A distributed airship radar ground moving target accurate positioning method

Abstract

The application provides a distributed airship radar ground moving target accurate positioning method, and relates to the field of radar moving target detection. The method first determines a ground irradiation range according to system beam parameters and constructs a two-dimensional discretized virtual grid; then selects a main processing visual angle, filters out an initial distance ring representing a potential position of a target by matching a grid point theoretical equivalent distance with an actual detection value; then iterates and sequentially fuses the remaining auxiliary visual angles, and obtains an accurate positioning result of the target by using a geometric center method; finally, a multi-angle consistency dynamic verification mechanism is introduced, detection consistency features among the visual angles are counted, and a final verification result of the target is output according to a preset judgment criterion. The application realizes integrated processing of cross-platform track association, positioning and verification, overcomes the problem that the accuracy of the traditional method is limited, enhances the anti-false alarm and missing alarm interference capability, and effectively reduces the calculation complexity of the algorithm through grid reduction.

Description

Technical Field

[0001] This invention belongs to the field of radar moving target detection, specifically to a method for precise localization of ground moving targets using distributed airship radar. Background Technology

[0002] Currently, Ground Moving Target Indication (GMTI) technology is mainly divided into two detection modes: Wide Area Surveillance GMTI and Synthetic Aperture Radar GMTI. From the perspective of platform architecture, single-platform systems are simple in structure and easy to deploy, but their surveillance range and continuous surveillance capability are limited by the physical platform. Although distributed multi-platform systems can improve the detection dimension through multi-view collaboration, airborne platforms have the problem of limited endurance, while spaceborne platforms face the challenge of high deployment and maintenance costs.

[0003] Near space has attracted widespread attention in recent years due to its unique spatial location and atmospheric environment. Stratospheric airships, as important near-space detection platforms, possess outstanding advantages such as high payload capacity, long endurance, and good hovering stability. When the airship platform hovers above the target area, there is almost no relative motion between the ground clutter and the platform, resulting in a very narrow main lobe clutter spectrum in the received echo. This allows for the effective detection of slow-moving, weak targets without complex clutter suppression processing, providing a new development opportunity for GMTI (Geostationary Target Detection and Interference) technology. Invention patent CN109188387B discloses a distributed coherent radar target parameter estimation method based on interpolation compensation. The core technology of this invention is to use interpolation compensation to eliminate the extra phase caused by energy accumulation to the target echo, thereby accurately estimating the position coordinates of slow-moving targets on the ground. However, it only focuses on phase error correction and interpolation fitting under the coherent radar system and does not involve the underlying logic optimization of cross-platform association of target points in non-coherent multi-view detection scenarios. Furthermore, this parameter estimation model fails to fundamentally solve the false alarm interference problem in multi-platform collaborative detection, lacks a dynamic verification mechanism, and cannot provide a set of integrated positioning solutions that take into account both high positioning accuracy and high anti-interference capability for distributed systems. The invention patent CN110412559A discloses a noncoherent fusion target detection method for distributed UAV MIMO radar. The core technology of this invention is to construct a four-dimensional search grid of position and velocity, and to obtain the optimal detector through generalized likelihood ratio and maximum likelihood estimation for centralized fusion detection. However, it only focuses on replacing point matching between receivers with exhaustive search in all dimensions, without involving a computational optimization strategy of gradually reducing the candidate grid point set through sequential fusion iteration. Furthermore, the construction of the four-dimensional search grid and the calculation of clutter covariance matrix significantly increase the computational dimension and processing burden of the system, making it difficult to meet the stringent requirements of real-time performance and computational cost in practical engineering while ensuring high positioning accuracy. Invention patent CN120652413A discloses a ground moving target detection method based on a distributed airship platform. The core technology of this invention is to obtain the energy matrix of the distance-velocity two-dimensional search point based on RFT transform and perform multi-view energy superposition. It combines Kalman filtering and nearest neighbor algorithm for data association and trajectory tracking. However, it only focuses on the processing of multi-view energy accumulation and back-end trajectory tracking, without addressing the elimination of contradictory observation evidence from the underlying mechanism during the gridded positioning stage. Furthermore, it still relies on complex nearest neighbor data association algorithms to handle multi-target cross-tracks and lacks an endogenous verification mechanism to automatically eliminate false / missed alarms through the "majority" criterion of multi-view consistency. It is prone to association errors in complex electromagnetic environments and under high false alarm probabilities, and its robustness in engineering implementation is insufficient.

