A multi-sensor cross cue target tracking method and device

Abstract

The application discloses a multi-sensor cross prompting target tracking method and device, obtains working state information of all sensors, obtains a first number of sensors in a working state; compares the first number with a sensor working number threshold value, selects a sensor group according to a comparison result; obtains target detection information output by each sensor in the sensor group; fuses all target detection information to obtain target fused detection information; determines final target state information according to the target fused detection information; the application can coordinate and manage multiple sensors, avoids all sensors working at the same time, and can realize fast and accurate tracking of a target, improves working efficiency of the sensors; meanwhile, a weighted fusion method is adopted to fuse target detection information of the sensor group, more accurate final target state information is obtained, and the system has better target tracking performance.

Description

Technical Field

[0001] This invention belongs to the field of mobile target tracking, specifically relating to a multi-sensor cross-caution target tracking method and apparatus. Background Technology

[0002] Maneuvering target tracking technology utilizes sensor measurements combined with filtering algorithms to estimate target status, enabling modern defense systems to provide timely early warning and engagement. However, in modern warfare, as the performance of air attack weapons continues to improve, the detection range and reaction speed of defense systems are gradually decreasing, making it impossible for systems to react to targets in a timely, continuous, and effective manner. Therefore, how to efficiently utilize detection and defense methods in conjunction with target tracking technology to achieve early warning and tracking of targets is a current research hotspot in air defense systems.

[0003] Currently, the most effective solution for defending against air attacks is to establish a layered defense system. This system includes the comprehensive use of various early warning and detection devices to achieve timely and long-range target detection; the implementation of multi-layered interception to effectively defend against low-altitude / stealth targets; and the use of multi-sensor cross-caution technology to achieve continuous and high-probability target detection. Among these, multi-sensor cross-caution technology is a key technology for establishing an efficient detection and defense system, mainly because when a target attacks, a single sensor cannot accurately acquire all the information about the target; and when using multiple sensors, the limited sensor resources prevent them from working simultaneously.

[0004] For a mobile target tracking system that uses multi-sensor cooperative detection based on cross-cue technology, both sensor efficiency and target tracking performance are extremely important. However, most studies usually aim to improve the solution speed of the tracking algorithm in the system, but the solution quality is not improved. This results in technical problems such as insufficient target detection, low efficiency of multi-sensor operation, and easy loss of tracking signals. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-sensor cross-caution target tracking method and apparatus. By reasonably selecting the number of sensors in the working state, a sensor group is constructed and the working efficiency is improved, thereby improving the target tracking accuracy.

[0006] This invention adopts the following technical solution: a multi-sensor cross-caution target tracking method, comprising:

[0007] Obtain the operating status information of all sensors to get the initial number of sensors that are in operation;

[0008] The first quantity is compared with the sensor working quantity threshold, and the sensor group is selected according to the comparison result; wherein, the number of sensors in the sensor group is equal to the sensor working quantity threshold;

[0009] Acquire target detection information output by each sensor in the sensor group;

[0010] All target detection information is fused to obtain target fused detection information;

[0011] The final target status information is determined based on the target fusion detection information.

[0012] Furthermore, the sensor group selected based on the comparison results includes:

[0013] When the first quantity is not 0 and is less than the sensor working quantity threshold, calculate the first difference between the first quantity and the sensor working quantity threshold.

[0014] Select the sensors that correspond to the first difference number from the non-operating sensors and add them to the sensor group.

[0015] Furthermore, when the first quantity is 0, it includes:

[0016] Do not select the sensor to add to the sensor group;

[0017] When the first quantity is 0 and the first quantity at the previous moment is not 0, the sensor working quantity threshold is selected and added to the sensor group based on the final target state information at the previous moment.

[0018] Furthermore, selecting the sensor group based on the comparison results also includes:

[0019] When the first quantity is greater than the sensor working quantity threshold, select the sensor corresponding to the sensor working quantity threshold from the sensors that are in operation and add it to the sensor group.

[0020] Furthermore, selecting the sensor group based on the comparison results also includes:

[0021] When the first quantity equals the sensor working quantity threshold, all sensors in working state are added to the sensor group.

