A space-based passive cooperative multi-target positioning method in complex space environment

Abstract

The present invention discloses a space-based passive collaborative multi-target positioning method in a complex space environment, including S1, establishing a measurement clutter model and a missed detection model considering the actual space observation environment, and constructing a spacecraft observation simulation data generator; S2, constructing a cost matrix and correcting the cost matrix based on the detection probability and the clutter intensity; S3, calculating the number column through the corrected cost matrix, and estimating the number of targets by applying the k-means clustering algorithm; S4, processing the cost matrix according to the clustering result, applying the Hungarian algorithm to obtain the angle measurement data association result, and calculating the multi-target position according to the association result, so as to realize the correct association of the angle measurement data of multiple targets by the observation star and the multi-target collaborative positioning. The present invention eliminates the cluttered angle measurement data, correctly associates the measurement from the same target, and correctly estimates the number and position of the target at the current moment, and solves the problem of data missed detection and data association and positioning under clutter conditions in the collaborative multi-target observation of dual-stars in a cluttered environment.

Description

Technical Field

[0001] The present invention relates to the technical field of space non-cooperative target situation awareness, and in particular to a space-based passive collaborative multi-target positioning method in a complex space environment. Background Art

[0002] With the rapid development of human space technology, the surge in satellites and space debris in low-Earth orbit has made orbital space increasingly crowded and faced with security challenges. In order to cope with the possible collision risk between one's own satellites and non-cooperative targets such as space debris and failed satellites, it is urgent to develop space situational awareness technologies including target identification, orbit positioning, target tracking and maneuver detection technologies. The coordinated development of these key technologies will provide a solid foundation for the stability and security of orbital space and ensure the sustainable exploration and utilization of space in the future.

[0003] It is difficult for my country to spread ground-based observation platforms all over the world. For the increasing number of space targets in near-Earth orbit, it is difficult for ground-based observation platforms to achieve space target situation awareness tasks. Commonly used space-based observation methods include optical camera observation, active and passive radar observation, etc. Space-based optical observation sensors are widely used in various task links of space situation awareness due to their advantages such as small size, low cost and good concealment. In the initial orbit determination and tracking tasks of space-based passive multi-targets, a single observation platform is limited by factors such as observation angle, solar incidence angle, and earth occlusion. Multiple observation satellites are often used to observe multiple target satellites in coordination. In the actual space-based passive collaborative multi-target initial orbit determination, even if the observation satellites observe the same space area, due to the presence of clutter and missed detection in the sensors, the angle measurement data obtained by different observation satellites may be quite different, and it is impossible to determine whether the angle measurement data comes from the real target or clutter. How to associate the angle measurement data from the same target detected by different observation satellites, estimate the number of multiple targets and eliminate the influence of clutter and missed detection on the angle measurement data is the premise for realizing collaborative tracking and orbit determination.

[0004] At present, the main approaches to solving this problem at home and abroad are as follows: Assuming that all observation stars can observe all targets, construct a certain characteristic correlation parameter between the angle measurement data of different observation stars, use the parameter value to evaluate the possibility that the two angle measurement data come from the same target, and construct an n*n cost matrix to apply the allocation algorithm to obtain the correlation matrix with the minimum correlation parameter. However, in actual application scenarios, the number of targets observed by different observation stars may be different, and the actual measurement data obtained includes measurement data generated by clutter and measurement data generated by real targets. Most of the existing methods are difficult to apply in actual scenarios or have poor performance. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a space-based passive collaborative detection multi-target data association and multi-target positioning method, which can still maintain good angle measurement data association performance and positioning capability under a weaker measurement acquisition background.

[0006] A space-based passive collaborative multi-target positioning method in a complex space environment comprises the following steps: step 1, establishing a measurement clutter model and a missed detection model in consideration of an actual space observation environment, and constructing a spacecraft observation simulation data generator; step 2, constructing a cost matrix according to a data association probability index and correcting the cost matrix based on a detection probability and a clutter intensity; step 3, calculating a number column through the corrected cost matrix, and applying a k-means clustering algorithm to estimate the number of targets; step 4, processing the cost matrix according to the clustering result, applying a Hungarian algorithm to obtain an angle measurement data association result, and calculating the multi-target positions according to the association result; step 5, applying the above algorithm to an observation satellite, inputting clutter intensity and detection probability, and realizing the correct association of the angle measurement data of multiple targets by the observation satellite and the multi-target collaborative positioning.

