A method for judging the stability of UAV cluster configuration based on multi-source information fusion positioning

Abstract

The present invention discloses a method for judging the stability of a drone cluster configuration based on multi-source information fusion positioning, which belongs to the field of drone positioning and includes the following steps: S1, performing state matching based on multi-source sensor data to achieve data association of multi-source sensor data; S2, calculating errors using two adjacent frames of multi-source sensor data, and then confirming dynamic weights, achieving multi-source information fusion positioning results of a drone cluster based on multi-source sensor data fusion; S3, judging the stability of a drone cluster based on the boundary points of the multi-source information fusion positioning results of a drone cluster. The present invention adopts the above-mentioned method for judging the stability of a drone cluster configuration based on multi-source information fusion positioning, and realizes the positioning of a drone cluster and the judgment of the stability of a drone cluster by fusing multi-source sensor information, which can effectively avoid the problem of cluster positioning failure caused by complex environments and improve the system's ability to locate drone clusters.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle positioning, and in particular to a method for determining the stability of an unmanned aerial vehicle cluster configuration based on multi-source information fusion positioning. Background Art

[0002] Drone swarm warfare has become a new mode of warfare on the modern battlefield. Drones in drone swarm warfare are usually small in size, light in weight, and low in cost. By taking advantage of their numbers, they can make up for the shortcoming of the limited payload of a single drone and form a more powerful firepower and reconnaissance network. Accurate positioning of drone swarms is a prerequisite for drone swarm countermeasures. By accurately locating enemy drone swarms and analyzing their stability, important information about their technical specifications, combat capabilities, and intentions can be obtained, which helps to formulate more accurate response strategies and combat plans, and provides an information basis and reaction time for countering drone swarms.

[0003] The constantly changing configuration of enemy drone swarms makes their positioning particularly complex. Different drone swarms have different shapes, drive modes, characteristics and configurations, which puts higher demands on traditional swarm positioning technology.

[0004] Traditional drone cluster positioning technologies are as follows:

[0005] CN201811547570.7 discloses a multi-rotor UAV and control method based on optical flow sensor positioning and navigation, the method comprising: obtaining an image containing characteristic light spots collected by a visual sensor; determining the flight speed of the UAV according to the coordinate changes of the characteristic light spots in the image; determining the position of the UAV according to the preset starting position of the UAV, the inertial parameters measured by the inertial navigation component and the flight speed. By collecting an image containing characteristic light spots in a set search space through a visual sensor, the flight speed of the UAV can be determined. Then, the position of the UAV can be further determined based on the preset starting position, the determined flight speed and the collected inertial parameters, so that positioning can be achieved without a satellite positioning signal, and further, when the satellite positioning signal is unstable or cannot be received, the flight of the UAV will not be restricted.

[0006] It can be seen that traditional drone clusters are mostly based on a single type of sensor, so they have defects such as limited available information and are easily affected by various factors, which leads to a decrease in the system's ability to locate drone clusters. Summary of the invention

[0007] To solve the above problems, the present invention provides a method for judging the stability of a drone cluster configuration based on multi-source information fusion positioning. By fusing multi-source sensor information, the positioning of the drone cluster and the judgment of the stability of the drone cluster can be realized, which can effectively avoid the problem of cluster positioning failure caused by complex environment and improve the system's ability to locate the drone cluster.

[0008] To achieve the above object, the present invention provides a method for determining the stability of a UAV cluster configuration based on multi-source information fusion positioning, comprising the following steps:

[0009] S1. Perform state matching based on multi-source sensor data to achieve data association of multi-source sensor data;

[0010] S2, by initializing multi-source sensor data, and then using two adjacent frames of multi-source sensor data to calculate the error, and then confirm the dynamic weight, to achieve multi-source information fusion positioning results based on multi-source sensor data;

[0011] S3. Determine the stability of the UAV cluster based on the boundary points of the multi-source information fusion positioning results of the UAV cluster.

