On-orbit spacecraft skeleton structure identification method based on typical structure and device thereof
By using an on-orbit spacecraft skeleton structure identification method based on typical structures, key point labels and connection relationships are obtained, and identification is performed using a preset model. This solves the problem of poor spacecraft imaging quality, achieves high-precision morphological anomaly detection and real-time status monitoring, and improves space safety.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINESE PEOPLES LIBERATION ARMY UNIT 32035
- Filing Date
- 2023-02-17
- Publication Date
- 2026-06-09
Smart Images
Figure CN115953594B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of image processing technology, specifically relating to a method and apparatus for identifying the skeleton structure of an on-orbit spacecraft based on typical structures. Background Technology
[0002] With the rapid development of aerospace technology, an increasing number of space targets equipped with high-performance sensors are being sent into Earth orbit, providing crucial information for military reconnaissance, real-time communication, resource exploration, and other aerospace activities. As a large number of spacecraft are deployed to Earth orbit, space resources are continuously being compressed, leading to events such as space target disintegration and collisions. This creates a large amount of debris in the space environment, which not only occupies space resources but also poses a significant threat to the operation of spacecraft in orbit. Therefore, monitoring the motion status of spacecraft in orbit and conducting spacecraft behavior analysis, including attitude estimation, morphology estimation, structural dimension estimation, and anomaly detection, is of great significance for maintaining the safety and stability of space.
[0003] With the rapid development of deep learning, spacecraft condition monitoring and analysis based on neural networks are gradually emerging. In existing technologies, recognition is based on 3D models of spacecraft. The structures of each component of the spacecraft are complete, and there are basically no situations where some structures are unclear or cannot be labeled due to poor imaging quality or mutual occlusion of components. However, in reality, the imaging effect of optical / radar equipment is generally poor, and there will be adverse phenomena such as trailing and defocusing.
[0004] Therefore, it is urgent to improve the defects existing in the current technology. Summary of the Invention
[0005] To address the aforementioned problems in the existing technology, this invention provides a method and apparatus for identifying the skeleton structure of an on-orbit spacecraft based on typical structures. The technical problem to be solved by this invention is achieved through the following technical solution:
[0006] In a first aspect, the present invention provides a method for identifying the skeleton structure of an on-orbit spacecraft based on a typical structure, comprising:
[0007] Obtain raw images of on-orbit spacecraft with typical structures;
[0008] Based on the original images of a typical structured spacecraft in orbit, key point labels of the typical structured spacecraft in orbit are obtained; the key points are located on the main body, the first side solar panel, and the second side solar panel of the typical structured spacecraft in orbit.
[0009] Obtain the connection relationships between key points and construct the skeleton structure of a typical on-orbit spacecraft.
[0010] Based on the skeleton structure of a typical on-orbit spacecraft, a pre-set on-orbit spacecraft skeleton structure recognition model is trained to obtain the trained recognition model; wherein, the pre-set on-orbit spacecraft skeleton structure recognition model includes a feature extraction module, a skeleton structure prediction module, and a morphological anomaly detection module set in sequence.
[0011] The trained recognition model is used to identify images of spacecraft in orbit, obtain the skeleton structure of the spacecraft in orbit, and determine whether the shape of the spacecraft in orbit is abnormal based on the recognition results.
[0012] Secondly, the present invention also provides an on-orbit spacecraft skeleton structure identification device based on a typical structure, comprising:
[0013] The raw image acquisition module is used to acquire raw images of on-orbit spacecraft with typical structures;
[0014] The key point label acquisition module is used to acquire key point labels of a typical on-orbit spacecraft based on the original images of the typical on-orbit spacecraft. The key points are located on the main body, the first side solar panel, and the second side solar panel of the typical on-orbit spacecraft.
[0015] The skeleton structure acquisition module is used to acquire the connection relationships between key points and construct the skeleton structure of a typical on-orbit spacecraft.
