Non-cooperative target spacecraft intention judgment method and system

Abstract

The application discloses a non-cooperative target spacecraft intention research method and system, and the method comprises the following steps: S1, acquiring first data; S2, determining a target orbit according to the first data; S3, acquiring second data according to the target orbit; and inputting the second data into a preset intention research model to obtain an orbit behavior trend of the non-cooperative target spacecraft; wherein the orbit behavior trend comprises orbit keeping, approaching, avoiding, intercepting and flying with. The application can research the intention of the non-cooperative target spacecraft in advance, so as to improve the situation awareness capability of the spacecraft.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aerospace technology, in particular to a non-cooperative target spacecraft intention judgment method and system. BACKGROUND

[0002] In recent years, high-orbit space countermeasure equipment technology has developed rapidly, and high-orbit countermeasure has become an important field of military confrontation among countries. Among them, intention judgment is an important basis for building cognitive ability of spacecraft attack and defense confrontation, and intention judgment for non-cooperative targets is a key means to gain an advantage in future space confrontation.

[0003] In the prior art, most of the researches are to use the on-orbit motion state of the spacecraft to determine the orbit, without further considering the intention judgment of the non-cooperative target spacecraft.

[0004] Therefore, how to judge the intention of the non-cooperative target spacecraft in advance is one of the important problems to be solved in the field. SUMMARY

[0005] The purpose of the present application is to provide a non-cooperative target spacecraft intention judgment method and system to solve the problems in the prior art, which can judge the intention of the non-cooperative target spacecraft in advance to improve the situation awareness capability of the spacecraft.

[0006] The present application provides a non-cooperative target spacecraft intention judgment method, which comprises the following steps:

[0007] S1, obtaining first data;

[0008] S2, determining a target orbit according to the first data;

[0009] S3, obtaining second data according to the target orbit; and inputting the second data into a preset intention judgment model to obtain an orbit behavior trend of the non-cooperative target spacecraft;

[0010] The orbit behavior trend includes at least two of orbit keeping, approaching, evading, intercepting and flying with.

[0011] The non-cooperative target spacecraft intention judgment method as described above, wherein, optionally, the first data includes the relative position, relative speed, azimuth angle of the target spacecraft and the spacecraft, and the orbit parameters of the spacecraft.

[0012] The non-cooperative target spacecraft intention judgment method as described above, wherein, optionally, step S1 is to track the target spacecraft in real time and obtain the first data in real time.

[0013] The non-cooperative target spacecraft intention judgment method as described above, wherein, optionally, step S2 is to calculate the orbit of the target spacecraft as the target orbit through a high-precision integral algorithm and a high-precision dynamics model.

[0014] The non-cooperative target spacecraft intention judgment method as described above, wherein, optionally, in step S3,

[0015] S31, selecting a set number of second data from the target orbit;

[0016] S32, inputting the second data into a preset trained intention judgment model;

[0017] S33, obtaining the orbit behavior trend of the target spacecraft through the intention judgment model.

[0018] The non-cooperative target spacecraft intention judgment method as described above, wherein, optionally, the intention judgment model is a recurrent neural network model.

[0019] The non-cooperative target spacecraft intention judgment method as described above, wherein, optionally, the training process of the recurrent neural network model comprises,

[0020] S01, according to the orbit dynamics characteristics, the type of orbit behavior trend;

[0021] S02, according to the type of orbit behavior trend, making labels and inputting into an established recurrent neural network module;

[0022] S03, inputting the training sample into the neural network for calculation and outputting the result;

[0023] S04, comparing the training sample with the output result to calculate the model recognition rate;

[0024] S05, judging whether the model recognition rate can meet the requirements; if yes, a stable recurrent neural network model is obtained, and the training process is ended; if no, step S03 is entered.

[0025] The present application also proposes a non-cooperative target spacecraft intention judgment system for the method as described in any of the above, wherein,

[0026] The data acquisition unit is configured to acquire the first parameter;

[0027] The orbit determination module is in communication connection with the data acquisition unit and is configured to determine the target orbit of the target spacecraft according to the first parameter, and select a set number of second data from the target orbit;

[0028] The intention judgment model is in communication connection with the orbit determination module, configured to acquire the second data, perform target spacecraft intention judgment according to the second data, and obtain the orbit behavior trend of the target spacecraft.

[0029] The output module is in communication connection with the intention judgment model, configured to acquire the orbit behavior trend of the target spacecraft and output.