[0012] Preferably, step S1 specifically includes the following steps:

[0013] S11. Collect multi-source sensor data using multiple types of sensors, where the multi-source sensor data includes radar data, visual data, and radio data;

[0014] S12, taking the position state information of one type of sensor among the multiple types of sensors as a benchmark, traversing the remaining types of sensor data in pairs by using the nearest neighbor method;

[0015] S13, based on the data of multiple types of sensors obtained by traversal, determine whether they meet the distance constraint conditions; if they do, determine that the multi-source sensor data is successfully associated; otherwise, determine that part of the sensor data is successfully associated, and the data association ends.

[0016] Preferably, the multiple types of sensors in step S11 include radar sensors, optical sensors and radio sensors;

[0017] The radar data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space:

[0018] (1)

[0019] in, Indicates the number of drones detected by the radar sensor; For the The three-dimensional data results of the positioning of the UAV;

[0020] The visual data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space:

[0021] (2)

[0022] in, Indicates the number of drones detected by the optical sensor; For the The three-dimensional data results of the positioning of the UAV;

[0023] The radio data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space:

[0024] (3)

[0025] in, Indicates the number of drones detected by the radio sensor; For the The three-dimensional data results of the positioning of the UAV;

[0026] Step S12 specifically includes the following steps:

[0027] The first drone data detected by radar sensor for drone clusters As a benchmark, the cluster results are located in the optical sensor using the minimization function Medium Search Distance The nearest point, denoted as point :

[0028] (4)

[0029] in, , when the minimum value is found, the corresponding Denote the midpoint between the optical sensor and the radar sensor Associated points , at this time:

[0030] (5)

[0031] At the same time, the minimization function is used to locate the cluster results of radio sensors Medium Search Distance The nearest point, denoted as point :

[0032] (6)

[0033] in, , when the minimum value is found, the corresponding Denote the midpoint between the radio sensor and the radar sensor Associated points , at this time:

[0034] (7);

[0035] Step S13 specifically includes the following steps:

[0036] Add distance constraint to formula (4):

[0037] (8)

[0038] in, The distance constraint threshold is set.

[0039] Preferably, step S2 specifically includes the following steps:

[0040] S21, initializing the first frame of sensor data of multiple types of sensors;

[0041] S22, calculating the position state information error using sensor data of two adjacent frames of the same type of sensor;

[0042] S23, calculating a dynamic weight based on the position state information error;

[0043] S24, fusing multi-source sensor data according to dynamic weights to obtain a multi-source information fusion positioning result;

[0044] S25, determining whether the sensor data of all frames have achieved information fusion, if so, obtaining the positioning result after fusion of multi-source sensor data; if not, returning to step S24.

[0045] Preferably, step S21 specifically includes the following steps:

[0046] S211, let the error of radar sensor in three-dimensional coordinates be , the error of the optical sensor in three-dimensional coordinates is , the error of the radio sensor in three-dimensional coordinates is , satisfying the following relationship:

[0047] (9)

[0048] in, is the proportionality coefficient;

[0049] S212, the weight of radar sensor data fusion is , the weight of optical sensor data fusion is , the weight of radio sensor data fusion is ; then:

[0050] (10)

[0051] (11)

[0052] (12)

[0053] The weights and errors are equal to 1:

[0054] (13)

[0055] Get the fusion result of the first frame of sensor data:

[0056] (14)

[0057] in, It is the fusion result of the first frame of sensor data.

[0058] Preferably, step S22 specifically includes the following steps:

[0059] From the second frame sensor data start, ,in Indicates Frame data, using the previous frame sensor data fusion result As a benchmark, calculate the The data error of the frame sensor data in the three-dimensional coordinates, the error is defined as follows:

[0060] (15)

[0061] (16)

[0062] (17)

[0063] in, is the error of the radar sensor in three-dimensional coordinates, is the error of the optical sensor in three-dimensional coordinates, is the error of the radio sensor in three-dimensional coordinates, , , They represent the radar sensor, optical sensor and radio sensor after data association in the Frame sensor data, It is the sensor data fusion result of the previous frame.