[0016] The recognition model acquisition module is used to train a preset on-orbit spacecraft skeleton structure recognition model based on the skeleton structure of a typical on-orbit spacecraft and to acquire the trained recognition model. The preset on-orbit spacecraft skeleton structure recognition model includes a feature extraction module, a skeleton structure prediction module, and a morphological anomaly detection module set in sequence.
[0017] The recognition result acquisition module is used to identify images of on-orbit spacecraft using a trained recognition model, obtain the skeleton structure of the on-orbit spacecraft, and determine whether the shape of the on-orbit spacecraft is abnormal based on the recognition results.
[0018] The beneficial effects of this invention are:
[0019] This invention provides a method and apparatus for identifying the skeleton structure of an on-orbit spacecraft based on a typical structure. By acquiring key point labels of an on-orbit spacecraft with a typical structure and constructing the associations between these key points, the skeleton structure of the on-orbit spacecraft with the typical structure is expressed. This allows for training a pre-set on-orbit spacecraft skeleton structure identification model, obtaining the trained identification model, and using the trained model to identify the skeleton structure of the on-orbit spacecraft and determine whether the on-orbit spacecraft's shape is abnormal. The entire identification process is simple and effective, with fewer requirements on the on-orbit spacecraft's imaging status, making it more closely aligned with practical applications. Furthermore, this invention can identify anomalies in the on-orbit spacecraft's shape during the skeleton identification process, enabling real-time monitoring of the on-orbit spacecraft's status. This has significant practical implications for maintaining space safety, including real-time status monitoring, fault detection, and timely rescue of on-orbit spacecraft.
[0020] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0021] Figure 1 This is a flowchart of an on-orbit spacecraft skeleton structure identification method based on typical structures provided in an embodiment of the present invention;
[0022] Figure 2 This is a structural schematic diagram of a key point of an on-orbit spacecraft with a typical structure provided in an embodiment of the present invention;
[0023] Figure 3 This is a schematic diagram of a preset on-orbit spacecraft skeleton structure recognition model provided in an embodiment of the present invention;
[0024] Figure 4 This is a schematic diagram of a ResNet model provided in an embodiment of the present invention;
[0025] Figure 5 This is a schematic diagram of an on-orbit spacecraft skeleton structure identification device based on a typical structure provided in an embodiment of the present invention. Detailed Implementation
[0026] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.
[0027] Please see Figure 1 , Figure 1 This is a flowchart of an on-orbit spacecraft skeleton structure identification method based on typical structures provided by an embodiment of the present invention. The on-orbit spacecraft skeleton structure identification method based on typical structures provided by the present invention includes:
[0028] S101. Obtain raw images of on-orbit spacecraft with typical structures;
[0029] S102. Based on the original images of the typical structure of the on-orbit spacecraft, obtain the key point labels of the typical structure of the on-orbit spacecraft; wherein, the key points are located on the main body, the first side solar panel and the second side solar panel of the typical structure of the on-orbit spacecraft.
[0030] S103. Obtain the connection relationships between key points and construct the skeleton structure of a typical on-orbit spacecraft.
[0031] S104. Based on the skeleton structure of a typical on-orbit spacecraft, train a pre-set on-orbit spacecraft skeleton structure recognition model and obtain the trained recognition model; wherein, the pre-set on-orbit spacecraft skeleton structure recognition model includes a feature extraction module, a skeleton structure prediction module and a morphological anomaly detection module set in sequence.
[0032] S105. Use the trained recognition model to recognize the image of the spacecraft in orbit, obtain the skeleton structure of the spacecraft in orbit, and determine whether the shape of the spacecraft in orbit is abnormal based on the recognition results.