[0007] Preferably, step 1 is specifically as follows: step 1.1, establishing a spacecraft dynamics model and an optical camera line-of-sight angle measurement model that consider the non-spherical perturbations of the earth and multi-body gravitational perturbations in the J2000 coordinate system; step 1.2, considering the actual spacecraft observation environment, establishing a measurement clutter model and a sensor missed detection model and constructing a spacecraft observation simulation data generator.

[0008] Preferably, the measurement clutter model construction method in step 1.2 is: whether the clutter generated by each pixel in the field of view of the optical sensor conforms to the Bernoulli distribution, then the number of clutter existing in all pixels in the field of view of the sensor conforms to the binomial distribution. Since the number of pixels is large, the Poisson distribution can be used to more accurately approximate the binomial distribution, so a clutter number distribution model that obeys the Poisson distribution can be constructed.

[0009] As a preferred method, the sensor missed detection model construction method in step 1.2 is: construct the observation equation ,h k (x k ) is the definition of the observation equation for x, represents the angle measurement noise at this moment. The detection probability of the sensor detecting the target is , then the missed detection model can be written as , where the subscript k represents the current time, Indicates the target position at the current moment. The goal for this moment The observed value of To obtain the likelihood function of the current measurement from the target, is the set of measurements at time k.

[0010] Preferably, step 2 is specifically as follows: step 2.1, constructing a cost matrix using the minimum distance model as an evaluation index of the angle measurement data association probability; step 2.2, correcting the cost matrix considering the sensor detection probability and the clutter intensity.

[0011] Preferably, the formula for correcting the cost matrix based on sensor detection probability in step 2.2 is: ,in is the cost matrix before correction, is the modified cost matrix.

[0012] As a preferred method, the method of correcting the cost matrix based on clutter intensity in step 2.2 is: for the cost matrix Each column in is corrected separately, and the corrected matrix elements are recorded as ,in is the clutter strength of the sensor, are the elements in the cost matrix. The calculation formula is ,in is the parameter of the Poisson distribution, which represents the expected value of the number of clutter. is the spatial distribution function of clutter.

[0013] As a preferred embodiment, step 3 is specifically as follows: Step 3.1: by using the modified cost matrix Calculate the number column, process it with the k-means clustering algorithm, and estimate the actual number of targets and the actual number of clutter at the current moment.

[0014] Preferably, step 4 is as follows: pruning the cost matrix according to the clustering results of the logarithmic series to ensure that the number of rows of the cost matrix is ​​equal to the actual number of targets, using the Hungarian algorithm to obtain the association matrix, calculating the minimum distance line segment according to the minimum distance model based on the association results of the angle measurement data, and calculating the target position on the minimum distance line segment.

[0015] Beneficial effects:

[0016] (1) The present invention solves the problem of missed angle measurement data and angle measurement data association and positioning under clutter conditions when dual-satellite coordinated multi-target observation is performed in a clutter environment. By constructing a model and proposing an algorithm, it is possible to eliminate angle measurement data from clutter, correctly associate measurements from the same target, and correctly estimate the number of targets and target positions at the current moment.

[0017] (2) The present invention can eliminate the measurement originating from clutter and consider the influence of missed detection on the correlation performance through the improvement of the algorithm, and finally realize the correct correlation of the angle measurement data from the same target between different sensors without the need to introduce additional equipment.

[0018] (3) The present invention estimates the number of targets at the current moment by processing the number series using a clustering algorithm, without manually selecting a threshold to process the number series, thereby improving the systematicness and robustness of the algorithm.

[0019] (4) The present invention calculates the association matrix by using the Hungarian algorithm, without the need to traverse the cost matrix, thereby ensuring the running speed and performance of the algorithm.

[0020] (5) The present invention uses dual-satellite collaboration to perform initial multi-target positioning, thereby solving the problem of weak observability of a single sensor and improving the positioning accuracy of the target. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of an actual angle measurement data association scene according to an embodiment of the present invention;

[0022] Figure 2 This is a flowchart of an algorithm implementation of an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of a minimum distance model according to an embodiment of the present invention;

[0024] Figure 4 In one embodiment of the present invention, the orbital inclination is 0.01° and the angle measurement error is 1×10 -4 Correlation result diagram when rad;

[0025] Figure 5 In one embodiment of the present invention, the orbital inclination is 0.01° and the angle measurement error is 1×10 -5 Correlation result diagram when rad;

[0026] Figure 6 In one embodiment of the present invention, the orbital inclination is 0.001° and the angle measurement error is 1×10 -5 rad. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] The present invention discloses a method for associating and positioning angle measurement data of dual-satellite passive collaborative multi-targets in a complex space environment. Aiming at the problem of how to correctly associate angle measurement data of the same target observed by different observation stars with the multi-target orbital position determination when the measurements obtained when dual-satellite collaboratively observes multiple targets under a weak measurement acquisition background contain clutter and missed detection, a method for associating space-based passive dual-satellite collaborative multi-target observation data and determining the multi-target orbital positions based on parameter correction and cluster estimation is proposed.