[0064] Preferably, the dynamic weight calculation process in step S23 is as follows:

[0065] The following formula is used to obtain the Scaling factor for frame sensor data :

[0066] (18)

[0067] Get the first Dynamic weighting of frame sensor data:

[0068] (19)

[0069] (20)

[0070] (twenty one)

[0071] in, For the Dynamic weighting of frame radar data, For the Dynamic weights of frame visual data, For the Dynamic weighting of frame radio data;

[0072] And satisfy:

[0073] (twenty two);

[0074] The fusion formula of the multi-source sensor data in step S24 is as follows:

[0075] (twenty three).

[0076] Preferably, step S3 specifically includes the following steps:

[0077] S31, obtaining the boundary points of the drone cluster based on the multi-source information fusion positioning results of the drone cluster, calculating the center of mass of the drone cluster, and analyzing the movement law of the drone cluster;

[0078] S32. Determine the stability of the drone cluster based on the boundary points of the drone cluster.

[0079] Preferably, the calculation formula of the centroid of the drone cluster described in step S31 is as follows:

[0080] (twenty four)

[0081] (25)

[0082] (26)

[0083] in, express The centroid of the direction, Indicates A drone on the projection surface coordinate, represents the center of mass in the Y direction, Indicates A drone on the projection surface coordinate, Indicates the position of the drone on the projection surface Number of racks, represents the centroid of the real position of the projection surface, represents the centroid accuracy, Represents the multi-representation sensor fusion centroid.

[0084] Preferably, step S32 specifically includes the following steps:

[0085] The stability index of the UAV cluster plane configuration is defined as , Stability Threshold , Stability Labeling , then the stability judgment formula is as follows:

[0086] (27)

[0087] in, express The area enclosed by the boundary nodes at the moment, express The area enclosed by the boundary nodes at the moment, To set the threshold.

[0088] The present invention has the following beneficial effects:

[0089] By fusing multi-source sensor information to achieve the positioning of drone clusters and determine the stability of drone clusters, we can effectively avoid the problem of cluster positioning failure caused by complex environments and improve the system's ability to locate drone clusters.

[0090] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 It is a logic flow chart of the method for determining the stability of the UAV cluster configuration based on multi-source information fusion positioning of the present invention;

[0092] Figure 2 It is a schematic diagram of the actual position of the arrow-shaped formation UAV cluster moving at a uniform speed in the simulation example of the present invention;

[0093] Figure 3 This is a result diagram of positioning the arrow-shaped formation cluster by only the radar sensor in the simulation example of the present invention;

[0094] Figure 4 This is a result diagram of positioning the arrow formation cluster by only the radio sensor in the simulation example of the present invention;

[0095] Figure 5 This is a result diagram of the positioning of the arrow formation cluster by only the visual sensor in the simulation example of the present invention;

[0096] Figure 6 It is a three-dimensional perspective diagram of cluster positioning after the fusion of three types of sensor information in the simulation example of the present invention;

[0097] Figure 7 It is a graph showing the change of positioning error over time of cluster positioning after the fusion of three types of sensor information in the simulation example of the present invention;

[0098] Figure 8 It is a plane configuration diagram of an arrow-shaped formation obtained after multi-sensor fusion of a simulation example of the present invention;

[0099] Fig. 9 It is a centroid distribution diagram of the projection surface obtained by real and multi-sensor fusion at different moments of the simulation example of the present invention;

[0100] Fig.10 It is the X-direction centroid motion diagram obtained by real and multi-sensor fusion at different moments of the simulation example of the present invention;

[0101] Fig.11 It is a distribution diagram of stability index and stability mark at different moments of the simulation example of the present invention. DETAILED DESCRIPTION

[0102] In order to make the purpose, technical scheme and advantages disclosed in the embodiments of the present invention clearer, the embodiments of the present invention are further described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not intended to limit the embodiments of the present invention. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions.

[0103] It should be noted that the terms "include", "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or devices.