[0004] Existing research has proposed using stratospheric airships as radar platforms for Earth observation, but most studies focus on system conceptual design and preliminary engineering verification. Currently, mature technical solutions have not yet been developed to address key technical challenges in distributed airship platform collaborative detection scenarios, such as multi-view echo information fusion, cross-platform point correlation, and comprehensive interpretation of false alarms and missed alarms. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of the existing technology by providing a distributed airship radar method for precise localization of moving ground targets. This method constructs a distributed multi-platform observation configuration and discretizes the monitored area into a grid. It utilizes sequential fusion and iteration of multi-view information to achieve rapid convergence and precise localization of potential target areas. Furthermore, it integrates a dynamic verification module based on a consistency criterion, giving it advantages such as the ability to achieve cross-platform point correlation and integrated discrimination of false alarms and missed alarms, significantly improved detection reliability in complex environments, effectively reduced algorithm computational complexity, and convenient engineering implementation.

[0006] The technical solution to achieve the purpose of this invention is as follows:

[0007] A method for precise localization of ground moving targets using distributed airship radar includes the following steps:

[0008] Step S1: Obtain the cooperative detection echo of the distributed radar, determine the ground monitoring area and construct a discretized virtual grid, and initialize the grid point status;

[0009] Step S2: Select the main processing viewpoint and calculate the theoretical equivalent distance corresponding to the virtual grid point; match it with the actual target distance detected by the radar under this viewpoint, filter out the grid points that meet the matching conditions, and form a range loop representing the potential position of the target.

[0010] Step S3: Traverse the remaining auxiliary views, perform consistency matching between the theoretical equivalent distance of the grid points in the candidate set and the actual detection distance of each auxiliary view, and dynamically remove mismatched grid points; based on the final converged grid point set, calculate and obtain the accurate positioning result of the target;

[0011] Step S4: Introduce a multi-view consistency dynamic verification mechanism, statistically analyze the detection consistency features between different views, determine the authenticity of the detected target according to the preset decision criteria, and output the final positioning verification result.

[0012] Furthermore, the initialization of the grid point state specifically includes: setting a two-dimensional discretized virtual grid around it. , its in shaft and The resolutions of the axes are respectively and Each grid point This represents a potential target location and is assigned a binary state variable. This is used to indicate the probability of a target existing at that location. The initial state is zero, representing no target.

[0013] In the formula, It is a two-dimensional discretized virtual mesh. It is a binary state variable.

[0014] Furthermore, one of the airship platforms is selected as a launch station to transmit radar pulse signals to the ground monitoring area from the primary processing perspective; the remaining airship platforms serve as receiving stations, arranged in a distributed spatial geometry with the launch station, to collaboratively receive the echo signals formed by the reflection of the radar pulse signals from moving targets on the ground.

[0015] Furthermore, after initializing the grid point states, the algorithm begins processing a detected target in the main processing viewpoint, recording the actual detection value of its corresponding equivalent one-way distance history as... The equivalent one-way distance journey is the distance from the target to the launch station. Distance and main processing perspective Half of the sum of the distances from the target is determined by the target's perspective. The distance cell containing the CFAR detection result is directly obtained;

[0016] ground grid Each grid point within Calculate its distance to the launch station The distance to the main processing viewpoint is used as the equivalent one-way distance history of that grid point in the main processing viewpoint. :

[0017] In the formula, and These represent the geometric distance from the virtual grid point on the ground to the radar transmitting station, and the geometric distance from the virtual grid point on the ground to the main processing viewpoint, respectively.

[0018] Traversing all grid points, grid points with a state of 1 form a distance loop. The distance loop is a narrow annular region contained only within the main processing view. Given the constraints of observation information, the set of all possible initial positions of the target. : .

[0019] Furthermore, the traversal of the remaining auxiliary views specifically includes: calculating the theoretical equivalent distance of the current candidate grid point under each auxiliary view, and performing consistency matching with the actual detection distance of the auxiliary view; dynamically updating the state variables of the candidate grid point according to the matching result; gradually reducing the set of candidate grid points through multi-view sequential fusion iteration; and obtaining the accurate positioning result of the target based on the finally converged set of grid points.