[0030] The non-cooperative target spacecraft intention judgment system as described above, wherein, optionally, further comprising a model training and updating unit;

[0031] The model training and updating unit comprises a behavior intention library, a sample input module, a weight updating module and a model verification module;

[0032] The behavior intention library is configured to set the orbit behavior trends of different types of target spacecrafts;

[0033] The sample input module is configured to acquire training samples corresponding to the types of the orbit behavior trends in the behavior intention library;

[0034] The behavior intention library, the model verification module and the sample input module are in communication connection with the weight updating module, and the weight updating module is configured to update the weights according to the types of the orbit behavior trends in the behavior intention library and the training samples;

[0035] The model verification module is in communication connection with the intention judgment model and the weight updating module, and the model verification module is configured to verify the intention judgment model during training, update the simulation training data set when the verification result does not meet the condition, and repeat the training until the verification result meets the condition.

[0036] The non-cooperative target spacecraft intention judgment system as described above, wherein, optionally, further comprising a data preprocessing module;

[0037] The data preprocessing module is in communication connection with the data acquisition unit and the orbit determination module, and the data preprocessing module is configured to acquire the first parameters, perform principal component analysis processing on the first parameters, and output the processed first parameters to the orbit determination module.

[0038] Compared with the prior art, the non-cooperative target spacecraft intention judgment method provided by the application determines the orbit of the non-cooperative target spacecraft through real-time monitoring of the non-cooperative target spacecraft and orbit determination of the non-cooperative target spacecraft according to the first data monitored, that is, the target orbit. A certain amount of second data is extracted from the determined target orbit, and the intention of the non-cooperative target spacecraft is judged by the intention judgment model using the second data, so as to obtain the intention of the non-cooperative target spacecraft, which can provide effective defense prediction and counteraction decision for our side, thereby improving the situation awareness capability of our spacecraft. The first data is used to determine the target orbit, and then the second data is extracted on the target orbit, which can greatly improve the data amount for intention judgment, and is beneficial to improve the accuracy of intention judgment. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 is the overall step flowchart of embodiment 1 of the application;

[0040] Figure 2 is the specific step flowchart of step S3 in embodiment 1 of the application;

[0041] Figure 3 is the step flowchart of recurrent neural network model training in embodiment 1 of the application;

[0042] Figure 4 is the structure block diagram of embodiment 2 of the application. DETAILED DESCRIPTION

[0043] The embodiments described below with reference to the drawings are exemplary and are only used to explain the application, and cannot be explained as a limitation of the application.

[0044] Embodiment 1

[0045] Please refer to Figures 1 to 3 The embodiment provides a non-cooperative target spacecraft intention judgment method, which comprises the following steps:

[0046] S1, acquiring first data; specifically, the first data includes the relative position, relative speed, azimuth angle of the target spacecraft and the spacecraft of our side, and the orbit parameters of the spacecraft of our side. The target spacecraft is tracked in real time, and the first data is acquired in real time. In the implementation, the observation quantity obtained by our side through space-based / ground-based means during the on-orbit period of the non-cooperative target spacecraft is the position, speed, azimuth angle and other state parameters of the non-cooperative target high-orbit spacecraft relative to the spacecraft of our side. In the specific implementation, the orbit parameters of the spacecraft of our side include the position of our side, the speed of our side and the running orbit of our side.

[0047] S2, determining the target orbit according to the first data; that is, determining the orbit of the target spacecraft according to the first data. The method of determining the target orbit through the first data is: combining the orbit dynamics to preprocess the observation, and calculating to determine the orbit where the non-cooperative target high-orbit spacecraft is, that is, the target orbit. Since the target spacecraft can maneuver and change orbit at any time, it is necessary to track the target spacecraft in real time and determine the orbit, and then realize the prediction of the spacecraft state at future time. Specifically, through high-precision integration algorithm and high-precision dynamics model, the accurate description and prediction of the satellite orbit are realized to obtain the accurate target orbit. It should be pointed out that since the target spacecraft can maneuver and change orbit at any time, the target orbit determined at different times can be different.

[0048] S3, obtaining the second data according to the target orbit; and inputting the second data into the preset intent judgment model to obtain the orbit behavior trend of the non-cooperative target spacecraft. Specifically, please refer to Figure 2 , this step includes the following specific steps:

[0049] S31, selecting a set number of second data from the target orbit; that is, determining the target orbit and obtaining the second data according to the target orbit, that is, predicting the state of the non-cooperative target spacecraft at future time. The obtained second data is used for intent judgment. Specifically, the second data contains data at each place on the target orbit and the corresponding time. Further, the second data can contain only data in the future period of time, or contain both data in the past period of time and data in the future period of time. Compared with only containing data in the future period of time, containing both data in the past period of time and data in the future period of time can be beneficial to improve the accuracy of intent judgment. In specific implementation, the data amount of the second data is greater than the set value to ensure the accuracy of intent judgment. Before inputting the second data into the intent judgment model, the second data is subjected to principal component analysis. S32, inputting the second data into the preset trained intent judgment model. That is, inputting the second data subjected to principal component analysis into the preset trained intent judgment model. S33, obtaining the orbit behavior trend of the target spacecraft through the intent judgment model calculation.