[0033] For details, please continue to see Figure 1 As shown in this embodiment, a method for identifying the skeleton structure of an on-orbit spacecraft based on a typical structure is proposed. This method acquires key point labels of on-orbit spacecraft with typical structures and constructs the relationships between these key points to express the skeleton structure of the on-orbit spacecraft with the typical structure. This is used to train a pre-set on-orbit spacecraft skeleton structure identification model, obtain the trained identification model, and then use the trained model to identify the skeleton structure of the on-orbit spacecraft and determine whether the spacecraft's shape is abnormal. The entire identification process is simple and effective, requires less attention to the imaging state of the on-orbit spacecraft, and is closer to practical applications. Furthermore, in this embodiment, during the process of identifying the skeleton of the on-orbit spacecraft, abnormalities in the spacecraft's shape can also be identified, enabling real-time monitoring of the spacecraft's state. This has significant practical implications for maintaining space safety, including real-time status monitoring, fault detection, and timely rescue of on-orbit spacecraft.
[0034] In an optional embodiment of the present invention, please refer to Figure 2 As shown, Figure 2 This is a structural schematic diagram of a key point of a typical on-orbit spacecraft provided in an embodiment of the present invention. The main body of the typical on-orbit spacecraft includes a first key point 0 and a second key point 1. The first key point 0 is located at the first end of the main body along the flight direction of the on-orbit spacecraft, and the second key point 1 is located at the second end of the main body along the flight direction of the on-orbit spacecraft.
[0035] The first side panel of a typical on-orbit spacecraft includes a third key point 2 and a fourth key point 3. The third key point 2 is located at the connection between the first side panel and the main body, and the fourth key point 3 is located at the midpoint of the outer end of the first side panel.
[0036] The second side solar panel of a typical on-orbit spacecraft includes a fifth key point 4 and a sixth key point 5. The fifth key point 4 is located at the connection between the second side solar panel and the main body, and the sixth key point 5 is located at the midpoint of the outer end of the second side solar panel.
[0037] For details, please continue to see Figure 2 As shown, in this embodiment, six key points are used to describe the typical structure of an on-orbit spacecraft. Two key points are located on the main body, two on the first side solar panel, and two on the second side solar panel. The first key point 0 and the second key point 1 are located at opposite ends along the flight direction of the on-orbit spacecraft. If the main body has a flat panel or parabolic antenna load, one of the key points is located at the centroid of the load. The third key point 2 is located at the connection between the first side solar panel and the main body. The fourth key point 3 is located at the midpoint of the outermost extension of the first side solar panel, perpendicular to the flight direction of the on-orbit spacecraft. The fifth key point 4 is located at the connection between the second side solar panel and the main body. The sixth key point 5 is located at the midpoint of the outermost extension of the second side solar panel, perpendicular to the flight direction of the on-orbit spacecraft. Thus, the skeleton of a typical on-orbit spacecraft is described using these six key points.
[0038] In an optional embodiment of the present invention, please continue to refer to Figure 2 As shown, the skeleton structure of a typical on-orbit spacecraft includes:
[0039] Connect the first key point 0 and the third key point 2 to form the first connecting line;
[0040] Connect the first key point 0 and the fifth key point 4 to form the second connection;
[0041] Connect the second key point 1 and the third key point 2 to form the third connecting line;
[0042] Connect the second key point 1 with the fifth key point 4 to form the fourth connection line;
[0043] Connect the third key point 2 with the fifth key point 4 to form the fifth connecting line;
[0044] Connect the third key point 2 with the fourth key point 3 to form the sixth line;
[0045] Connect the fifth key point 4 with the sixth key point 5 to form the seventh line;
[0046] In this diagram, the sixth line represents the relative position of the first side solar panel, the seventh line represents the relative position of the second side solar panel, the fifth line represents the relative position of the first and second side solar panels with respect to the main body, the first and third lines represent the relative position of the first side solar panel with respect to the main body and its position along the flight direction of the spacecraft in orbit, and the second and fourth lines represent the relative position of the second side solar panel with respect to the main body and its position along the flight direction of the spacecraft in orbit.
[0047] For details, please continue to see Figure 2 As shown in this embodiment, based on the key points of a typical structured on-orbit spacecraft, the key points are associated by connecting lines to represent the relative positions of each component of the typical structured on-orbit spacecraft, as well as the morphological information of each component.