[0020] Furthermore, a sequential approach is adopted to integrate information from the remaining auxiliary perspectives, from the perspective... Start, proceed in reverse to For each auxiliary perspective to be merged at present ( Specifically, perform the following operations:

[0021] Iterate through the set of all grid points whose current state is 1. (Initial) ),make The grid points in the auxiliary viewpoint The number of traversals gradually decreases;

[0022] For each grid point Calculate its perspective The equivalent one-way distance history :

[0023] The equivalent one-way distance history and perspective The results of CFAR detection and point aggregation were compared; assuming a perspective Detected in The equivalent one-way distance journey corresponding to each objective is: ( );

[0024] If the equivalent one-way distance journey and perspective A certain detection target in Corresponding equivalent one-way distance history If the difference is within the distance resolution, then the point is considered to be... This auxiliary perspective was used. The verification is performed, its state is kept at 1, and the detected target is... Associate the point with the current target being processed; otherwise, consider the point to be unrelated to the perspective. The observational evidence is contradictory, so its state is reset to 0;

[0025] when After all auxiliary perspectives participate in the fusion, the final set of remaining grid points It will converge to a region due to the influence of distance resolution and grid discretization. It may contain multiple adjacent grid points; in this case, the final position estimate of the target is... This can be obtained by calculating the geometric center of the point set:

[0026] In the formula, The number of points in the set.

[0027] Furthermore, the multi-view consistency dynamic verification mechanism in step S4 specifically includes: obtaining an initial ground grid assignment matrix representing the potential location of the target based on the detection results of the main processing view; introducing a multi-view consistency dynamic verification mechanism to sequentially verify the initial ground grid assignment matrix using the detection information of each auxiliary view, and statistically analyzing the detection consistency features among the multiple views; determining the authenticity of the detected target according to the detection consistency features and a preset decision criterion, and outputting the target's positioning verification result.

[0028] Furthermore, before the multi-view consistency dynamic verification, the core variables need to be defined and initialized, and auxiliary statistical variables need to be introduced. These variables need to be re-initialized every time the main processing view is switched, as shown below:

[0029] Introducing a consistency counter The number of auxiliary viewpoints whose detection information matches the current target position assumption during the statistical fusion process is counted, and the initial value of the consistency counter is set to 1, representing the main processing viewpoint itself;

[0030] Introducing an auxiliary fusion perspective index This represents the auxiliary viewpoint number in the current multi-view information sequential fusion process. When the viewpoint of airship number 1 is selected as the primary processing viewpoint... Iterate through numbers 2 to 3 in any order. From the auxiliary perspective, the consistency counter is initialized to 2;

[0031] Introduce a valid matrix pointer This is used to indicate the view index corresponding to the latest non-zero ground assignment matrix that should be continued to be used. The initial value of the valid matrix pointer is 1, representing the main processing view itself, that is, the ground grid point assignment matrix in the initial state. Equivalent to ;

[0032] Constructing a consistent state vector The consistency state vector has a length equal to the total number of viewpoints. The vector, Each element represents the detection consistency state of the target being processed from the corresponding viewpoint, where each auxiliary element... Corresponding to the The detection state of each auxiliary viewpoint, namely:

[0033] Furthermore, the multi-view consistency dynamic verification traverses all auxiliary views, and for each auxiliary view... Assign matrix based on current ground grid points As input, with auxiliary perspective The detection results are matched and integrated for location fusion processing. Based on the processing results, a new ground grid point assignment matrix is ​​obtained. For matrices To determine if a matrix is ​​zero, a viewpoint consistency check is performed, which can be categorized into the following cases:

[0034] like , indicating perspective The detection results did not contain target information that matched the targets detected from the main processing perspective. This was addressed through the consistency state vector. Marking perspective The detection consistency for the current target is inconsistent, specifically:

[0035] At the same time, due to perspective The detection consistency is abnormal, the ground grid point assignment matrix is ​​not updated, and the valid matrix pointer is maintained. Keep it unchanged, ensuring the next auxiliary view continues to be used. Perform information fusion and consistency verification;

[0036] like , indicating perspective The detection results contain target information that matches the current target; that is, for the current moving target, the main processing view and the auxiliary view... The test results are consistent, at which point the consistency counter needs to be updated. :

[0037] And update the consistency state vector. Valid matrix pointer : ; .

[0038] Furthermore, when all After all auxiliary perspectives have been traversed, the dynamic verification algorithm will use the consistency counter. Make a final judgment based on the value:

[0039] When the target is determined to exist, then the consistency state vector is determined. All views corresponding to elements with a value of 0 in the vector resulted in missed detections for that target; when a target is determined to be a false alarm, it is then determined that: the detection consistency state vector... For all elements with a value of 1, the corresponding CFAR detection output results show false alarms.