[0050] In specific implementation, the orbit behavior trend includes at least two of orbit keeping, approaching, evading, intercepting and flying with. In one implementation, the orbit behavior trend can be divided into 5 categories of orbit keeping, approaching, evading, intercepting and flying with according to the tactical purpose. In application, according to actual needs, the categories of the orbit behavior trend can be increased or reduced.

[0051] Through the above process, the non-cooperative target spacecraft is monitored in real time, and the non-cooperative target spacecraft is orbit-determined according to the first data monitored, so as to determine the orbit, i.e. the target orbit. A certain amount of second data is extracted from the determined target orbit, and the intention of the non-cooperative target spacecraft is judged by the intention judgment model using the second data, so as to obtain the intention of the non-cooperative target spacecraft. This can provide effective defense prediction and counteraction action decision for our side, thereby improving the situation awareness capability of our spacecraft. The first data is used to determine the target orbit, and then the second data is extracted on the target orbit, which can greatly improve the amount of data used for intention judgment, and is beneficial to improve the accuracy of intention judgment.

[0052] The intention judgment model is a recurrent neural network model. In specific implementation, the training of the recurrent neural network model should be completed in advance, i.e. before step S1.

[0053] Please refer to Figure 3 The training process of the recurrent neural network model includes the following specific steps:

[0054] S01, according to the orbit dynamics characteristics, the category of the orbit behavior trend. In this step, the behavior intention library is constructed for different scenes, so as to label the training samples, i.e. set corresponding data labels, so as to classify the labels for different scenes, and on this basis, learn the orbit training samples to form a recurrent neural network. The learning process is as shown in S03-S05.

[0055] S02, according to the category of the orbit behavior trend, make labels and input to the established recurrent neural network module; specifically, the category of the behavior trend is orbit keeping, approaching, evading, intercepting and flying with.

[0056] S03, input the training sample into the neural network for calculation, and output the result;

[0057] S04, compare the training sample with the output result to calculate the model recognition rate;

[0058] S05, judge whether the model recognition rate can meet the requirements; if yes, a stable recurrent neural network model is obtained, and the training process is ended; if not, step S03 is entered.

[0059] That is, through the recurrent neural network model, the prediction result is calculated and the memory is passed, a large number of training samples are input, the output result is compared with the data label to calculate the data error, and the model recognition rate is repeatedly trained until a stable model is generated to meet the requirements, and the expected goal is achieved. The stable model referred to here means that the model recognition rate can meet the requirements. Specifically, the condition that the model recognition rate meets the requirements can be, for example, that the accuracy of intent judgment is greater than a certain value, such as 95%.

[0060] Embodiment 2

[0061] This embodiment provides a non-cooperative target spacecraft intent judgment system for the method of embodiment 1, and the same parts will not be repeated. Only the differences will be described below.

[0062] Please refer to Figure 4 The non-cooperative target spacecraft intent judgment system proposed in this embodiment includes a data acquisition unit, an orbit determination module, an intent judgment model, and an output module.

[0063] The data acquisition unit is used to acquire the first parameter; in specific implementation, the first data includes the relative position, relative velocity, azimuth angle of the target spacecraft and our spacecraft, and the orbit parameters of our spacecraft.

[0064] The orbit determination module is in communication connection with the data acquisition unit, and is used to determine the target orbit of the target spacecraft according to the first parameter, and select a set number of second data from the target orbit.

[0065] The intent judgment model is in communication connection with the orbit determination module, and is used to acquire the second data, perform target spacecraft intent judgment according to the second data, and obtain the orbit behavior trend of the target spacecraft. The intent judgment model can be a recurrent neural network model as disclosed in embodiment 1.

[0066] The output module is in communication connection with the intent judgment model, and is used to acquire the orbit behavior trend of the target spacecraft and output. Specifically, the output module can be a human-computer interface for outputting the type of orbit behavior trend, which can be a display, an audible and light alarm, etc.

[0067] In specific implementation, a model training and updating unit is further included. The model training and updating unit is used to train the recurrent neural network model through training samples before application to obtain a stable model.

[0068] The model training and updating unit includes a behavior intent library, a sample input module, a weight updating module, and a model verification module.

[0069] The behavior intent library is used to set the orbit behavior trend of different types of target spacecraft. Specifically, the orbit behavior trend includes orbit keeping, approaching, evading, intercepting, and flying with.

[0070] The sample input module is configured to obtain training samples corresponding to the categories of the track behavior trends in the behavior intention library.

[0071] The behavior intention library, the model verification module and the sample input module are in communication connection with the weight updating module, and the weight updating module is configured to update the weights according to the categories of the track behavior trends in the behavior intention library and the training samples.