[0048] It should be noted that, from the perspective of the two-dimensional image, the first, second, and fifth lines form an isosceles triangle, and the third, fourth, and fifth lines also form an isosceles triangle.
[0049] It should be noted that, as Figure 2 As shown, four forms are displayed respectively. Among them, (a) is a spacecraft skeleton structure model that only shows the first side solar panel, (b) is a spacecraft skeleton structure model that only shows the main body, (c) is a complete spacecraft skeleton structure model, and (d) is a spacecraft skeleton structure model that only shows the second side solar panel.
[0050] In an optional embodiment of the present invention, please refer to Figure 3 As shown, Figure 3 This is a schematic diagram of a preset on-orbit spacecraft skeleton structure recognition model provided in an embodiment of the present invention. The feature extraction module includes a ResNet model and a deconvolution model.
[0051] Specifically, in this embodiment, please continue to refer to Figure 3 As shown, the feature extraction module includes a ResNet model (residual network model) that can perform bottom-up feature extraction. It uses deeper layers of the neural network to extract semantic information from the image, obtains the distribution of key points in the skeleton structure, and then uses deconvolution to expand the feature map size and calculate the features of the target key points.
[0052] It should be noted that, please refer to Figure 4 As shown, Figure 4 This is a schematic diagram of a ResNet model provided in an embodiment of the present invention. The basic idea of the ResNet model is to introduce a "shortcut connection" that can skip one or more layers, such as... Figure 4 As shown. ResNet proposes two mapping methods: one is identity mapping, such as... Figure 4The image shows a "curved curve." Another type of residual mapping refers to the portion other than the "curved curve," so the final output is... Here, identity mapping refers to itself, i.e., in the formula. The residual mapping refers to the "difference", which is... Therefore, residual refers to Partially; if the network has already reached its optimal state, further deepening the network will push the residual mapping to 0, leaving only the identity mapping. Theoretically, the network will remain in an optimal state indefinitely, and its performance will not decrease as the depth increases.
[0053] In an optional embodiment of the present invention, please continue to refer to Figure 3 As shown, the skeleton structure prediction module includes three layers of convolutional neural network.
[0054] Specifically, in this embodiment, please continue to refer to Figure 3 As shown, the skeleton structure prediction module includes three convolutional neural network layers. Based on convolution operations, it directly performs regression prediction on the image features extracted by the feature extraction module to obtain the position coordinates of the key points.
[0055] It should be noted that a three-layer neural network can include an input layer, a hidden layer, and an output layer, and ultimately obtains regression predictions of key points through the three layers of the neural network.
[0056] In an optional embodiment of the present invention, please continue to refer to Figure 3 As shown, the morphological anomaly detection module includes a binary classification model.
[0057] Specifically, in this embodiment, please continue to refer to Figure 3 As shown, based on the regression results obtained by the skeleton structure prediction module, a binary classification model is used to determine whether the morphology of the spacecraft in orbit is abnormal. If the binary classification model outputs 1, it is an abnormal morphology of the spacecraft in orbit; if the binary classification model outputs 0, it is a normal morphology of the spacecraft in orbit. Optionally, an abnormal morphology of the spacecraft in orbit includes the failure of the first side solar panel and / or the second side solar panel to open normally.
[0058] It should be noted that the binary classification model provided in this embodiment can be the softmax function.
[0059] In an optional embodiment of the present invention, the method further includes: obtaining the mean squared error loss function of a pre-trained on-orbit spacecraft skeleton structure recognition model, the expression of which is:
[0060] ;
[0061] in, The number of feature points, , For the first The real labels of key points For the first Predicted labels for key points To predict the Euclidean distance between key points.
[0062] In an optional embodiment of the present invention, the identification of the morphology of an on-orbit spacecraft is achieved through the following process.
[0063] Construct training and testing sample sets of simulated images of typical on-orbit spacecraft structures.