[0022] Furthermore, the methods for selecting sensors include:

[0023]

[0024]

[0025] Where P is the detection accuracy of the sensor array, D is the relative distance between the sensor array and the target, and p i Let r be the detection accuracy of the i-th sensor in the sensor group. i Let x be the detection distance of the i-th sensor in the sensor group. i ,y i ,z i(x, y, z) represents the coordinates of the i-th sensor in the sensor group; (x, y, z) represents the coordinates of the target; d i Let represent the relative distance between the i-th sensor in the sensor group and the target, and N represent the number of sensors in the sensor group.

[0026] Furthermore, the fusion of all target detection information includes:

[0027] Calculate the weighting factor for each sensor in the sensor group;

[0028] All target detection information is fused based on the weighting factor of each sensor.

[0029] Furthermore, the weighting factor is calculated as follows:

[0030]

[0031] Among them, w j R is the weighting factor for the j-th sensor in the sensor group. j Let be the observation noise covariance of the j-th sensor in the sensor group.

[0032] Furthermore, the final target state information determined based on the target fusion detection information includes:

[0033] The target fusion detection information is used as the observation information, and the interactive multi-model unscented Kalman filter method is used to estimate the target state information to obtain the final target state information.

[0034] Another technical solution of the present invention: a multi-sensor cross-caution target tracking device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned multi-sensor cross-caution target tracking method.

[0035] The beneficial effects of this invention are: by constructing a sensor group, this invention can coordinate and manage multiple sensors, avoid all sensors working at the same time, and achieve fast and accurate tracking of targets, thereby improving the working efficiency of the sensors; at the same time, by using a weighted fusion method to fuse the target detection information of the sensor group, more accurate final target state information is obtained, thereby ensuring that the system has better target tracking performance. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the multi-type, multi-level sensor collaborative detection system constructed in an embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram of the flight trajectories of enemy aircraft and early warning aircraft and the distribution of radar in an embodiment of the present invention;

[0038] Figure 3 This is a comparison diagram of the real-time sensor response between the method of this invention and the target tracking method without cross-clues.

[0039] Figure 4 This is a comparison chart of the root mean square error of the position of the method in this embodiment of the invention and the target tracking method without cross-cutting clues;

[0040] Figure 5 This is a comparison chart of the root mean square error of the speed of the method in this embodiment of the invention and the target tracking method without cross-cutting clues;

[0041] Figure 6 This is a comparison chart of the mean root mean square errors of position and velocity between the method of this invention and the target tracking method without cross-cutting prompts. Detailed Implementation

[0042] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0043] This invention discloses a multi-sensor cross-caution target tracking method, comprising: acquiring the working status information of all sensors to obtain a first number of sensors in working status; comparing the first number with a sensor working number threshold, and selecting a sensor group based on the comparison result; wherein the number of sensors in the sensor group is equal to the sensor working number threshold; acquiring target detection information output by each sensor in the sensor group; fusing all target detection information to obtain target fused detection information; and determining the final target status information based on the target fused detection information.

[0044] This invention constructs a sensor group, which can coordinate and manage multiple sensors, avoid all sensors working at the same time, and achieve fast and accurate target tracking, thus improving the working efficiency of the sensors. At the same time, a weighted fusion method is used to fuse the target detection information of the sensor group to obtain more accurate final target state information, thereby ensuring that the system has better target tracking performance.

[0045] In this embodiment of the invention, a multi-type, multi-level sensor collaborative detection system is established, which enables the system to comprehensively utilize the detection information of the sensors to achieve early warning, accurate detection and tracking of aerial targets, while facilitating the air defense system to quickly deploy defensive measures.

[0046] First, multiple types of sensor detection equipment are selected. Ground-based platform sensors are susceptible to the effects of the Earth's curvature and terrain, resulting in detection blind spots. Furthermore, their fixed positions make them vulnerable to interference or attacks from the opposing side, leading to a decrease in sensor detection accuracy. In contrast, aerial platform sensor equipment is located at high altitudes and is mobile, which can overcome the shortcomings of ground-based sensors to some extent. It has a wider detection range and operational range, which is beneficial for extending early warning time and implementing guidance and command. Therefore, this embodiment of the invention mainly uses different combinations of sensors from both land and aerial platforms.