[0104] Like reference numerals and letters denote similar items in the following drawings, and thus, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0105] like Figure 1 As shown, the method for judging the stability of the UAV cluster configuration based on multi-source information fusion positioning includes the following steps:

[0106] S1. Perform state matching based on multi-source sensor data to achieve data association of multi-source sensor data;

[0107] Step S1 specifically includes the following steps:

[0108] S11. Collect multi-source sensor data using multiple types of sensors, where the multi-source sensor data includes radar data, visual data, and radio data;

[0109] The multiple types of sensors in step S11 include radar sensors, optical sensors and radio sensors;

[0110] The radar data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space:

[0111] (1)

[0112] in, Indicates the number of drones detected by the radar sensor; For the The three-dimensional data results of the positioning of the UAV;

[0113] The visual data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space:

[0114] (2)

[0115] in, Indicates the number of drones detected by the optical sensor; For the The three-dimensional data results of the positioning of the UAV;

[0116] The radio data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space:

[0117] (3)

[0118] in, Indicates the number of drones detected by the radio sensor; For the The three-dimensional data results of the positioning of the UAV;

[0119] S12, taking the position state information of one type of sensor among the multiple types of sensors as a benchmark, traversing the remaining types of sensor data in pairs by using the nearest neighbor method;

[0120] Step S12 specifically includes the following steps:

[0121] The first drone data detected by radar sensor for drone clusters As a benchmark, the cluster results are located in the optical sensor using the minimization function Medium Search Distance The nearest point, denoted as point :

[0122] (4)

[0123] in, , when the minimum value is found, the corresponding Denote the midpoint between the optical sensor and the radar sensor Associated points , at this time:

[0124] (5)

[0125] At the same time, the minimization function is used to locate the cluster results of radio sensors Medium Search Distance The nearest point, denoted as point :

[0126] (6)

[0127] in, , when the minimum value is found, the corresponding Denote the midpoint between the radio sensor and the radar sensor Associated points , at this time:

[0128] (7);

[0129] S13, based on the data of multiple types of sensors obtained by traversal, determine whether they meet the distance constraint conditions; if they do, determine that the multi-source sensor data is successfully associated; otherwise, determine that part of the sensor data is successfully associated, and the data association ends.

[0130] Step S13 specifically includes the following steps:

[0131] Add distance constraint to formula (4):

[0132] (8)

[0133] in, The distance constraint threshold is set. Only the minimum value within this range is considered to meet the requirements. Otherwise, it is considered that the positioning result of the optical sensor cluster is There is no midpoint with the radar sensor Associated points If the positioning result of the radio sensor cluster is There are related points , it is assumed that there are only two sets of associated data in the data association stage, and only two sets of data are fused in the subsequent data fusion stage. There is no connection point , then the data that can be used in the fusion stage is only the midpoint of the radar sensor It can be seen that if the distance constraint condition is met, all sensor data are successfully associated. If not, only part of the sensor data is successfully associated. So far, the data association based on state matching is completed.

[0134] S2, by initializing multi-source sensor data, and then using two adjacent frames of multi-source sensor data to calculate the error, and then confirm the dynamic weight, to achieve multi-source information fusion positioning results based on multi-source sensor data;

[0135] Multi-source data fusion can only be performed after data association is completed. When multi-source data is fused, the first frame of data must be initialized first; then starting from the second frame of data, the error of the second frame of data is calculated based on the fusion result obtained by fusion of the first frame of data, and the multi-source data fusion weight is adjusted dynamically in real time according to the error. The third frame of data is then continuously iterated, and the weight of each frame of data is different when data is fused, thereby improving the accuracy of multi-source data fusion.

[0136] Step S2 specifically includes the following steps:

[0137] S21, initializing the first frame of sensor data of multiple types of sensors;

[0138] Step S21 specifically includes the following steps:

[0139] S211, let the error of radar sensor in three-dimensional coordinates be , the error of the optical sensor in three-dimensional coordinates is (Indicator error converted to three-dimensional coordinates), the error of the radio sensor in three-dimensional coordinates is (Indicator error is converted into three-dimensional coordinates), satisfying the following relationship:

[0140] (9)

[0141] in, is the proportionality factor, which is used to adjust the error weight to keep the probability sum equal to 1;

[0142] S212, the weight of radar sensor data fusion is , the weight of optical sensor data fusion is , the weight of radio sensor data fusion is ; then:

[0143] (10)

[0144] (11)

[0145] (12)

[0146] The weights and errors are equal to 1:

[0147] (13)

[0148] Get the fusion result of the first frame of sensor data:

[0149] (14)

[0150] in, It is the fusion result of the first frame of sensor data.