[0029] Combination Figures 1 to 3 , a dual-satellite passive cooperative multi-target angle measurement data association positioning method in a complex space environment, comprising the following steps:

[0030] Step 1: Considering the actual space observation environment, the measurement clutter model and missed detection model are established, and a spacecraft observation simulation data generator is constructed.

[0031] Step 1.1, establish the spacecraft dynamics model and optical camera line-of-sight angle measurement model considering the non-spherical perturbation of the earth and multi-body gravitational perturbation in the J2000 coordinate system.

[0032] Specifically, the spacecraft dynamics model of the target and the observation star is established in the J2000 coordinate system. The origin of the J2000 coordinate system is located at the center of the earth, the xy plane is the earth's equatorial plane, the x-axis points to the vernal equinox at the J2000 time, and the z-axis and the xy plane form a right-handed system. The influence of atmospheric drag perturbation and solar light pressure is ignored, and the absolute dynamics model including the non-spherical perturbation of the earth and the multi-body gravitational perturbation is constructed as follows:

[0033] ,

[0034] in, and are the non-spherical perturbation acceleration of the earth and the multi-body gravitational acceleration, is the spacecraft's position vector, is the Earth's gravitational constant. and The detailed expression is as follows:

[0035] ,

[0036] in, represents the position vector of the Moon, represents the lunar gravitational constant.

[0037] The following satellite optical camera sight angle measurement model is constructed, assuming that the state of the target in the absolute inertial system at time k is: , the state quantity of the observed star in the absolute inertial system at the current moment is recorded as , then the passive observation equation can be written as: ,in, is the measurement from the target at time k, and for The elements in To measure noise, and denote the azimuth and altitude respectively and are defined by:

[0038] ,

[0039] .

[0040] Step 1.2, considering the actual spacecraft observation environment, establish the measurement clutter model and sensor missed detection model and build a spacecraft observation simulation data generator.

[0041] Specifically, the reason for constructing the measurement clutter model and sensor missed detection model is that when the satellite-borne passive optical sensor extracts the target's sight vector through the star map, it is often unable to obtain completely accurate target information. Due to factors such as sensor error, sunlight angle, and earth occlusion, the actual measurement data obtained includes measurement data caused by clutter such as space debris and stars and missed detection of some targets, and the number of targets in each frame of the image is different. Establishing a reasonable clutter and missed detection model aims to take clutter and missed detection into account in the data association algorithm to eliminate the impact of clutter and missed detection on association performance and positioning accuracy, thereby realizing space-based passive dual-satellite collaborative multi-target observation data association and multi-target positioning.

[0042] When establishing a measurement clutter model, whether each pixel in the field of view of the optical sensor generates clutter can be considered to conform to the Bernoulli distribution. Then the number of clutter existing in all pixels in the sensor field of view conforms to the binomial distribution. Due to the large number of pixels, the Poisson distribution can be used to more accurately approximate the binomial distribution to describe the number of clutter in space. Therefore, the clutter number distribution model and clutter space distribution model that obey the Poisson distribution can be constructed as follows:

[0043] ,

[0044] Among them, the clutter model at time k can be recorded as , clutter set Each noise in is independent and identically distributed. is the number of clutter at the current moment, is the Poisson distribution parameter, is the spatial distribution function of the ith clutter, Po represents the Poisson distribution.

[0045] Clutter strength Integrating in the measurement space gives:

[0046] .

[0047] When constructing the sensor missed detection model, the observation equation can be written as , the detection probability of the sensor detecting the target is , that is, when the passive optical sensor extracts the target's measurement information through the star map, The probability of losing the measurement information of a certain target is , then the missed detection model can be written as:

[0048] ,

[0049] The subscript k represents the current time. Indicates the target position at the current moment. represents the angle measurement noise at this moment, The goal for this moment The observed value of is the detection probability, To obtain the likelihood function of the current measurement from the target, is the set of measurements at time k.