[0047] For land platforms, radar is the preferred sensor. Radar is a sensor capable of detecting distant targets, offering the advantages of all-weather, all-time operation. Based on its application, it is mainly divided into four types: long-range surveillance radar, search and surveillance radar, guidance and command radar, and altimeter radar.

[0048] The formula for radar detection range is as follows:

[0049]

[0050] Where, ρ max P is the maximum detection range of the radar in free space. t G is the pulse power radiated by the antenna, A is the effective area of ​​the antenna aperture, σ is the radar cross section, and P is the directional gain of the antenna. min It is the minimum signal power that the receiver can receive.

[0051] As shown in equation (1), the radar pulse power increases with the detection distance, which in turn increases the local energy supply pressure, resulting in a decrease in detection accuracy (i.e., the two are inversely proportional). Therefore, a longer detection distance is not always better when selecting a radar network. To ensure that the system achieves high measurement accuracy while realizing long-range detection, this embodiment uses a combination of search and early warning radar and altimeter radar to achieve long-range detection of aerial targets; at the same time, a guidance and command radar is used to obtain more accurate target information at close range.

[0052] In the air, early warning aircraft are used to detect targets. In this invention, the target refers to the invading Blue Force aircraft, which can be understood as Blue Force planes. The early warning aircraft places the entire radar system on the Red Force aircraft, increasing the height of the radar antenna, which can improve the radar search range and detection distance, providing better early warning and detection effects for the system. At the same time, after detecting a target, the early warning aircraft can continuously track the target throughout its flight process, increasing the early warning time; in wartime, it can also command and guide various friendly air defense weapons to carry out combat missions.

[0053] Secondly, a multi-layered sensor detection system should be established, and the geometric distribution of the detection equipment should be determined. Since targets can attack important facilities of the Red Force at any time along the direction of attack from outside the defense zone, multi-layered interception must be implemented to ensure that the system completes its air defense mission as early as possible.

[0054] To provide early warning of targets and maximize the detection capabilities of multiple sensors, this invention constructs a two-layered sensor detection system based on the target's trajectory, the types and performance of the detection equipment used. The first layer consists of search and surveillance radar and altimeter radar, primarily responsible for long-range detection and early warning, providing early warning and buying time for subsequent interception operations. The second layer consists of guidance and command radar, arranged in an equilateral triangular array according to the locations of important facilities and the target's initial position, primarily responsible for progress point defense detection, ensuring detection accuracy and effectiveness. Furthermore, an early warning aircraft provides continuous and reliable target indication information to the Red Force's detection system, extending the warning time.

[0055] In summary, the multi-type, multi-level sensor collaborative detection system structure constructed in the embodiments of the present invention is as follows: Figure 1 As shown.

[0056] A dynamic alliance model for multi-sensor cross-caution is established based on comprehensive detection requirements and system resources (i.e., the sensor group in this invention). The cross-caution mechanism is implemented according to the determined objective function and constraints, thereby improving the working efficiency of the sensors and ensuring the continuous visibility of the target.

[0057] Assume a sensor cooperative detection system has M sensors {S1, S2, ..., S}. M The target is detected by a sensor alliance. The number of sensors in the sensor group is pre-set to N (N≤M), and the actual number of sensors responding at each time step is n (i.e., the number of sensors responding at a given time when the method of this invention is not used). The objective of the sensor alliance in this invention is to achieve the highest detection accuracy P and the lowest relative distance D for the target by the multi-sensor detection network. Therefore, the dynamic alliance model can be abstracted into a problem of solving a multi-objective function optimization model. In other words, the sensor selection method in this invention includes:

[0058]

[0059]

[0060] Where P is the detection accuracy of the sensor array, D is the relative distance between the sensor array and the target, and p i Let r be the detection accuracy of the i-th sensor in the sensor group. i Let x be the detection distance of the i-th sensor in the sensor group. i ,y i ,z i(x, y, z) represents the coordinates of the i-th sensor in the sensor group; (x, y, z) represents the coordinates of the target; d i Let represent the relative distance between the i-th sensor in the sensor group and the target, and N represent the number of sensors in the sensor group.