[0151] S22, calculating the position state information error using sensor data of two adjacent frames of the same type of sensor;

[0152] Step S22 specifically includes the following steps:

[0153] From the second frame sensor data Start (there can be multiple frames of data in one second), ,in Indicates Frame data, using the previous frame sensor data fusion result As a benchmark, calculate the The data error of the frame sensor data in the three-dimensional coordinates, the error is defined as follows:

[0154] (15)

[0155] (16)

[0156] (17)

[0157] in, is the error of the radar sensor in three-dimensional coordinates, is the error of the optical sensor in three-dimensional coordinates, is the error of the radio sensor in three-dimensional coordinates, , , They represent the radar sensor, optical sensor and radio sensor after data association in the Frame sensor data, It is the sensor data fusion result of the previous frame.

[0158] S23, calculating a dynamic weight based on the position state information error;

[0159] The dynamic weight calculation process in step S23 is as follows:

[0160] The following formula is used to obtain the Scaling factor for frame sensor data :

[0161] (18)

[0162] Get the first Dynamic weighting of frame sensor data:

[0163] (19)

[0164] (20)

[0165] (twenty one)

[0166] in, For the Dynamic weighting of frame radar data, For the Dynamic weights of frame visual data, For the Dynamic weighting of frame radio data;

[0167] And satisfy:

[0168] (twenty two);

[0169] S24, fusing multi-source sensor data according to dynamic weights to obtain a multi-source information fusion positioning result;

[0170] The fusion formula of the multi-source sensor data in step S24 is as follows:

[0171] (twenty three).

[0172] S25, determining whether the sensor data of all frames have achieved information fusion, if so, obtaining the positioning result after fusion of multi-source sensor data; if not, returning to step S24.

[0173] S3. Determine the stability of the UAV cluster based on the boundary points of the multi-source information fusion positioning results of the UAV cluster.

[0174] Step S3 specifically includes the following steps:

[0175] S31, obtaining the boundary points of the drone cluster based on the multi-source information fusion positioning results of the drone cluster, calculating the center of mass of the drone cluster, and analyzing the movement law of the drone cluster;

[0176] The calculation formula of the centroid of the drone cluster described in step S31 is as follows:

[0177] (twenty four)

[0178] (25)

[0179] (26)

[0180] in, express The centroid of the direction, Indicates A drone on the projection surface coordinate, represents the center of mass in the Y direction, Indicates A drone on the projection surface coordinate, Indicates the position of the drone on the projection surface Number of racks, represents the centroid of the real position of the projection surface, represents the centroid accuracy, Represents the multi-representation sensor fusion centroid.

[0181] Based on the above analysis of the movement law of the drone cluster: Since the center of mass is the average value of the positions of all drones in the drone cluster, it can be used to represent the overall position of the drone cluster. Therefore, the position information of the drones can be used to calculate the center of mass of the drone cluster.

[0182] The overall movement trend of the drone cluster can be shown by plotting the position of the center of mass on the timeline (plotting the trajectory of the center of mass changing over time).

[0183] Based on the trajectory diagram of the center of mass of the drone cluster changing over time, we can analyze the shape, direction and speed changes of the center of mass trajectory, and then understand the movement laws of the drone cluster and the overall movement characteristics of the cluster, such as whether the cluster's movement state is uniform, accelerating, decelerating, etc.

[0184] S32. Determine the stability of the drone cluster based on the boundary points of the drone cluster.

[0185] Step S32 specifically includes the following steps:

[0186] The stability index of the UAV cluster plane configuration is defined as (The stability of the UAV cluster plane configuration refers to the ability to work together within a certain area in space) , Stability Labeling , then the stability judgment formula is as follows:

[0187] (27)

[0188] in, express The area enclosed by the boundary nodes at the moment, express The area enclosed by the boundary nodes at the moment, Formula (27) indicates that when the ratio of the difference between the area enclosed by the boundary nodes at this moment and the previous moment to the area enclosed by the boundary nodes at the previous moment is within the set threshold Inside, If the judgment is 1, the drone cluster is determined to be stable and the stability judgment ends.

