,000 fragments, a high-orbit long-distance scenario with an orbital altitude of ≥36,000 km, and a complex lighting interference scenario with light intensity fluctuation of ±20%. Space debris simulation data is generated using simulation tools, and the precise ephemeris data of the standard star is compared with the visualization report and structured data output in step S9. If discrepancies exist, the weight coefficients should be iteratively optimized using the gradient descent algorithm. , Matching threshold .

10. An autonomous cataloging system based on intelligent orbit determination of space debris, characterized in that, It includes a multi-source data preprocessing module, a cataloging and orbit determination module, a result output module, and a verification and optimization module connected in sequence; Space-based optical observation satellites and ground-based optical telescopes transmit observed space debris astronomical data to a multi-source data preprocessing module, which then preprocesses the data to obtain standardized astronomical data. The cataloging and orbit determination module sequentially solves for the optimal orbital parameters, performs clustering, obtains matching thresholds, determines similarity, determines consistency deviation, calculates cataloging confidence, and catalogs space debris within standardized astronomical data. The result output module organizes and outputs the optimal orbital parameters, cataloging confidence, and space debris cataloging information of the space debris into a visual report and structured data, thus completing the autonomous cataloging and orbit determination of space debris. Meanwhile, the verification and optimization module generates space debris simulation data through simulation tools, compares the precise ephemeris data of the standard star with the visualization report and structured data, and iteratively optimizes the data through the gradient descent algorithm.

Images