[0040] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0041] 1. This invention constructs a unified, discretized virtual ground grid, directly mapping echo observation information from different perspectives to the spatial geographic dimension. Employing a sequential fusion strategy, the candidate grid range containing potential targets achieves physical convergence under multi-view constraints. This mechanism fundamentally eliminates the risk of positioning failure due to matching errors in traditional methods, significantly improving the association accuracy in dense target environments and ensuring sub-meter-level positioning accuracy through fine-grained grid sampling.

[0042] 2. This invention introduces a multi-view consistency dynamic verification mechanism, enabling real-time recording and analysis of target information matching status under observation from different airship platforms. Utilizing a "majority" decision criterion, the system can automatically identify and eliminate false alarm interference generated by a single or few views, while effectively supplementing missed alarms under specific views. This integrated decision-making method based on the observation evidence chain effectively resists the impact of false alarms caused by clutter residue and electronic interference, greatly improving the robustness of the detection system in complex battlefield electromagnetic environments.

[0043] 3. This invention fully utilizes the unique advantages of stratospheric airship platforms, such as high-altitude hovering and strong stationary stability, effectively avoiding the severe clutter broadening caused by motion in traditional airborne platforms. In stratospheric applications, the radar main lobe clutter spectrum is extremely narrow, enabling the system to achieve sensitive detection of slow-moving, weak ground targets without the need for complex space-time adaptive processing (STAP) algorithms. Combined with multi-view geometric spatial gain, this scheme significantly improves the ability to acquire "low, small, and slow" targets while reducing the complexity of backend signal processing, possessing strong practical engineering guiding significance.

[0044] 4. This invention combines distance loop locking with sequential geometric consistency testing to gradually reduce the set of candidate grid points during the fusion process, effectively reducing computational complexity and avoiding unnecessary computational overhead, which is beneficial for achieving real-time target detection. Attached Figure Description

[0045] Figure 1 A flowchart of a method for precise localization of moving ground targets using distributed airship radar;

[0046] Figure 2 To determine the actual ground illumination range of the current wavelength;

[0047] Figure 3 Initialize the ground grid and lock the distance loop;

[0048] Figure 4 To utilize auxiliary perspective Perform observation information fusion;

[0049] Figure 5 To utilize auxiliary perspective Perform observation information fusion;

[0050] Figure 6 To achieve target matching and positioning through multi-perspective information fusion;

[0051] Figure 7 The flow of a dynamic verification algorithm for false alarms and missed alarms based on multi-view consistency;

[0052] Figure 8 The distance-Doppler domain processing results are for the echo from viewpoint 1 among the four airship platforms.

[0053] Figure 9 The range Doppler domain processing results for the echo from viewpoint 2 in the perspective of the four airship platforms;

[0054] Figure 10 The distance-Doppler domain processing results for the echoes from viewpoint 3 in the perspectives of the four airship platforms;

[0055] Figure 11 The distance-Doppler domain processing results for the echo at viewpoint 4 from the perspective of the four airship platforms;

[0056] Figure 12 The results of CFAR and spot aggregation processing of the echo from viewpoint 1 in the viewpoints of the four airship platforms;

[0057] Figure 13 The results of CFAR and spot aggregation processing of the echo from viewpoint 2 in the viewpoints of the four airship platforms;

[0058] Figure 14The results of CFAR and spot aggregation processing of the 3rd echo from the perspective of the four airship platforms are shown.

[0059] Figure 15 The results of CFAR and spot aggregation processing for the echo at viewpoint 4 from the perspective of four airship platforms;

[0060] Figure 16 The initial ground grid distance loop locking result for target 1 is processed when viewpoint 2 is the primary processing viewpoint;

[0061] Figure 17 The result of information fusion from auxiliary viewpoint 1 when viewpoint 2 is the primary viewpoint;

[0062] Figure 18 The result of fusing information from auxiliary viewpoint 4 after fusing information from auxiliary viewpoint 1 when viewpoint 2 is the primary processing viewpoint;

[0063] Figure 19 The result of averaging the information from all remaining viewpoints after fusing them together when viewpoint 2 is the primary viewpoint;

[0064] Figure 20 The initial ground grid distance loop locking result for target 2 is processed when viewpoint 4 is the primary processing viewpoint;