[0189] Simulation Example

[0190] This experiment is simulated and verified based on Matlab software.

[0191] At the same time, combined with the characteristics of drone clusters, the common cluster formation form is used for collective modeling. All moving nodes in the definition model are active in a three-dimensional area with a given length of 7km, width of 3km, and height of 10km. The relevant parameters are defined as follows:

[0192] The position X of the drone is a three-dimensional coordinate, which is defined as an activity within a three-dimensional area of ​​7km*3km*10km.

[0193] The speed V of the drone is a three-dimensional vector ranging from 102km / h to 130km / h.

[0194] Danger radius D, which is the minimum Euclidean distance of related drones that are too close to other drones, and its value is 0.3km;

[0195] Since different geometric formations are suitable for different mission scenarios, the arrow formation is suitable for breaking through the enemy area and carrying out strike missions, so the arrow formation is selected in this simulation example. Figure 2 As shown in the figure, a swarm of 30 drones adopts a two-layer arrow formation and moves forward at a uniform route speed of 102km / h. The height of each layer is no higher than 9100m, and the simulation time is 50s.

[0196] In this simulation example, the safe distance between two drones in the drone cluster is 300m, and the distance constraint can be taken as .

[0197] like Figure 3 As shown in the figure, the radar sensor is located at the coordinate (0, 0, 1), in km, simulating the radar sensor carried by the actual base station. The measurement range of the radar sensor is 360° in azimuth and 0-60° in elevation. The radar measurement distance accuracy index constraint is 20m. On the basis of the 20m accuracy index constraint of the measurement distance, the azimuth and elevation angles are constrained by 2° accuracy. The three-dimensional coordinates of the positioning result are generated based on the measurement distance, elevation angle and elevation angle.

[0198] The ambiguity problem of radar positioning: The resolution distance is limited to ≤30m, azimuth ≤2°, and pitch ≤2. It is believed that when the pitch angle or azimuth angle is less than or equal to 2°, the radar cannot separate the two targets and will regard the two targets as one target. During simulation, the positioning result of this target will be reflected by the average value of the two targets. The cluster individuals cannot be completely located, and the ambiguity problem is caused by the resolution problem.

[0199] Depend on Figure 3 It can be seen that if the arrow-shaped drone cluster is positioned only by radar sensors, due to the noise error and resolution of the radar sensor of only 2°, it may not be possible to locate the complete drone cluster. In this case, only 11 drones can be located, which is far from meeting the requirements.

[0200] like Figure 4As shown in the figure, the radio sensor is located at the coordinate (0, 0, 1), in km, simulating the radio sensor carried by the actual base station. The measurement range of the radio sensor is 360° in azimuth and 0-60° in elevation. The error of the radio sensor is mainly converted through the direction finding accuracy. By using the relationship between the radius, polar angle and azimuth angle in the spherical coordinate system of the point and the three-dimensional coordinates, the direction finding accuracy of 3° is added to the direction finding angle to obtain the positioning result of the radio sensor. The 3° direction finding accuracy will result in an error of hundreds of meters in the three-dimensional position coordinates of the positioning result.

[0201] Depend on Figure 4 It can be seen that although radio measurement can observe each individual drone cluster in the arrow-shaped formation, the direction-finding accuracy is only 3°. When reflected in the three-dimensional position distance coordinates, the maximum error can be 433m, which is not convenient for cluster positioning and subsequent configuration judgment.

[0202] and Figure 2 The actual position of the representative drone is compared. Figure 4 The position of the drone cluster obtained by the radio detection represented has a large error of 400m, which exceeds the 300m interval between drones in the drone cluster. It has disrupted the configuration distribution of the drone cluster and does not reflect the actual situation of the drone cluster at all.