[0061] Based on the sensor selection method described above, the selection of sensors at various times is divided into the following cases:

[0062] When the first quantity is not 0 and is less than the sensor working quantity threshold, the first difference between the first quantity and the sensor working quantity threshold is calculated; among the non-working sensors, the sensor corresponding to the first difference quantity is selected and added to the sensor group. That is, when n < N, the number of responding sensors is less than the number of sensor alliances. The system will use a cross-prompt mechanism, causing responding sensors to sequentially prompt non-responding sensors with a relative distance d. i The smallest sensor is pointed at the target, preparing for collaborative detection, that is:

[0063]

[0064] in, This indicates the i-th sensor that has responded. This indicates the sensor that is the i-th closest to the target among the unresponsive sensors.

[0065] More specifically, when the first quantity is 0, it includes: not selecting a sensor to join the sensor group.

[0066] In this case, when k=1, k is the time sequence number, and the operating states of each sensor are initialized by letting:

[0067] UnionSensors(k,1:N)=0 (5)

[0068] UnionSensors(k,1:N) represents the N responding sensors in the sensor alliance at time k. 0 indicates that no sensor responds. This state represents the first moment after the system starts. At this time, all sensors do not respond because the blue target has not been detected. At this time, there is no need to select a sensor group. Therefore, there are 0 responding sensors in the sensor alliance.

[0069] Furthermore, if none of the sensors detect the target at times k-1 and k, the sensor consortium remains consistent with its state at the previous time step, i.e.:

[0070] UnionSensors(k,1:N)=UnionSensors(k-1,1:N) (6)

[0071] At this point, it indicates that no sensor responded in the previous moments, meaning that no target was detected. Therefore, the number of sensors in the sensor consortium is still 0.

[0072] Another scenario involves a sensor group where the first quantity is 0 and the first quantity at the previous time step was not 0. The sensor group is then selected based on the final target state information from the previous time step, using a threshold for the number of working sensors. That is, if n≠0 at time k-1 and n=0 at time k, the target is lost. In this case, the system should prompt the other N sensors that have not responded and are at the closest relative distance to search for the target's possible movement range. Simultaneously, a tracking model should be used to predict the target's trajectory. If target information is still not obtained after a certain time (e.g., 100 seconds), initialization is restored. The corresponding sensor alliance is:

[0073]

[0074] in, The i-th sensor (i.e., the i-th sensor sorted from smallest to largest) is the one that did not respond in the previous time step.

[0075] Additionally, when the initial number equals the threshold for the number of active sensors, all active sensors are added to the sensor group. That is, when n = N, the number of responding sensors equals the number of sensor consortia. The sensor consortium is quickly built and begins operation, i.e.:

[0076]

[0077] Finally, there's another scenario: when the initial number exceeds the sensor operational quantity threshold, a sensor corresponding to the threshold is selected from the operational sensors and added to the sensor group. That is, when n > N, the number of responding sensors exceeds the target sensor alliance number. To conserve sensor resources while maintaining detection accuracy, suitable sensors need to be selected for collaborative detection. Assuming the sensor detection accuracy weight α = 0.8 and the relative distance weight β = 0.2, the overall detection effect η is obtained by combining the objective function. i (i = 1…n), the sensor alliance is obtained as follows:

[0078]

[0079] in, The i-th sensor is used to comprehensively detect the effects.

[0080] Thus, the establishment of the multi-sensor cross-caution optimization model and sensor resource management are complete. When a visible target is lost, the sensor alliance prompts the nearest sensor that has not yet responded to search the target's flight area sequentially. This reduces sensor detection cycle time and allows for faster target location acquisition. When the number of responding sensors is less than the number of sensors in the alliance, the nearest sensor that has not yet responded is also prompted to stand by, ensuring continuous visibility of the target during sensor handover. If the number of responding sensors exceeds the number of sensors in the alliance, simultaneous responses from multiple sensors would be wasteful. Therefore, by considering multiple objective functions, sensors with high detection accuracy and large absolute relative distances are selected to operate. This ensures high tracking accuracy and keeps the target visible for a period of time. Based on the above analysis, the cross-caution technique used in the sensor alliance process not only ensures continuous target visibility but also improves sensor efficiency, thus enhancing the overall performance of the multi-sensor collaborative detection system.