[0072] The model verification module is in communication connection with the intention judgment model and the weight updating module, and the model verification module is configured to verify the intention judgment model during training, and generate a weight adjustment instruction for adjusting the weights when the verification result does not meet the condition, so that the weight updating model updates the weights according to the weight adjustment instruction, so as to repeat the training until the verification result meets the condition.

[0073] In a specific implementation, in order to improve the accuracy of track determination, a data preprocessing module is further included. The data preprocessing module is in communication connection with the data acquisition unit and the orbit determination module, and the data preprocessing module is configured to obtain the first parameters, perform principal component analysis on the first parameters, and output the processed first parameters to the orbit determination module. Specifically, by using a high-precision integral algorithm and a high-precision dynamics model, the satellite orbit can be accurately described and predicted to obtain an accurate target orbit. It should be noted that the target spacecraft can be maneuvered to change the orbit at any time, and the target orbit determined at different times can be different.

[0074] The recurrent neural network obtained by data training is subjected to error calculation and reverse gradient propagation, and finally a stable model is obtained, which can realize intention judgment. From the perspective of space safety, the situation is comprehensively evaluated in combination with the information such as the safe distance of the spacecraft in orbit, so as to ensure the safety of the spacecraft in orbit.

[0075] The above embodiments according to the drawings illustrate the structure, features and effects of the present application. The above description is only the preferred embodiment of the present application, but the present application is not limited by the drawings. Any changes or modifications made in accordance with the concept of the present application, or equivalent embodiments with equivalent changes, shall be within the scope of the present application.

Claims

1. A method for determining the intent of a non-cooperative target spacecraft, characterized in that: The method comprises the following steps: S1, obtaining first data; S2, determining a target orbit according to the first data; S3, obtaining second data according to the target orbit; and inputting the second data into a preset intention judgment model to obtain an orbit behavior trend of the non-cooperative target spacecraft; wherein the orbit behavior trend comprises at least two of orbit keeping, approaching, evading, intercepting and flying with; the first data comprises relative positions, relative speeds, azimuth angles of the target spacecraft and the spacecraft of our side, and orbit parameters of the spacecraft of our side; and the second data comprises data at each place on the target orbit and corresponding time points; step S1 is to track the target spacecraft in real time and obtain the first data in real time; step S2 is to calculate the orbit of the target spacecraft as the target orbit through a high-precision integral algorithm and a high-precision dynamics model; in step S3, S31, selecting a set number of second data from the target orbit; S32, inputting the second data into a preset trained intention judgment model; S33, obtaining the orbit behavior trend of the target spacecraft through the intention judgment model; the intention judgment model is a recurrent neural network model.

2. The method of claim 1, wherein: The training process of the recurrent neural network model comprises S01, according to the orbit dynamics characteristics, the types of the orbit behavior trend; S02, according to the types of the orbit behavior trend, making labels and inputting them into the established recurrent neural network module; S03, inputting the training samples into the neural network for calculation and outputting the results; S04, comparing the training samples with the output results to calculate the model recognition rate; S05, judging whether the model recognition rate meets the requirements; if yes, a stable recurrent neural network model is obtained, and the training process is ended; if not, step S03 is entered.

3. A non-cooperative target spacecraft intent inference system for use in a method as claimed in claim 1 or 2, characterized in that: comprises a data acquisition unit for acquiring first parameters; an orbit determination module in communication connection with the data acquisition unit, for determining a target orbit of a target spacecraft according to the first parameters, and selecting a set number of second data from the target orbit; an intention judgment model in communication connection with the orbit determination module, for obtaining the second data, performing intention judgment of the target spacecraft according to the second data, and obtaining an orbit behavior trend of the target spacecraft; an output module in communication connection with the intention judgment model, for obtaining the orbit behavior trend of the target spacecraft and outputting.

4. The non-cooperative target spacecraft intent determination system of claim 3, wherein: It further comprises a model training and updating unit; the model training and updating unit comprises a behavior intention library, a sample input module, a weight updating module and a model verification module; the behavior intention library is used to set different types of orbit behavior trends of target spacecrafts; the sample input module is used to obtain training samples corresponding to the types of the orbit behavior trends in the behavior intention library; the behavior intention library, the model verification module and the sample input module are all in communication connection with the weight updating module, and the weight updating module is used to update weights according to the types of the orbit behavior trends in the behavior intention library and the training samples; The model verification module is in communication connection with the intention judgment model and the weight updating module respectively, and is configured to verify the intention judgment model during training, and update the simulation training data set when the verification result does not meet the condition, and repeat the training until the verification result meets the condition.

5. The non-cooperative target spacecraft intent determination system of claim 3, wherein: The data preprocessing module is further included; The data preprocessing module is in communication connection with the data acquisition unit and the orbit determination module, and is configured to acquire the first parameter, perform principal component analysis on the first parameter, and output the processed first parameter to the orbit determination module.

Images