[0065] Figure 21 The result of information fusion from auxiliary viewpoint 1 when viewpoint 4 is the primary viewpoint;

[0066] Figure 22 The result of fusing information from auxiliary viewpoint 1 and auxiliary viewpoint 2 when viewpoint 4 is the primary processing viewpoint;

[0067] Figure 23 The result is the average of the data from all remaining viewpoints after the data from viewpoint 4 is fused together. Detailed Implementation

[0068] To more clearly describe the ideas, technical solutions, and advantages of the present invention, specific embodiments are illustrated through examples and accompanying drawings. Obviously, the described embodiments are only a portion, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0069] Combination Figure 1 The flowchart shown illustrates a method for precise localization of moving ground targets using a distributed airship radar. This embodiment provides a method for precise localization of moving ground targets using a distributed airship radar, specifically including the following steps:

[0070] Step S1, as follows Figure 2As shown, the cooperative detection echo of the distributed radar is obtained to determine the actual ground illumination range of the current wave position and construct a discretized virtual grid. The grid point states are initialized, and a two-dimensional discretized virtual grid is set around them. , its in shaft and The resolutions of the axes are respectively and Each grid point represents a potential target location and is assigned a binary state variable. .

[0071] Step S2, as follows Figure 3 The diagram illustrates the ground grid initialization and range loop locking process. A primary processing viewpoint is selected, and the theoretical equivalent range corresponding to each virtual grid point is calculated. This range is then matched with the actual target range detected by the radar under that viewpoint, and grid points meeting the matching criteria are selected to form a range loop representing the potential location of the target.

[0072] Step S3, as follows Figure 4 and Figure 5 As shown, these are examples of using auxiliary perspectives. and auxiliary perspective The process of fusion of observational information involves iterating through the remaining auxiliary viewpoints, matching the theoretical equivalent distances of grid points in the candidate set with the actual detection distances of each auxiliary viewpoint, and dynamically eliminating mismatched grid points. For example... Figure 6 As shown, the target matching and localization are finally achieved through the sequential fusion of multi-view information. Based on the final converged set of grid points, the accurate localization result of the target is calculated.

[0073] Step S4, as follows Figure 7 The diagram illustrates the algorithm flow for dynamic verification of false alarms and missed alarms based on multi-view consistency. A multi-view consistency dynamic verification mechanism is introduced, which statistically analyzes the detection consistency features across different views and determines the authenticity of detected targets according to preset decision criteria, outputting the final location verification result.

[0074] After verifying the above theoretical process, this embodiment built a specific simulation system. Some simulation system parameters used in this embodiment are shown in Table 1 below; the platform and point target motion parameter settings are shown in Table 2; and the relevant simulation parameter settings in the simulation experiment are shown in Table 3.

[0075] Table 1. Parameters of Partial Simulation System

[0076] Table 2. Platform and Point Target Motion Parameter Settings

[0077] Table 3. Simulation parameter settings in the simulation experiment.

[0078] Under the aforementioned simulation background and parameters, the echoes received by the four airship platforms were processed into the range-Doppler domain, as shown below. Figure 8 , Figure 9 , Figure 10 , Figure 11 As shown. By Figure 8 , Figure 9 , Figure 10 , Figure 11 It can be seen that the distance units of the two moving targets are significantly different from different perspectives. Figure 12 , Figure 13 , Figure 14 , Figure 15 The image shows the processing results after CFAR detection and point aggregation, where false alarms are observed at viewpoints 1, 2, and 3 (i.e., the areas within the cyan-green boxes in the image). It should be noted that, for clearer presentation, the temporal range cells have been truncated: the original range sampling points were truncated from the 25,000th to the 35,000th. Therefore, there is a 25,000-cell offset between the range cells displayed for detected targets and their actual locations in the image. For example... Figure 13 The target detected in the 6151st range cell and the 56th Doppler cell should actually be in the 31150th range cell, which is consistent with Table 2.

[0079] To demonstrate the processing effect of this algorithm, we will take the detection of target 1 as the primary processing viewpoint 2 and the detection of target 2 as the primary processing viewpoint 4 as examples. The detection results of target 1 as the primary processing viewpoint 2 are as follows: Figure 16 , Figure 17 , Figure 18 , Figure 19 As shown, the detection results of target 2 are processed primarily from viewpoint 4. Figure 20 , Figure 21 , Figure 22 , Figure 23 As shown. The remaining test results, especially for... Figure 12 , Figure 13 , Figure 14 The dynamic verification process for each false alarm is summarized in Table 4.