[0081] Next, all target detection information needs to be fused. The specific methods include: calculating the weighting factor of each sensor in the sensor group; and fusing all target detection information based on the weighting factor of each sensor.

[0082] Specifically, an optimal weighted fusion method is used for detection information fusion. After the sensors complete the measurement of the target parameters, a certain sensor node is used as the fusion center to achieve detection information fusion of the sensor alliance under the minimum variance criterion, which is beneficial to improving the target tracking accuracy.

[0083] The observation equation for sensor j is:

[0084] Z j (k)=h(X(k))+V j (k) (10)

[0085] Where X(k) and Z j (k) represents the system state vector at time k and the observed value of sensor j, respectively; h(·) is the nonlinear function describing the observation equation; and V is the observation noise of sensor j. j (k) is zero-mean white Gaussian noise, satisfying V j (k)~N(V j (k); 0,R j ).

[0086] Assuming that the observations from each sensor are unbiased and independent, the fusion estimation of observation information... The corresponding noise covariance R can be expressed as:

[0087]

[0088]

[0089]

[0090] Among them, w j Let N be the weight of the observation data of sensor j, and N be the number of sensors responding at time k (i.e., the number of sensor alliances or the number of sensors in a sensor group).

[0091] To minimize the noise covariance of equation (12), an auxiliary function containing Lagrange multipliers is constructed:

[0092]

[0093] Differentiating the auxiliary function (14) yields:

[0094]

[0095] Solving equations (11)-(15) simultaneously yields the optimal weighting factor w. j for:

[0096]

[0097] Among them, w j R is the weighting factor for the j-th sensor in the sensor group. j Let be the observation noise covariance of the j-th sensor in the sensor group.

[0098] w j Substituting into equations (11)-(12) yields the optimal weighted fusion estimate of the observations. The corresponding noise covariance R is as follows:

[0099]

[0100]

[0101] In summary, having obtained the fused target detection information, the next step is to determine the final target state information based on the fused target detection information. Specifically, the fused target detection information is used as observation information, and the interactive multi-model unscented Kalman filter method is used to estimate the target state information to obtain the final target state information.

[0102] The fused observation results (i.e. fused target detection information) obtained in the above steps are substituted into the system equations for maneuvering target tracking as observations. The target state estimation results are obtained through filtering algorithms, thus realizing a target tracking method based on multi-sensor cross-caution.

[0103] For a maneuvering target tracking system, the system state equation and observation equation satisfy a nonlinear dynamic system with additive noise, as shown below:

[0104] X(k)=f(X(k-1))+W(k) (19)

[0105] Z(k)=h(X(k))+V(k) (20)

[0106] Where X(k)∈R n Z(k)∈R m Let f(·) and h(·) be the state vector and observation vector of the system at time k, respectively; f(·) and h(·) are the nonlinear functions describing the system state model and measurement model, respectively; and the process noise W(k)∈R. n And measurement noise V(k)∈R m The noise consists of uncorrelated, zero-mean white Gaussian noise with covariance matrices Q(k)∈R. n×n and R(k)∈R m×m .

[0107] Since the target is in a maneuvering state, and the Interactive Multi-Model Unscented Kalman Filter (IMM-UKF) algorithm can use multiple motion models to match different maneuvering states of the target, and can change the model used according to the current motion characteristics of the target, it has good adaptability. Therefore, this invention adopts IMM-UKF as the tracking algorithm for maneuvering targets.

[0108] Based on the nonlinear system described by equations (19)-(20), assuming that the model set in IMM-UKF consists of r mathematical models, the specific steps of the algorithm are as follows:

[0109] First, input interaction is performed to calculate the initial mixture state estimate of sub-filter j (j = 1…r). Covariance Estimation

[0110]

[0111]

[0112]

[0113]

[0114] in, Let μ be the state estimate and corresponding covariance of model i at time k-1. i (k-1) represents the probability of model i at time k-1. To calculate the normalization constant for the input interaction probability from model i to model j, p ij Let be the probability of transitioning from model i to model j.