[0050] According to the above clutter model and missed detection model, the following data generator is constructed: the set of observation data that the sensor can obtain It should be the union of the measurement set and the clutter set, which can be used As artificially synthesized observation data simulates real observation data, a possible angle measurement data association scenario is as follows Figure 1 As shown in the figure, two observation satellites are used to observe three targets. For observation satellite 1, a total of 5 measurements are obtained at this moment, including 3 measurements from the target and 2 measurements from clutter; for observation satellite 2, a total of 3 measurements are obtained at this moment, including 2 measurements from the target and 1 measurement from clutter. At this moment, observation satellite 2 misses detection and does not detect target 3. The detailed algorithm implementation process is as follows Figure 2 shown.

[0051] Step 2: construct a cost matrix based on the data association probability index and modify the cost matrix based on the detection probability and clutter intensity.

[0052] Step 2.1, construct the cost matrix using the minimum distance model as the evaluation index of the angle measurement data association probability.

[0053] Specifically, for the scenario where two observation satellites observe multiple targets, each observation satellite obtains an observation data set, which includes measurements from clutter and measurements from targets. In order to describe the possibility that any two measurements between different observation satellites come from the same target, a minimum distance model is introduced, such as Figure 3As shown. The two measurements obtained by different observation stars constitute two line-of-sight vectors in space. By constructing a distance model between line-of-sight vectors and performing optimization, the minimum distance segment between any two line-of-sight vectors can be calculated. Let's assume that at a certain moment there are m measurements in the observation data set of observation star 1 and n measurements in the observation data set of observation star 2. The minimum distance between the i-th measurement of observation star 1 and the j-th measurement of observation star 2 is ,make , we can get the following cost matrix :

[0054] .

[0055] Cost Matrix Describes the possibility that measurements from different sensors come from the same target, The larger the value, the more likely it is that the i-th measurement of observation star 1 and the j-th measurement of observation star 2 come from the same target; The smaller it is, the less likely it is that the i-th measurement of observation star 1 and the j-th measurement of observation star 2 come from the same target.

[0056] In step 2.2, the cost matrix is ​​modified by considering the sensor detection probability and clutter intensity.

[0057] Specifically: the above cost matrix Although the possibility that measurements from different sensors come from the same target is described, the impact of measurements from clutter and sensor missed detection on the cost matrix is ​​not considered. Therefore, the cost matrix needs to be corrected by considering the above two factors. Assuming that the detection probability of observed star 1 is , the clutter intensity of the observation star 2 is , the formula for correcting the cost matrix based on the sensor detection probability is: ,in is the cost matrix before correction, is the corrected cost matrix. After completing the detection probability correction, the method of correcting the cost matrix based on the clutter intensity is: for the cost matrix Each column in is corrected separately, and the corrected matrix elements are ,in is the clutter strength of the sensor, The cost matrix The elements in . The calculation formula is ,in is the parameter of the Poisson distribution, which represents the expected value of the number of clutter. is the spatial distribution function of clutter, and the final corrected cost matrix is ​​recorded as .

[0058] Step 3: Calculate the number column through the modified cost matrix and apply the k-means clustering algorithm to estimate the number of targets.

[0059] Specifically: the corrected cost matrix By summing up the columns, we get a matrix with m rows and 1 column. This matrix is ​​defined as the number column. The values ​​of the elements in the number column express the number of explicit targets at the current moment. An explicit target refers to a target whose observation data sets of both sensors at the current moment contain measurements from the target. To determine the orbital position of a target, the necessary condition is that the observation data sets of both sensors at this moment contain measurements from the target. After calculating the number column through the modified cost matrix, the values ​​in the number column often show partial polarization. The m measurements observed by sensor 1 correspond to m rows respectively. When there are measurements from the same target in sensor 2, the values ​​of the elements in the corresponding row in its number column are larger; when there are no measurements from the same target in sensor 2, the values ​​of the elements in the corresponding row in its number column are smaller; therefore, the number column can be processed by the k-means clustering algorithm to distinguish a larger value array from a smaller value array. The row corresponding to the element in the smaller numerical array can be considered as the row where the clutter of sensor 1 is located, and the number of elements in the larger numerical array can be considered as the number of explicit targets at the current moment. Therefore, the number of explicit targets at the current moment can be estimated through the clustering algorithm.

[0060] Step 4: Process the cost matrix according to the clustering results, apply the Hungarian algorithm to obtain the angle measurement data association results, and calculate the multi-target positions according to the association results.