[0080] Table 4. Simulation verification of all experimental results.

[0081] Analysis shows that the ICLV algorithm integrates detection information from four perspectives, accurately associates the same target, and locates it by iteratively updating the ground grid point assignment matrix and taking the average value, with an average error of about 2.4 meters.

[0082] Analyzing Table 4, when the main processing viewpoint is viewpoint 1, 5 targets are detected, of which the first detected target... The value is 2; analyze the corresponding... It can be seen that only viewpoint 1 and viewpoint 2 possess information about the detected target, while the second and fourth detected targets... The value is 1. Based on the dynamic verification algorithm's criteria for determining false alarms, these three targets are false alarms. When the main processing viewpoint is viewpoint 2, three targets are detected, with the third target being... The value is 2, and only viewpoints 1 and 2 have this target information; when the main processing viewpoint is viewpoint 3, only the third target is detected. The value is 1, and only viewpoint 3 itself possesses this target information; when the main processing viewpoint is viewpoint 4, The values ​​represent the total number of viewpoints, indicating that neither of the two detected targets is a false alarm, and all four viewpoints contain information about two targets.

[0083] In summary, all four viewpoints successfully detected the two designated moving targets. Viewpoint 1 generated three false alarms, viewpoints 2 and 3 each generated one false alarm, and viewpoint 4 generated no false alarms. Furthermore, the false alarm generated by viewpoint 2 was one of the false alarms generated by viewpoint 1. This conclusion is consistent with... Figure 8 The prior knowledge is consistent, and the final ICLV algorithm has an average straight-line positioning error of 2.3m for the two targets, which is a relatively good level under the simulation system parameters in this paper.

[0084] Since the above simulation experiments can only verify the processing effect of the algorithm when false alarms are generated from different perspectives, in order to further verify the effectiveness of the algorithm, the simulation results of the effectiveness of the algorithm under three complex false alarm and missed alarm scenarios are given below, as shown in Tables 5, 6 and 7 respectively.

[0085] Table 5. View 1: Missed Alarm Target 1; View 2: False Alarm Target 1; View 3 and View 4: Normal.

[0086] Analysis of Table 5 shows that when the main processing viewpoint is Viewpoint 1, only one target is detected. A value of 4 indicates that the target actually exists; when the main processing view is view 2, three targets are detected, one of which is... The value is 2, and only the assignment matrices of viewpoint 2 and viewpoint 1 have non-zero values, therefore this target is determined to be a false alarm; the remaining two targets The values ​​are 3 and 4 respectively, which meet the judgment criteria, indicating that they truly exist, and When the value is 3, by It can be seen that only the assignment matrix after processing by viewpoint 1 is a 0 matrix, indicating that there is no target information in viewpoint 1, which also meets the condition that viewpoint 1 misses a target in this case; when the main processing viewpoints are viewpoints 3 and 4, two targets are detected. The values ​​all satisfy the judgment conditions, and The values ​​of 3 are all because the target information is not present in viewpoint 1. In summary, the judgment results of this algorithm are consistent with the simulated scenarios in Table 5.

[0087] Table 6. View 1: Normal; View 2: False Alarm - One New Target with Simultaneous Missed Alarms; Target 2, View 3, and View 4: Normal.

[0088] Analysis of Table 6 shows that when the main processing viewpoint is viewpoint 1, the two detected targets correspond to... If both values ​​are greater than half of the total number of views, the two targets are determined to exist; when the main processing view is view 2, the first target is detected. The value is 2, corresponding to The consistency state variables of both viewpoints 3 and 4 are 0, indicating that the target is a false alarm; the second detected target A value of 4 indicates that the target truly exists; when the main processing viewpoint is viewpoint 3, both detected targets are determined to be real by formula (3-13), but the second detected target... The value is 3, and the assignment matrix after verification from viewpoint 2 is 0, indicating that the detected target information is not in viewpoint 2. The situation of viewpoint 4 is consistent with that of viewpoint 3, indicating that viewpoint 2 missed the corresponding target. In summary, the judgment result of this algorithm is consistent with the simulated situation in Table 6.

[0089] Table 7. Viewpoint 1: Normal; Viewpoints 2 and 3: Simultaneous false alarms for the same target; Viewpoint 4: Normal.

