Method and system for classifying relative motion trails of spatial non-cooperative targets

Abstract

A method and a system for classifying a relative motion trajectory of a spatial non-cooperative target are disclosed, the normalized relative distance of a trajectory in an ROI region satisfying a detection distance is obtained through trajectory preprocessing, the trajectory is segmented according to the normalized relative distance, then the obtained sub-trajectories are classified and judged, and then the relative motion trajectory of the spatial non-cooperative target is classified according to the classification result of each sub-trajectory. And the classification of the whole relative movement track is completed. According to the method, the spatial non-cooperative target trajectory is identified and classified based on the solution of the relative motion equation, so that the requirement of an own spacecraft on situation awareness of the spatial non-cooperative target in a region of interest (ROI) in a complex spatial environment is met; and the perception of the space non-cooperative target on five motion situations of flying around, vortex, approaching, leaving and hovering of the spacecraft on the own side is completed.

Description

Technical Field

[0001] The present invention relates to a technology in the field of satellite control, in particular to a relative motion trajectory classification method and system for space non-cooperative targets. Background Art

[0002] The relative motion between the non-cooperative space target and the friendly spacecraft satisfies the constraints of orbital dynamics. Without considering orbital perturbations and atmospheric interference, assuming that the orbits of the non-cooperative space target and the friendly spacecraft are both circular orbits, the relative motion between the non-cooperative space target and the friendly spacecraft satisfies the solution of the orbital dynamics relative motion equation. The defects and shortcomings of the existing space target trajectory classification technology are: it is impossible to accurately classify the specific form of the relative motion between the non-cooperative space target and the friendly spacecraft, and it is impossible to perform adaptive segmentation and classification processing for complex relative motion trajectories. Summary of the invention

[0003] Aiming at the problem that the existing technology cannot meet the situational awareness and classification of the relative motion of one's own spacecraft to non-cooperative space targets, and the deficiency that the existing trajectory processing technology cannot determine the specific relative motion type of the tracking target or cannot perform segmented classification processing on the complex trajectory, the present invention proposes a relative motion trajectory classification method and system for non-cooperative space targets. The trajectories of non-cooperative space targets are identified and classified based on the solution of the relative motion equation, so as to meet the situational awareness of non-cooperative space targets in a region of interest (ROI) of one's own spacecraft in a complex space environment, and complete the perception of five motion situations of non-cooperative space targets to one's own spacecraft, namely, circling, mediation, approaching, moving away and hovering.

[0004] The present invention is achieved through the following technical solutions:

[0005] The present invention relates to a relative motion trajectory classification method for a space non-cooperative target, comprising:

[0006] Step 1: Trajectory preprocessing: Delineate the ROI area according to the detection distance (dr) input by the user and extract the trajectory within the ROI area, and calculate the relative distance rd of each point on the relative motion trajectory of the spatial non-cooperative target on the xy plane xy Then normalization is performed.

[0007] The ROI area range meets the following requirements:

[0008] The relative distance

[0009] The normalization process refers to: Where: rd nar (x, y) is rd xy(x,y) normalized relative distance, max(rd xy ) is rd xy The maximum value in (x,y), min(rd xy ) is rd xy Minimum value in (x,y).

[0010] Step 2: Track segmentation:

[0011] 2.1 Relative distance rd nar (x, y) is first-order derived to obtain the minimum point location set Ass min and the maximum point location set Ass max ;

[0012] 2.2 Judge the minimum and maximum points and determine the trajectory segmentation point set Ass seg , including:

[0013] 2.2.1 Traverse the minimum point, if the current minimum point Min current and the previous minimum point Min current-1 Satisfy|Min current -Min current-1 |>1, then the current minimum point Min current PositionIdxMin current Add to the track segmentation point collection Ass seg

[0014] 2.2.2 Traverse the maximum value point, if the current maximum value point Max current and the previous maximum point Max current-1 Meet|Max current -Max current-1 |>1, then the current maximum point Max current PositionIdxMax current Add to the track segmentation point collection Ass seg

[0015] 2.3 According to the trajectory segmentation point set Ass seg , divide the trajectory into several m sub-trajectories.