[0061] Specifically: Process the cost matrix according to the results of the k-means clustering algorithm , delete all elements in the row corresponding to the smaller value array, so that the cost matrix It becomes a matrix with k rows and n columns, denoted as , where k is the estimated number of targets at the current moment. The Hungarian optimal allocation algorithm is used to calculate the correlation matrix, and the correlation matrix is ​​defined as follows:

[0062] ,

[0063] .

[0064] Find the optimal incidence matrix through the Hungarian optimal assignment algorithm , the optimal incidence matrix It indicates which measurement data from observation star 2 the measurement data from each target obtained by observation star 1 should be associated with, and then the multi-target position is calculated according to the association result of the angle measurement data as follows:

[0065] According to the optimal correlation matrix Take out each pair of associated angle measurement data. According to the minimum distance model mentioned in the above steps, there is a minimum distance line segment between the two sight vectors formed by any two measurement data. Let's assume that the coordinates of the two endpoints of the minimum distance line segment in space are and , then the estimated position of the target can be considered as the midpoint on the minimum distance line segment:

[0066] ,

[0067] The optimal incidence matrix All the associated pairs of angle measurement data in are taken out, and the estimated positions of multiple targets can be obtained by calculating according to the above formula.

[0068] Step 5, apply the above algorithm to the observation satellite, input the clutter intensity and detection probability, and realize the correct association of the angle measurement data of multiple targets by the two observation satellites and the multi-target collaborative positioning.

[0069] The feasibility of the present invention is illustrated with reference to the following examples.

[0070] Set the following calculation conditions and technical parameters:

[0071] 1) Two observation satellites are deployed in GEO orbit. The eccentricity of the two observation satellites is 0, the argument of perigee is 0°, the right ascension of the ascending node is 0°, and the true anomaly angles are 89.8836° and 90.1164° respectively. The orbital inclination of observation satellite 1 is set as a dynamic parameter, which can be 0.01° or 0.001°, and the orbital inclination of observation satellite 2 is 0°;

[0072] 2) The ten target satellites are deployed in a GEO neighborhood orbit 1000km higher than the GEO orbit. The eccentricity of the ten target satellites is 0, the orbit inclination is 0°, the perigee angle is 0°, the right ascension of the ascending node is 0°, the true anomaly angle between each two target satellites differs by about 0.13°, and the interval between the target satellites is about 100km;

[0073] 3) The angular measurement error of the sensor is set as a dynamic parameter to 1×10 -4 rad, 1×10 -5 rad.

[0074] Based on the dual-satellite collaborative angle measurement data correlation positioning method of the present invention and the above-mentioned calculation conditions and technical parameters, a simulation experiment was carried out using MATLAB software. For each set of dynamic parameters, 1000 Monte Carlo target shooting simulation experiments were carried out. The correlation accuracy under different conditions was obtained as follows: Figure 4 , Figure 5 and Figure 6 As shown in , the correct association is defined as the correct association of the angle measurement data from the same target between different sensors based on the correct estimation of the number of explicit targets. Figure 4 The orbital inclination of the observation star 1 is 0.01°, and the angle measurement error of the two stars is 1×10 -4 Correlation result diagram when rad; Figure 5 The orbital inclination of the observation star 1 is 0.01°, and the angle measurement error of the two stars is 1×10 -5 Correlation result diagram when rad; Figure 6 The orbital inclination of the observation star 1 is 0.001°, and the angle measurement error of the two stars is 1×10 -5 rad. When the angle measurement accuracy of the binary star is 1×10 -5 rad, the proposed method can maintain good correlation performance under different sensor detection probabilities and clutter expectations. When the orbital inclination is 0.01°, the algorithm correlation accuracy remains above 90%; when the orbital inclination is 0.001°, the algorithm correlation accuracy remains above 85%. -4 rad, the proposed method gradually decreases with the increase of the clutter expectation value, but when the clutter expectation value is 10, the association accuracy is still above 65%. By differentiating the estimated position of multiple targets with the true position, the positioning accuracy of the algorithm can be calculated. At a relative distance of 1000km, when the angle measurement accuracy of the sensor is 1×10 -4 rad, the average position error of the target is 0.7416km; when the angle measurement accuracy of the sensor is 1×10 -5 rad, the average position error of the target is 0.0680 km.

[0075] Therefore, the method of the present invention can correctly associate the angle measurement data of the two observation satellites coordinating to observe multiple targets and calculate the corresponding multi-target positions in the weak measurement background containing clutter and missed detection. Compared with the traditional passive data association method, the method of the present invention solves the problem that the traditional method ignores sensor clutter and missed detection, resulting in the inability to be applied in practice or poor application performance. Experiments have proved that the proposed method has superior performance in the GEO orbital space.