[0090] Analysis of Table 7 shows that although false alarms occur for the same target from both perspectives, when perspectives 2 and 3 are used as the primary processing perspectives, the corresponding... The values ​​are all 2, and they are obtained through their respective corresponding values. The data shows that the false alarm target information exists only in viewpoints 2 and 3, and both are determined to be false alarms. Combining this with the results of other processing, target 1 and target 2 are both confirmed to exist. In conclusion, it is ultimately determined that there is one false alarm target in each of viewpoints 2 and 3, and that target 1 and target 2 are both confirmed to exist. This determination is consistent with the scenario simulated in Table 7.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for precise localization of moving ground targets using distributed airship radar, characterized in that, Includes the following steps: Step S1: Obtain the cooperative detection echo of the distributed radar, determine the ground monitoring area and construct a discretized virtual grid, and initialize the grid point status; Step S2: Select the main processing viewpoint and calculate the theoretical equivalent distance corresponding to the virtual grid point; match it with the actual target distance detected by the radar under this viewpoint, filter out the grid points that meet the matching conditions, and form a range loop representing the potential position of the target. Step S3: Traverse the remaining auxiliary views, perform consistency matching between the theoretical equivalent distance of the grid points in the candidate set and the actual detection distance of each auxiliary view, and dynamically remove mismatched grid points; based on the final converged grid point set, calculate and obtain the accurate positioning result of the target; Step S4: Introduce a multi-view consistency dynamic verification mechanism, statistically analyze the detection consistency features between different views, determine the authenticity of the detected target according to the preset decision criteria, and output the final positioning verification result.

2. The method for precise ground target localization using distributed airship radar according to claim 1, characterized in that, The initialization of the grid point state specifically includes: setting a two-dimensional discretized virtual grid around it. , its in shaft and The resolutions of the axes are respectively and Each grid point This represents a potential target location and is assigned a binary state variable. This is used to indicate the probability of a target existing at that location. The initial state is zero, representing no target. ； In the formula, It is a two-dimensional discretized virtual mesh. It is a binary state variable.

3. The method for precise ground target localization using distributed airship radar according to claim 2, characterized in that, One of the airship platforms is selected as the launch station to transmit radar pulse signals to the ground monitoring area from the primary processing perspective; the remaining airship platforms serve as receiving stations, arranged in a distributed spatial geometry with the launch station, to collaboratively receive the echo signals formed by the reflection of the radar pulse signals from moving targets on the ground.

4. The method for precise ground target localization using distributed airship radar according to claim 3, characterized in that, After initializing the grid point states, the algorithm begins processing a detected target in the main processing viewpoint, recording the actual detection value of its corresponding equivalent one-way distance history as... The equivalent one-way distance journey is the distance from the target to the launch station. Distance and main processing perspective Half of the sum of the distances from the target is determined by the target's perspective. The distance cell containing the CFAR detection result is directly obtained; ground grid Each grid point within Calculate its distance to the launch station The distance to the main processing viewpoint is used as the equivalent one-way distance history of that grid point in the main processing viewpoint. : ； In the formula, and These represent the geometric distance from the virtual grid point on the ground to the radar transmitting station, and the geometric distance from the virtual grid point on the ground to the main processing viewpoint, respectively. Traversing all grid points, grid points with a state of 1 form a distance loop. The distance loop is a narrow annular region contained only within the main processing view. Given the constraints of observation information, the set of all possible initial positions of the target. : 。 5. A method for precise ground target localization using distributed airship radar according to claim 4, characterized in that, The traversal of the remaining auxiliary views specifically includes: calculating the theoretical equivalent distance of the current candidate grid point under each auxiliary view, and performing consistency matching with the actual detection distance of the auxiliary view; dynamically updating the state variables of the candidate grid point according to the matching results; gradually reducing the set of candidate grid points through multi-view sequential fusion iteration; and obtaining the accurate positioning result of the target based on the finally converged set of grid points.