[0016] Step 3, trajectory classification: judge and classify the m sub-trajectories one by one, and complete the classification of the entire relative motion trajectory by synthesizing all classification results, including:

[0017] 3.1 Calculate the standard deviation std of the x-axis, y-axis, and z-axis of each sub-track x 、std y and std z , specifically: Where: meanaxis They are the average values ​​of the x-axis, y-axis, and z-axis, n is the number of trajectory points, and axis j The size of each point on the x, y and z axes;

[0018] 3.2 When std x 、std y and std z In the three axes, std x 、std y Both axes are less than 1, std x 、std z If the two axes are less than 1, the sub-trace is hovering;

[0019] 3.3 When std x 、std y and std y are greater than 1, the x-axis of the sub-trajectory is divided into the first half of the sub-trajectory x before =[x1, x2, ...x n / 2 ] and the second half of the sub-trajectory x after =[x n / 2 , x n / 2 +1, ...x n ], and similarly, the y-axis of the sub-trajectory is evenly divided into the first half of the sub-trajectory y before =[y1,y2,..y n / 2 ] and the second half of the sub-trajectory y after =[y n / 2 ,y n / 2 +1, ...y n ]; According to the formula: Where y0 is the y-axis coordinate of the orbiting point. When the trajectory is an orbiting point, y0=0; tho is the orbiting radius, and d is the semi-major axis. Calculate a set of semi-major axes d temp =[d1,d2,d3,...d n / 2 ], Calculate a set of orbit radius rho temp =[rho1, rho2, ....rho n / 2 ], every point

[0020] 3.4 Calculation of d temp and rho temp Standard deviation d and std rho , when std d and std rho are all less than 1, and std d Greater than std rho , then the sub-trajectory is a fly-by; when stdd and std rho are all less than 1, and std d Less than std rho , then the sub-trajectory is a mediation; when std d and std rho The difference is less than 1 and rd nar [1] greater than rd nar [n], then the sub-trajectory is close; when std d and std rho The difference is less than 1 and rd nar [1] is less than rd nar [n], then the sub-trajectory is far away;

[0021] 3.5 Output the classification results of each sub-trajectory to complete the classification of the entire relative motion trajectory. Technical Effects

[0022] The present invention decomposes the complete mixed trajectory into sub-trajectories with single orbital characteristics by finding the points where the motion trend of the relative motion trajectory changes rapidly. Compared with the prior art, the present invention is more suitable for the classification problem of on-orbit relative motion trajectories that conforms to engineering practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram of the system of the present invention;

[0024] Figure 2 is a flow chart of an embodiment;

[0025] Figure 3 It is a schematic diagram of the trajectory of the embodiment;

[0026] Figure 4 This is the test data result graph;

[0027] In the figure: (a) is the confusion matrix for noise-free trajectory classification; (b) is the confusion matrix for noisy trajectory classification. DETAILED DESCRIPTION

[0028] like Figure 1As shown, a relative motion trajectory classification system of a space non-cooperative target involved in this embodiment includes: a trajectory preprocessing module, a trajectory segmentation module and a trajectory classification module, wherein: the trajectory preprocessing module extracts the motion trajectory of the space non-cooperative target in the ROI area according to the range of the ROI input by the user, and performs dimension reduction and normalization processing on the trajectory; the trajectory segmentation module segments the preprocessed relative motion trajectory, considers the points where the motion trend of the space non-cooperative target changes to perform trajectory segmentation, extracts trajectory segmentation points, and divides the trajectory into multiple sub-trajectories according to the trajectory segmentation points; the trajectory classification module judges each sub-trajectory, judges the sub-trajectory in five scenarios of moving away, approaching, circling, hovering, and mediating, and finally outputs the classification result of the entire trajectory.