[0076] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A space-based passive collaborative multi-target positioning method in a complex space environment, characterized in that: The steps include: Step 1: Consider the actual space observation environment to establish the measurement clutter model and missed detection model, and build a spacecraft observation simulation data generator. The measurement clutter model conforms to the Poisson distribution and is expressed as: ,in, is the clutter model at time k, the clutter set Each noise in is independent and identically distributed. is the number of clutter at the current moment, is the Poisson distribution parameter, For the i The spatial distribution function of the clutter; The missed detection model is expressed as: , where subscript k Indicates the current moment, Indicates the target position at the current moment. represents the angle measurement noise at this moment, The goal for this moment The observed value of To obtain the likelihood function of the current measurement from the target, is the set of measurements at time k; Step 2: Construct a cost matrix based on the data association probability index and modify the cost matrix based on the detection probability and clutter intensity. In order to describe the possibility that any two measurements between different observation stars come from the same target, the minimum distance model is used as the angle measurement data association probability evaluation index to construct the cost matrix, and based on the detection probability Modify the cost matrix, the modified cost matrix is ​​expressed as: ,in, is the cost matrix before correction, , is the minimum distance between the i-th measurement of observation star 1 and the j-th measurement of observation star 2, The larger it is, the more likely it is that the two measurements come from the same target; is the modified cost matrix; Step 3, calculate the number column through the modified cost matrix, and apply the k-means clustering algorithm to estimate the number of targets; Step 4, process the cost matrix according to the clustering results, apply the Hungarian algorithm to obtain the angle measurement data association results and calculate the multi-target positions according to the association results; Step 5, apply the algorithm of steps 1-4 to the observation satellite, input the clutter intensity and detection probability, and realize the correct association of the observation satellite's angle measurement data of multiple targets and multi-target collaborative positioning.

2. The space-based passive collaborative multi-target positioning method in a complex space environment according to claim 1, characterized in that: The data generator is the union of the measurement set and the clutter set, expressed as: , is the set of measurements at time k, A collection of clutter.

3. The space-based passive collaborative multi-target positioning method in a complex space environment according to claim 1, characterized in that: The method of correcting the cost matrix based on clutter intensity is: Each column in is corrected separately, and the corrected matrix elements are recorded as ,in is the clutter strength of the sensor, is the element in the cost matrix; ,in is the parameter of the Poisson distribution, which represents the expected value of the number of clutter. is the spatial distribution function of clutter.

4. The space-based passive collaborative multi-target positioning method in a complex space environment according to claim 3 is characterized in that: In step 3, the modified cost matrix Add the sum by column, and get a matrix with m rows and 1 column, which is defined as the number column, expressing the number of explicit targets at the current moment; The number column is processed based on the k-means clustering algorithm to distinguish between large value arrays and small value arrays. The rows corresponding to the elements in the small value array are considered to be the rows where the clutter is located, and the number of elements in the large value array is considered to be the number of explicit targets at the current moment.

5. The space-based passive collaborative multi-target positioning method in a complex space environment according to claim 4, characterized in that: In step 4, the cost matrix is ​​pruned according to the clustering results of the logarithmic series to ensure that the number of rows in the cost matrix is ​​equal to the actual number of targets. The Hungarian algorithm is used to obtain the association matrix. The minimum distance segment is calculated based on the minimum distance model based on the association results of the angle measurement data, and the target position is calculated on the minimum distance segment.

6. The space-based passive collaborative multi-target positioning method in a complex space environment according to claim 5, characterized in that: In step 4, the pruning cost matrix is: All elements in the row corresponding to the small value array make the cost matrix becomes a matrix with k rows and n columns , where k is the estimated number of targets at the current moment, and the pruned cost matrix The Hungarian optimal assignment algorithm was used to calculate the incidence matrix.

7. The space-based passive collaborative multi-target positioning method in a complex space environment according to claim 5 or 6, characterized in that: In step 4, the incidence matrix is ​​expressed as: ， ; Find the optimal incidence matrix through the Hungarian optimal assignment algorithm , the optimal incidence matrix It shows the data correlation results between the observed stars, according to the optimal correlation matrix Take out each pair of associated angle measurement data, and assume that the coordinates of the two endpoints of the minimum distance line segment in space are and , the estimated position of the target is considered to be the midpoint on the minimum distance line segment: ; The optimal incidence matrix All the associated pairs of angle measurement data in are taken out to obtain the estimated positions of multiple targets.

Images