[0203] like Figure 5 As shown, two base stations equipped with visual sensors are set on the XY plane. , The location of the base station is determined by Four parameters determine the is the base station coordinate, It is the angle between the main optical axis of the visual sensor and the X-axis, ranging from 0 to 90 degrees. It is the angle between the main optical axis of the visual sensor and the Z axis, ranging from 0 to 90 degrees. The data after 3D reconstruction roughly retains the distribution characteristics of the arrow formation, but there is still a deviation from the actual position of the arrow formation.

[0204] like Figure 6 and Figure 7 As shown in the figure, data fusion is performed based on the original data of three sensors, namely, visual sensor, radio sensor and radar sensor. The original data of the three sensors have been matched using the state matching method. In the data fusion stage, the fusion result is obtained by using the multi-source data fusion method based on dynamic weights. It can be seen that the positioning error of the result after the fusion of the data based on visual sensor, radio sensor and radar sensor is smaller, and the average positioning error of 50 seconds is 7.16%, which improves the ability of cluster positioning.

[0205] like Figure 8As shown in the figure, it can be observed that the boundary point recognition algorithm finds the convex polygon that envelops all drone points and identifies the plane configuration of the drone cluster. The plane configuration of the arrow-shaped formation can reflect the motion law of the corresponding drone cluster in three-dimensional space to a certain extent.

[0206] like Fig. 9 As shown in the figure, the actual position of the arrow formation in the Y-axis direction of the projection plane is 1.5 km, and the centroid accuracy of the multi-source sensor fusion at each moment is within the range of 0.9~1.1; the X-axis actual position direction at different moments corresponds to The equal intervals indicate that it is a uniform motion.

[0207] like Fig.10 As shown, the slope of the real position has not changed, indicating that its acceleration is 0. At the same time, the average speed of the cluster during this period can be obtained based on the slope.

[0208] like Fig.11 As shown in the figure, the threshold is set to 20%. According to the stability definition of the UAV cluster planar configuration, it is easy to know that every moment corresponding to the real position is a stable configuration, and the arrow-shaped formation flag is equal to 1 most of the time.

[0209] By analyzing the simulation example results, it can be concluded that the method for judging the stability of UAV cluster configuration based on multi-source information fusion positioning described in the present invention has a good effect.

[0210] Therefore, the present invention adopts the above-mentioned method for judging the configuration stability of a drone cluster based on multi-source information fusion positioning. By fusing multi-source sensor information, the positioning of the drone cluster and the judgment of the stability of the drone cluster can be realized, which can effectively avoid the problem of cluster positioning failure caused by complex environment and improve the system's ability to locate the drone cluster.