[0029] The trajectory preprocessing module includes: a user input information unit, a ROI area trajectory extraction unit and a trajectory relative distance calculation unit, wherein: the user input information unit sets the ROI area according to the information input by the user, outputs the x, y, z axis thresholds, and the x rang ,y rang , z rang ;ROI area trajectory extraction unit based on x rang ,y rang , z rang , cut the trajectory and keep the relative motion trajectory Tr in the ROI area roi ; The trajectory relative distance calculation unit calculates the relative motion trajectory Tr roi The relative distance rd of each point on the xy plane xy , according to the normalized formula, the relative distance rd xy Perform normalization to obtain the normalized distance rd nor .

[0030] The trajectory segmentation module includes: a first-order derivative unit, a segmentation point judgment unit and a trajectory segmentation unit, wherein: the first-order derivative unit is used to determine the normalized distance rd nor Perform first-order difference processing to obtain the minimum and maximum points. The minimum points form the minimum set Ass min , the maximum points form the maximum set Ass max ; The segmentation point judgment unit is based on Ass min and Ass max , perform extreme point judgment processing, and obtain the trajectory segmentation point set Ass seg , the trajectory segmentation unit is based on Ass seg , segment the trajectory and obtain m sub-trajectories, namely Tr roi ={STr1, STr2, STr3,..., STr m}.

[0031] The trajectory classification module includes: a sub-trajectory sorting unit, a sub-trajectory classification unit and a classification result integration unit, wherein: the sub-trajectory sorting unit performs sorting processing according to each sub-trajectory and outputs the sub-trajectory STr in sequence i (i≤m), sub-track classification unit for sub-track STr i Classification processing is performed to obtain sub-trajectory classification results. The classification result integration unit outputs the complete trajectory classification result based on the output results of all sub-trajectory classification units.

[0032] like Figure 2 As shown, this embodiment is based on the classification of relative motion trajectories of spatial non-cooperative targets of the above system, specifically including:

[0033] The first step is to perform trajectory preprocessing, specifically: based on the input detection distance dr: 180km, calculate the x-axis and y-axis ranges as [-120km, 120km] and the z-axis range as [-60km, 60km]. Extract the relative motion trajectory Tr of the spatial non-cooperative target in the ROI area roi , calculate the relative distance r of the trajectory on the xy plane dxy , for r dxy Normalize to get rd nar ;

[0034] The second step is to normalize the relative distance rd nar Taking the first-order derivative, rd nar Contains 7 minimum points and 6 maximum points, Ass min is [36, 67, 127, 177, 227, 277, 327], Ass max is [48, 99, 149, 199, 249, 299],

[0035] The third step is to obtain the segment point set Ass by gradually comparing the size of the minimum point and the maximum point. seg = [48, 327];

[0036] Step 4: Input segment point set Ass seg , Tr roi Divided into STr1, STr2, STr3;

[0037] The fifth step is to do a classification experiment on STr1. First, calculate std x 、std y and std z , found std x 、std y and std z All greater than 1;

[0038] Step 6: Calculate d of STr1temp and rho temp , calculate std respectively d and std rho , get std d and std rho All greater than 1;

[0039] Step 7: Calculate the rd of STr1 nar [1] and rd nar

[47] , rd nar [1]=1>rd nar

[47] = 0.1, the STr1 type is close;

[0040] Step 8: Perform classification experiments on STr2. First, calculate std x 、std y and std z , found std x 、std y and std z All greater than 1;

[0041] Step 9: Calculate the semi-major axis d of STr2 temp and the orbit radius rho temp and its corresponding standard deviation std d and std rho , get std d and std rho are all less than 1, and std d =53.21>std rho =20.18, and STr2 is obtained as fly-around;

[0042] Step 10: Calculate STr3 first by calculating std x 、std y and std z , found std x 、std y and std z All greater than 1;

[0043] Step 11: Calculate d of STr3 temp and rho temp , calculate std respectively d and std rho , get std d and std rho All greater than 1;

[0044] Step 12: Calculate the rd of STr3 nar [1] and rd nar

[478] , rd nar[1] = 0.1 <rd nar

[47] = 1, the STr3 type is far away;

[0045] Step 13: Integrate the classification results of the three sub-trajectories STr1, STr2 and STr3 and output Tr roi The trajectory classification results are as follows: Figure 3 shown.