6. The method for precise ground target localization using distributed airship radar according to claim 5, characterized in that, The remaining auxiliary perspectives are integrated in a sequential manner, from the perspective... Start, proceed in reverse to For each auxiliary perspective to be merged at present ,in Specifically, perform the following operations: Iterate through the set of all grid points whose current state is 1. Initially ,make The grid points in the auxiliary viewpoint The number of traversals gradually decreases; For each grid point Calculate its perspective The equivalent one-way distance history : ； The equivalent one-way distance history and perspective The results of CFAR detection and point aggregation were compared; assuming a perspective Detected in The equivalent one-way distance journey corresponding to each objective is: ; If the equivalent one-way distance journey and perspective A certain detection target in Corresponding equivalent one-way distance history If the difference is within the distance resolution, then the point is considered to be... This auxiliary perspective was used. The verification is performed, its state is kept at 1, and the detected target is... Associate the point with the current target being processed; otherwise, consider the point to be unrelated to the perspective. The observational evidence is contradictory, so its state is reset to 0; when After all auxiliary perspectives participate in the fusion, the final set of remaining grid points It will converge to a region due to the influence of distance resolution and grid discretization. It may contain multiple adjacent grid points; in this case, the final position estimate of the target is... This can be obtained by calculating the geometric center of the point set: ； In the formula, The number of points in the set.

7. A method for precise ground target localization using distributed airship radar according to claim 6, characterized in that, The multi-view consistency dynamic verification mechanism in step S4 specifically includes: obtaining an initial ground grid assignment matrix representing the potential location of the target based on the detection results of the main processing view; introducing a multi-view consistency dynamic verification mechanism to sequentially verify the initial ground grid assignment matrix using the detection information of each auxiliary view, and statistically analyzing the detection consistency features among the multiple views; determining the authenticity of the detected target according to the detection consistency features and a preset decision criterion, and outputting the target's positioning verification result.

8. A method for precise ground target localization using distributed airship radar according to claim 7, characterized in that, Before the multi-view consistency dynamic verification, the core variables need to be defined and initialized, and auxiliary statistical variables need to be introduced. These variables need to be re-initialized every time the main processing view is switched, as shown below: Introducing a consistency counter The number of auxiliary viewpoints whose detection information matches the current target position assumption during the statistical fusion process is counted, and the initial value of the consistency counter is set to 1, representing the main processing viewpoint itself; Introducing an auxiliary fusion perspective index This represents the auxiliary viewpoint number in the current multi-view information sequential fusion process. When the viewpoint of airship number 1 is selected as the primary processing viewpoint... Iterate through numbers 2 to 3 in any order. From the auxiliary perspective, the consistency counter is initialized to 2; Introduce a valid matrix pointer This is used to indicate the view index corresponding to the latest non-zero ground assignment matrix that should be continued to be used. The initial value of the valid matrix pointer is 1, representing the main processing view itself, that is, the ground grid point assignment matrix in the initial state. Equivalent to ; Constructing a consistent state vector The consistency state vector has a length equal to the total number of viewpoints. The vector, Each element represents the detection consistency state of the target being processed from the corresponding viewpoint, where each auxiliary element... Corresponding to the The detection state of each auxiliary viewpoint, namely:

9. A method for precise ground target localization using distributed airship radar according to claim 8, characterized in that, The multi-view consistency dynamic verification iterates through all auxiliary views, and for each auxiliary view... Assign matrix based on current ground grid points As input, with auxiliary perspective The detection results are matched and integrated for location fusion processing. Based on the processing results, a new ground grid point assignment matrix is ​​obtained. For matrices To determine if a matrix is ​​zero, a viewpoint consistency check is performed, which can be categorized into the following cases: like , indicating perspective The detection results did not contain target information that matched the targets detected from the main processing perspective. This was addressed through the consistency state vector. Marking perspective The detection consistency for the current target is inconsistent, specifically: ； At the same time, due to perspective The detection consistency is abnormal, the ground grid point assignment matrix is ​​not updated, and the valid matrix pointer is maintained. Keep it unchanged, ensuring the next auxiliary view continues to be used. Perform information fusion and consistency verification; like , indicating perspective The detection results contain target information that matches the current target; that is, for the current moving target, the main processing view and the auxiliary view... The test results are consistent, at which point the consistency counter needs to be updated. : ； And update the consistency state vector. Valid matrix pointer : ; 。 10. A method for precise ground target localization using distributed airship radar according to claim 9, characterized in that, When all After all auxiliary perspectives have been traversed, the dynamic verification algorithm will use the consistency counter. Make a final judgment based on the value: ； When the target is determined to exist, then the consistency state vector is determined. All views corresponding to elements with a value of 0 in the vector resulted in missed detections for that target; when a target is determined to be a false alarm, it is then determined that: the detection consistency state vector... For all elements with a value of 1, the corresponding CFAR detection output results show false alarms.

Images