[0115] Secondly, each model performs parallel filtering, The state estimate is obtained by UKF filtering as input. and the corresponding covariance P j (k|k).

[0116] Then, the likelihood function is constructed using the innovation vector of the sub-filter and the corresponding covariance matrix, and the probability μ of model j at time k is updated. j (k):

[0117]

[0118]

[0119]

[0120] Among them, Λ j (k) is the likelihood function of model j at time k, and c is the normalization constant used to calculate the probability of model j. Let P be the innovation vector of sub-filter j at time k. ZZ(j) (k) represents the corresponding information covariance, and n is the dimension of the measurement vector Z(k).

[0121] Finally, the states and covariances obtained from each sub-filter are weighted and summed to complete the output interaction, resulting in the mixed state estimate and covariance, calculated as follows:

[0122]

[0123]

[0124] In summary, this invention employs multi-sensor cross-caution technology in the target tracking system. First, it improves the working method of sensors in the air defense system by establishing a sensor alliance and its optimization model. This transforms the individual detection method of sensors into a collaborative detection method of the sensor alliance. Under the cross-caution mechanism, sensor resources can be saved and sensor efficiency can be improved while ensuring target tracking accuracy. At the same time, an optimal fusion algorithm is used to fuse sensor detection data before the target tracking algorithm, further improving the tracking performance of the system.

[0125] To further verify the technical effect of the method of the present invention, the following verification embodiments were also carried out.

[0126] First, set the initial dynamic parameters of the target under the following scenario:

[0127] A Cartesian coordinate system is established with the command center as the origin, and the x, y, and z axes representing east, north, and altitude, respectively, with a unit length of kilometers. Assume the coordinates of the important facility are (200, 700, 0); the enemy aircraft maneuvers towards the important facility from an initial position of (2000, 2500, 30) with an initial velocity of (0, -500 m / s, 0).

[0128] Assume the Red Force has eight detection sensors to detect targets. The sensor types corresponding to the sensor numbers are as follows: sensors 1 and 2 are search and surveillance radars; sensors 3 and 4 are altimeter radars; sensors 5, 6, and 7 are guidance and command radars; and sensor 8 is an early warning aircraft. The detection ranges of the search and surveillance radar, altimeter radar, and guidance and command radar are 700 km, 650 km, and 300 km, respectively, and their measurement noise covariances are R0, ... 1，2 =diag([10 2 0.03 2 ]) R 3,4 =0.5 2 R 5,6,7 =diag([2 2 0.001 2 1 2 The detection range of the early warning aircraft is 500km, and the measurement noise covariance R = diag([5 2 0.003 2 2 2 The red team, starting from coordinates (1000, 0, 10), sets off northward at a speed of (0, 230 m / s, 0) to patrol and protect the security of important facilities. The detection accuracy matrix is ​​[-10, -10, -10, -10, -2, -2, -2, -8], where the i-th element represents the detection accuracy of sensor i; the sensor alliance number is set to 3. The flight paths of enemy aircraft and early warning aircraft, and the radar distribution are as follows: Figure 2 As shown.

[0129] The model set in the filtering algorithm consists of a uniform motion model and a "current" statistical model, where the maximum acceleration 'a' in the "current" statistical model is... M =0.03m·s -2 The maneuver frequency α = 0.05; the Markov probability transition matrix P and the model probability u are respectively:

[0130]

[0131] u = [0.5 0.5] (31)

[0132] The initial states for filtering in the three axes are as follows: The initial state covariance is

[0133] Implement the target tracking method with cross-clues according to the above steps, and compare and analyze it with the target tracking algorithm without cross-clues under the same conditions. The simulation time is 2500s, and 100 Monte Carlo simulations are performed on the entire process. The root mean square error (RMSE) is used as the evaluation criterion, which is defined as:

[0134]

[0135]

[0136] Where M is the number of Monte Carlo simulations, k = 1, 2, ..., N′, N′ is the number of samples, and X i (k) represents the true value in the i-th Monte Carlo simulation. Let represent the estimated value in the i-th Monte Carlo simulation, and ARMSE be the root mean square error.