[0046] A single orbital characteristic dataset is used as the test set to verify the accuracy of the orbital characteristic classification algorithm. The test samples are 6960 in total, divided into two categories: noisy and noise-free. Each category contains five different types of orbital data: hovering, approaching, circling, flying around, and moving away. The test results are as follows: Figure 3 and Figure 4 As shown in the figure, in order to intuitively show the classification effect, hovering, approaching, mediating, circling, and moving away are represented by 0, 1, 2, 3, 4, and 5 respectively. Figure 4 The calculation of various indicators of the algorithm is shown in Table 1:

[0047] Table 1 Evaluation table of track characteristics classification algorithm situation Precision Recall F1Score Accuracy Noise-free 96.39％ 96.15％ 96.12％ 96.43％ Noise 94.68％ 94.37％ 94.34％ 94.64％

[0048] Compared with the prior art, the present invention has a high accuracy rate in identifying the trajectory characteristics of space non-cooperative targets, has a certain adaptability, is more in line with the actual relative motion trajectory on orbit, has a low time complexity, and has the possibility of real-time processing; finally, it can have a certain degree of anti-interference against the false motion of space non-cooperative targets and extract the real motion trend of space non-cooperative targets.

[0049] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.

Claims

1. A relative motion trajectory classification method for non-cooperative targets in space, characterized in that: Through trajectory preprocessing, the normalized relative distance of the trajectory in the ROI area that meets the detection distance is obtained. After the trajectory is segmented according to the normalized relative distance, the obtained sub-trajectories are classified and judged. Then, according to the classification result of each sub-trajectory, the classification of the entire relative motion trajectory is completed.

2. The relative motion trajectory classification method of a space non-cooperative target according to claim 1 is characterized in that: The trajectory preprocessing is to define the ROI area according to the detection distance (dr) input by the user and extract the trajectory within the ROI area, and calculate the relative distance rd of each point on the relative motion trajectory of the spatial non-cooperative target on the xy plane. xy Then normalization is performed; The ROI area range meets the following requirements: The relative distance The normalization process refers to: Where: rd nar (x,y) is rd xy (x,y) normalized relative distance, max(rd xy ) is rd xy The maximum value in (x,y), min(rd xy ) is rd xy Minimum value in (x,y).

3. The relative motion trajectory classification method of a space non-cooperative target according to claim 1, characterized in that: The said segments specifically include: 2.1 Relative distance rd xy Perform first-order derivative to obtain the minimum point location set Ass min and the maximum point location set Ass max ; 2.2 Judge the minimum and maximum points and determine the trajectory segmentation point set Ass seg , specifically including: 2.2.1 Traverse the minimum point, if the current minimum point Min current and the previous minimum point Min current-1 Satisfaction|Min current -Min current-1 |>1, then the current minimum point Min current Add to the track segmentation point collection Ass seg ; 2.2.2 Traverse the maximum value point, if the current maximum value point Max current and the previous maximum point Max current-1 Meet|Max current -Max current-1 |>1, then the current maximum point Max current Add to the track segmentation point collection Ass seg ; 2.3 According to the trajectory segmentation point set Ass seg , divide the trajectory into several m sub-trajectories.