[0211] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for judging the stability of UAV cluster configuration based on multi-source information fusion positioning, characterized by: The following steps are involved: S1. Perform state matching based on multi-source sensor data to achieve data association of multi-source sensor data; Step S1 specifically includes the following steps: S11. Collect multi-source sensor data using multiple types of sensors, where the multi-source sensor data includes radar data, visual data, and radio data; S12, taking the position state information of one type of sensor among the multiple types of sensors as a reference, traversing the remaining types of sensor data in pairs by using the nearest neighbor method; S13, based on the data of multiple types of sensors obtained by traversal, determine whether they meet the distance constraint condition; if they do, determine that the multi-source sensor data is successfully associated; otherwise, determine that some sensor data is successfully associated, and the data association ends; The multiple types of sensors in step S11 include radar sensors, optical sensors and radio sensors; The radar data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space: （1） in, Indicates the number of drones detected by the radar sensor; For the The three-dimensional data results of the positioning of the UAV; The visual data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space: （2） in, Indicates the number of drones detected by the optical sensor; For the The three-dimensional data results of the positioning of the UAV; The radio data described in step S11 is: At this moment, the three-dimensional data of the positioning of the drone cluster The results are stored in matrix space: （3） in, Indicates the number of drones detected by the radio sensor; For the The three-dimensional data results of the positioning of the UAV; Step S12 specifically includes the following steps: The first drone data detected by radar sensor for drone clusters As a benchmark, the cluster results are located in the optical sensor using the minimization function Medium Search Distance The nearest point, denoted as point : （4） in, , when the minimum value is found, the corresponding Denote the midpoint between the optical sensor and the radar sensor Associated points , at this time: （5） At the same time, the minimization function is used to locate the cluster results of radio sensors Medium Search Distance The nearest point, denoted as point : （6） in, , when the minimum value is found, the corresponding Denote the midpoint between the radio sensor and the radar sensor Associated points , at this time: （7）； Step S13 specifically includes the following steps: Add distance constraint to formula (4): （8） in, is the distance constraint threshold set; S2, by initializing multi-source sensor data, and then using two adjacent frames of multi-source sensor data to calculate the error, and then confirm the dynamic weight, to achieve multi-source information fusion positioning results based on multi-source sensor data; Step S2 specifically includes the following steps: S21, initializing the first frame of sensor data of multiple types of sensors; Step S21 specifically includes the following steps: S211, let the error of radar sensor in three-dimensional coordinates be , the error of the optical sensor in three-dimensional coordinates is , the error of the radio sensor in three-dimensional coordinates is , satisfying the following relationship: （9） in, is the proportionality coefficient; S212, the weight of radar sensor data fusion is , the weight of optical sensor data fusion is , the weight of radio sensor data fusion is ; then: （10） （11） （12） The weights and errors are equal to 1: （13） Get the fusion result of the first frame of sensor data: （14） in, is the fusion result of the first frame of sensor data; S22, calculating the position state information error using sensor data of two adjacent frames of the same type of sensor; Step S22 specifically includes the following steps: From the second frame sensor data start, ,in Indicates Frame data, using the previous frame sensor data fusion result As a benchmark, calculate the The data error of the frame sensor data in the three-dimensional coordinates, the error is defined as follows: （15） （16） （17） in, is the error of the radar sensor in three-dimensional coordinates, is the error of the optical sensor in three-dimensional coordinates, is the error of the radio sensor in three-dimensional coordinates, , , They represent the radar sensor, optical sensor and radio sensor after data association in the Frame sensor data, It is the sensor data fusion result of the previous frame; S23, calculating a dynamic weight based on the position state information error; The dynamic weight calculation process in step S23 is as follows: The following formula is used to obtain the Scaling factor for frame sensor data : （18） Get the first Dynamic weighting of frame sensor data: （19） （20） （21） in, For the Dynamic weighting of frame radar data, For the Dynamic weights of frame visual data, For the Dynamic weighting of frame radio data; And satisfy: （22）； S24, fusing multi-source sensor data according to dynamic weights to obtain a multi-source information fusion positioning result; The fusion formula of the multi-source sensor data in step S24 is as follows: （23）； S25, determining whether the sensor data of all frames have achieved information fusion, if so, obtaining the positioning result after fusion of multi-source sensor data; if not, returning to step S24; S3, judging the stability of the drone cluster based on the boundary points of the multi-source information fusion positioning results of the drone cluster; Step S3 specifically includes the following steps: S31, obtaining the boundary points of the drone cluster based on the multi-source information fusion positioning results of the drone cluster, calculating the center of mass of the drone cluster, and analyzing the movement law of the drone cluster; S32. Determine the stability of the drone cluster based on the boundary points of the drone cluster.

2. The method for determining the stability of a UAV cluster configuration based on multi-source information fusion positioning according to claim 1 is characterized in that: The calculation formula of the centroid of the drone cluster described in step S31 is as follows: （24） （25） （26） in, express The centroid of the direction, Indicates A drone on the projection surface coordinate, represents the center of mass in the Y direction, Indicates A drone on the projection surface coordinate, Indicates the position of the drone on the projection surface Number of racks, represents the centroid of the real position of the projection surface, represents the centroid accuracy, Represents the multi-representation sensor fusion centroid.

3. The method for determining the stability of a UAV cluster configuration based on multi-source information fusion positioning according to claim 2 is characterized in that: Step S32 specifically includes the following steps: The stability index of the UAV cluster plane configuration is defined as , Stability Threshold , Stability Labeling , then the stability judgment formula is as follows: （27） in, express The area enclosed by the boundary nodes at the moment, express The area enclosed by the boundary nodes at the moment, To set the threshold.

Images