[0064] Specifically, 100 arc segments of laser imaging simulation sequence data were acquired, with each arc segment having more than 30 effective images. The target TLE roots were used for imaging simulation. The image sample size in all sequences was 384×384. The key points were labeled in the order of the first key point, the second key point 1, the third key point 2, the fourth key point 3, the fifth key point 4, and the sixth key point 5 to obtain the labels of each key point.
[0065] Twenty image sequences were used as the test sample set, and eighty image sequences were used as the training sample set.
[0066] Based on the training sample set, a pre-set on-orbit spacecraft skeleton structure recognition model is trained to obtain the trained recognition model.
[0067] Specifically, the image sequences in the training sample set are input into the feature extraction module. The sequence images are first processed by the residual structure of the ResNet model to obtain a feature sequence of size 24×24×512. Then, two deconvolution operations are performed to obtain a key point feature sequence of size 96×96×512. The obtained key point sequence is input into the skeleton structure prediction module to obtain a regression result of size 384×384×6. The obtained regression result is used to use a binary classifier to determine whether the morphology of the on-orbit spacecraft is abnormal. The judgment criterion is that if there are missing key points in a single image sequence, that is, the number of identified key points is less than 6, then the morphology of the on-orbit spacecraft is judged to be abnormal; otherwise, the morphology of the on-orbit spacecraft is normal. Through the above process, the preset on-orbit spacecraft skeleton structure recognition model is trained with 150 iterations, and the trained recognition model is saved.
[0068] The trained recognition model is evaluated using a test sample set. The test sample set is input into the trained recognition model, and the output is the skeleton structure of the spacecraft in orbit, in order to determine the accuracy of the trained recognition model.
[0069] The trained recognition model is used to identify the skeleton of the spacecraft in orbit and to determine whether the spacecraft's shape is abnormal.
[0070] In this embodiment, the detection accuracy in the on-orbit spacecraft morphological anomaly detection task reached 97.8%, which has high application value. Based on the extraction of the spacecraft "skeleton" structure, it can also complete tasks such as spacecraft size estimation and attitude estimation, and has a strong development prospect in the field of aerospace identification.
[0071] Based on the same inventive concept, please refer to Figure 5 , Figure 5 This is a schematic diagram of an on-orbit spacecraft skeleton structure identification device based on a typical structure provided in an embodiment of the present invention. The present invention also provides an on-orbit spacecraft skeleton structure identification device based on a typical structure, applied to the on-orbit spacecraft skeleton structure identification method based on a typical structure provided in the above embodiments of the present invention. For specific embodiments of the identification method, please refer to the above description; further details will not be repeated here. The identification device includes:
[0072] The raw image acquisition module 201 is used to acquire raw images of on-orbit spacecraft with typical structures.
[0073] The key point label acquisition module 202 is used to acquire key point labels of the typical structure of the on-orbit spacecraft based on the original image of the typical structure of the on-orbit spacecraft; wherein the key points are located on the main body, the first side solar panel and the second side solar panel of the typical structure of the on-orbit spacecraft.
[0074] The skeleton structure acquisition module 203 is used to acquire the connection relationship between key points and construct the skeleton structure of a typical on-orbit spacecraft.
[0075] The recognition model acquisition module 204 is used to train a preset on-orbit spacecraft skeleton structure recognition model based on the skeleton structure of a typical on-orbit spacecraft and to acquire the trained recognition model; wherein, the preset on-orbit spacecraft skeleton structure recognition model includes a feature extraction module, a skeleton structure prediction module and a morphological anomaly detection module set in sequence.
[0076] The recognition result acquisition module 205 is used to recognize the image of the spacecraft in orbit using the trained recognition model, obtain the skeleton structure of the spacecraft in orbit, and determine whether the shape of the spacecraft in orbit is abnormal based on the recognition result.