4. The relative motion trajectory classification method of a space non-cooperative target according to claim 1, characterized in that: The trajectory classification refers to: judging and classifying the m sub-trajectories divided into the trajectory one by one, and completing the classification of the entire relative motion trajectory by synthesizing all classification results, which specifically includes: 3.1 Calculate the standard deviation std of the x-axis, y-axis, and z-axis of each sub-track x 、std y and std z , specifically: Where: mean axis They are the average values ​​of the x-axis, y-axis, and z-axis, n is the number of trajectory points, and axis j The size of each point on the x, y and z axes; 3.2 When std x 、std y and std z In the three axes, std x 、std y Both axes are less than 1, std x 、std z If the two axes are less than 1, the sub-trace is hovering; 3.3 When std x 、std y and std y are greater than 1, the x-axis of the sub-trajectory is divided into the first half of the sub-trajectory x before =[x1,x2,...x n / 2] and the second half of the sub-trajectory x after =[x n / 2,x n / 2+1,…x n ], and similarly, the y-axis of the sub-trajectory is evenly divided into the first half of the sub-trajectory y before =[y1,y2,...y n / 2] and the second half of the sub-trajectory y after =[y n / 2,y n / 2+1,…y n ]; Calculate a set of semi-major axes d in sequence temp =[d1,d2,d3,…d n / 2], And a set of orbit radius 3.4 Calculation of d temp and rho temp The corresponding standard deviation std d and stdr ho , when std d and stdr ho are all less than 1, and std d Greater than std rho , then the sub-trajectory is a fly-by; when std d and std rho are all less than 1, and std d Less than std rho , then the sub-trajectory is a mediation; when std d and std rho The difference is less than 1 and rd nar [1] greater than rd nar [n], then the sub-trajectory is close; when std d and std rho The difference is less than 1 and rd nar [1] is less than rd nar [n], then the sub-trajectory is far away; 3.5 Output the classification results of each sub-trajectory to complete the classification of the entire relative motion trajectory.

5. A relative motion trajectory classification system for space non-cooperative targets implementing the method described in any one of claims 1 to 4, characterized in that: include: The trajectory preprocessing module, the trajectory segmentation module and the trajectory classification module, wherein: the trajectory preprocessing module extracts the motion trajectory of the spatial non-cooperative target in the ROI area according to the range of the ROI input by the user, and performs dimension reduction and normalization on the trajectory; the trajectory segmentation module segments the preprocessed relative motion trajectory, considers the points where the motion trend of the spatial non-cooperative target changes to perform trajectory segmentation, extracts trajectory segmentation points, and divides the trajectory into multiple sub-trajectories according to the trajectory segmentation points; the trajectory classification module judges each sub-trajectory, and judges the sub-trajectory in five scenarios: moving away, approaching, circling, hovering, and mediating, and finally outputs the classification result of the entire trajectory.

6. The relative motion trajectory classification system according to claim 5, characterized in that: The trajectory preprocessing module includes: a user input information unit, a ROI area trajectory extraction unit and a trajectory relative distance calculation unit, wherein: the user input information unit sets the ROI area according to the information input by the user, outputs the x, y, z axis thresholds, and the x rang ,y rang , z rang ;ROI area trajectory extraction unit based on x rang ,y rang , z rang , cut the trajectory and keep the relative motion trajectory Tr in the ROI area roi ; The trajectory relative distance calculation unit calculates the relative motion trajectory Tr roi The relative distance rd of each point on the xy plane xy , according to the normalized formula, the relative distance rd xy Perform normalization to obtain the normalized distance rd nor .

7. The relative motion trajectory classification system according to claim 5, characterized in that: The trajectory segmentation module includes: a first-order derivative unit, a segmentation point judgment unit and a trajectory segmentation unit, wherein: the first-order derivative unit is used to determine the normalized distance rd nor Perform first-order difference processing to obtain the minimum and maximum points. The minimum points form the minimum set Ass min , the maximum points form the maximum set Ass max ; The segmentation point judgment unit is based on Ass min and Ass max , perform extreme point judgment processing, and obtain the trajectory segmentation point set Ass seg , the trajectory segmentation unit is based on Ass seg , segment the trajectory and obtain m sub-trajectories, namely Tr roi = {STr1, STr2, STr3, ..., STr m }.

8. The relative motion trajectory classification system according to claim 5, characterized in that: The trajectory classification module includes: a sub-trajectory sorting unit, a sub-trajectory classification unit and a classification result integration unit, wherein: the sub-trajectory sorting unit performs sorting processing according to each sub-trajectory and outputs the sub-trajectory STr in sequence i (i≤m), sub-track classification unit for sub-track STr i Classification processing is performed to obtain sub-trajectory classification results. The classification result integration unit outputs the complete trajectory classification result based on the output results of all sub-trajectory classification units.

Images