[0077] For details, please continue to see Figure 5As shown in this embodiment, an on-orbit spacecraft skeleton structure recognition device based on a typical structure is provided. The device acquires key point labels of the on-orbit spacecraft with a typical structure through a key point label acquisition module 202, constructs the association between key points through a skeleton structure acquisition module 203 to express the skeleton structure of the on-orbit spacecraft with the typical structure, trains a preset on-orbit spacecraft skeleton structure recognition model through a recognition model acquisition module 204, acquires the trained recognition model, uses the trained recognition model to identify the skeleton structure of the on-orbit spacecraft, and determines whether the shape of the on-orbit spacecraft is abnormal. The entire recognition process is simple and effective, with fewer requirements on the imaging state of the on-orbit spacecraft, making it closer to practical applications. Furthermore, in this embodiment, during the process of recognizing the skeleton of the on-orbit spacecraft, abnormalities in the shape of the on-orbit spacecraft can also be identified, enabling real-time monitoring of the status of the on-orbit spacecraft. This has significant practical implications for real-time status monitoring, fault detection, and timely rescue of on-orbit spacecraft, thus ensuring space safety.
[0078] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations are intended to cover non-exclusive inclusion, such that an article or device comprising a list of elements includes not only those elements but also other elements not expressly listed. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device comprising said element. Terms such as "connected" or "linked" are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect. The orientations or positional relationships indicated by terms such as "upper," "lower," "left," and "right" are based on the orientations or positional relationships shown in the accompanying drawings and are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the invention.
[0079] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. In addition, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.
[0080] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.
Claims
1. A method for identifying the skeleton structure of an on-orbit spacecraft based on typical structures, characterized in that, include: Obtain raw images of on-orbit spacecraft with typical structures; Based on the original images of the typical structure of the on-orbit spacecraft, key point labels of the typical structure of the on-orbit spacecraft are obtained; wherein, the key points are located on the main body, the first side solar panel, and the second side solar panel of the typical structure of the on-orbit spacecraft; the main body of the typical structure of the on-orbit spacecraft includes a first key point and a second key point, the first key point being located at a first end of the main body along the flight direction of the on-orbit spacecraft, and the second key point being located at a second end of the main body along the flight direction of the on-orbit spacecraft; the first side solar panel of the typical structure of the on-orbit spacecraft includes a third key point and a fourth key point, the third key point being located at the connection between the first side solar panel and the main body, and the fourth key point being located at the midpoint of the outer end of the first side solar panel; the second side solar panel of the typical structure of the on-orbit spacecraft includes a fifth key point and a sixth key point, the fifth key point being located at the connection between the second side solar panel and the main body, and the sixth key point being located at the midpoint of the outer end of the second side solar panel; Obtain the connection relationships between the key points and construct the skeleton structure of the typical on-orbit spacecraft; constructing the skeleton structure of the typical on-orbit spacecraft includes: Connect the first key point with the third key point to form a first connecting line; Connect the first key point with the fifth key point to form a second line; Connect the second key point to the third key point to form a third connecting line; Connect the second key point with the fifth key point to form a fourth connecting line; Connect the third key point with the fifth key point to form the fifth connecting line; Connect the third key point with the fourth key point to form the sixth connecting line; Connect the fifth key point with the sixth key point to form the seventh line; Wherein, the sixth line represents the relative position of the first side solar panel, the seventh line represents the relative position of the second side solar panel, the fifth line represents the relative position of the first and second side solar panels with respect to the main body, the first and third lines represent the relative position of the first side solar panel with respect to the main body and its position along the flight direction of the spacecraft in orbit, and the second and fourth lines represent the relative position of the second side solar panel with respect to the main body and its position along the flight direction of the spacecraft in orbit. Based on the skeleton structure of the typical on-orbit spacecraft, a preset on-orbit spacecraft skeleton structure recognition model is trained to obtain the trained recognition model; wherein, the preset on-orbit spacecraft skeleton structure recognition model includes a feature extraction module, a skeleton structure prediction module, and a morphological anomaly detection module arranged sequentially. The trained recognition model is used to identify images of spacecraft in orbit, obtain the skeleton structure of the spacecraft in orbit, and determine whether the shape of the spacecraft in orbit is abnormal based on the recognition results.
2. The method for identifying the skeleton structure of an on-orbit spacecraft based on typical structures according to claim 1, characterized in that, The feature extraction module includes a ResNet model and a deconvolution model.
3. The method for identifying the skeleton structure of an on-orbit spacecraft based on typical structures according to claim 1, characterized in that, The skeleton structure prediction module includes three layers of convolutional neural network.
4. The method for identifying the skeleton structure of an on-orbit spacecraft based on a typical structure according to claim 1, characterized in that, The morphological anomaly detection module includes a binary classifier.
5. The method for identifying the skeleton structure of an on-orbit spacecraft based on a typical structure according to claim 1, characterized in that, Also includes: Obtain the mean squared error loss function of the pre-set in-orbit spacecraft skeleton structure recognition model, the expression of which is: ； in, The number of feature points, For the first One feature point, , For the first The real labels of key points For the first Predicted labels for key points To predict the Euclidean distance between key points.
6. A device for identifying the skeleton structure of an on-orbit spacecraft based on a typical structure, characterized in that, include: The raw image acquisition module is used to acquire raw images of on-orbit spacecraft with typical structures; A key point label acquisition module is used to acquire key point labels of an on-orbit spacecraft with a typical structure based on the original image of the spacecraft with the typical structure. The key points are located on the main body, the first side solar panel, and the second side solar panel of the on-orbit spacecraft with the typical structure. The main body of the on-orbit spacecraft with the typical structure includes a first key point and a second key point. The first key point is located at a first end of the main body along the flight direction of the on-orbit spacecraft, and the second key point is located at a second end of the main body along the flight direction of the on-orbit spacecraft. The first side solar panel of the on-orbit spacecraft with the typical structure includes a third key point and a fourth key point. The third key point is located at the connection between the first side solar panel and the main body, and the fourth key point is located at the midpoint of the outer end of the first side solar panel. The second side solar panel of the on-orbit spacecraft with the typical structure includes a fifth key point and a sixth key point. The fifth key point is located at the connection between the second side solar panel and the main body, and the sixth key point is located at the midpoint of the outer end of the second side solar panel. A skeleton structure acquisition module is used to acquire the connection relationships between the key points and construct the skeleton structure of the typical spacecraft in orbit; constructing the skeleton structure of the typical spacecraft in orbit includes: Connect the first key point with the third key point to form a first connecting line; Connect the first key point with the fifth key point to form a second line; Connect the second key point to the third key point to form a third connecting line; Connect the second key point with the fifth key point to form a fourth connecting line; Connect the third key point with the fifth key point to form the fifth connecting line; Connect the third key point with the fourth key point to form the sixth connecting line; Connect the fifth key point with the sixth key point to form the seventh line; Wherein, the sixth line represents the relative position of the first side solar panel, the seventh line represents the relative position of the second side solar panel, the fifth line represents the relative position of the first and second side solar panels with respect to the main body, the first and third lines represent the relative position of the first side solar panel with respect to the main body and its position along the flight direction of the spacecraft in orbit, and the second and fourth lines represent the relative position of the second side solar panel with respect to the main body and its position along the flight direction of the spacecraft in orbit. The recognition model acquisition module is used to train a preset on-orbit spacecraft skeleton structure recognition model based on the skeleton structure of the typical on-orbit spacecraft and acquire the trained recognition model; wherein, the preset on-orbit spacecraft skeleton structure recognition model includes a feature extraction module, a skeleton structure prediction module and a morphological anomaly detection module arranged in sequence. The recognition result acquisition module is used to identify images of on-orbit spacecraft using a trained recognition model, obtain the skeleton structure of the on-orbit spacecraft, and determine whether the shape of the on-orbit spacecraft is abnormal based on the recognition results.