[0137] Figure 3 The response times of each sensor in target tracking algorithms with and without cross-caution are presented. It is evident that the target tracking algorithm without cross-caution experiences a brief system failure due to target loss within 1536-1599 seconds; furthermore, the method fails to acquire target height information promptly upon initial target detection; and multiple sensors operate simultaneously during certain time periods, leading to wasted sensor resources. In contrast, the system employing the cross-caution method of this invention allows unresponsive sensors to search near the target's movement range when the target is lost, and predicts the target model in gaps in multi-sensor coverage, thus achieving continuous target visibility. Furthermore, by utilizing only a subset of sensors at each time point, the system can reduce resource consumption while maintaining tracking performance.

[0138] It is worth noting that because the defense system is deployed before the use of cross-clueing technology in this invention, there are not many sensors near the protection facility. Therefore, the use of cross-clueing algorithm does not significantly reduce the overall resource consumption rate of the system. However, when there are many sensors near the protection facility, the saving effect is more obvious.

[0139] Figure 4 and Figure 5 The root mean square errors of position and velocity for target tracking algorithms with and without cross-clues are given respectively. Figure 6These are the ARMSE values ​​for position and velocity in the two methods. It is evident that the target tracking algorithm with cross-caution has higher overall accuracy. This is mainly because the method fuses data from multiple sensors to improve the accuracy of the observation data, thereby obtaining a more accurate target state estimation. Furthermore, due to the use of cross-caution technology with the objective function of improving system detection accuracy and minimizing the relative distance between sensors and the target, when the target is lost or the number of working sensors is less than the number of sensors in the coalition, the sensor coalition first prompts the sensor closest to the target, enabling the system to obtain target detection information in a timely manner, reducing sensor detection cycle time and improving target tracking performance.

[0140] This invention focuses on maneuvering target tracking and proposes a multi-sensor cross-caution technique for target tracking to improve the efficiency of multi-sensor operation and target tracking accuracy. First, the invention improves the operation of sensors in air defense systems by establishing a sensor alliance and its optimization model. This transforms the individual sensor detection method into a collaborative target detection method by the sensor alliance, saving sensor resources while ensuring target tracking accuracy. Second, an objective function is established to minimize the relative distance between the sensor and the target and maximize the sensor's detection accuracy, thereby improving sensor efficiency. Finally, the algorithm achieves sensor detection information fusion through optimal weighted fusion under the minimum variance criterion, further enhancing target tracking performance. This invention improves the sensor detection efficiency and tracking performance of maneuvering target tracking systems.

[0141] The present invention also discloses a multi-sensor cross-caution target tracking device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aforementioned multi-sensor cross-caution target tracking method.

[0142] It should be noted that the information interaction and execution process between the above-mentioned devices are based on the same concept as the method embodiments of the present invention. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0143] The device can be a computing device such as a desktop computer, laptop, handheld computer, radar, or cloud server. The device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that it may include more or fewer components, or a combination of certain components, or different components; for example, it may also include input / output devices, network access devices, etc.

[0144] The processor referred to can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0145] In some embodiments, the memory may be an internal storage unit of the extraction device, such as the hard drive or memory of the extraction device. In other embodiments, the memory may be an external storage device of the extraction device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the extraction device. Furthermore, the memory may include both internal storage units and external storage devices of the extraction device. The memory is used to store operating systems, applications, bootloaders, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.

Claims

1. A multi-sensor cross-caution target tracking method, characterized in that, include: Obtain the operating status information of all sensors to get the initial number of sensors that are in operation; The first quantity is compared with the sensor working quantity threshold, and a sensor group is selected based on the comparison result; wherein, the number of sensors in the sensor group is equal to the sensor working quantity threshold; Obtain the target detection information output by each sensor in the sensor group; All the target detection information is fused to obtain target fused detection information; The final target status information is determined based on the target fusion detection information; Based on the comparison results, the sensor group selected includes: When the first quantity is not 0 and is less than the sensor working quantity threshold, calculate the first difference between the first quantity and the sensor working quantity threshold. Select the sensors that correspond to the first difference number from the sensors that are not in operation and add them to the sensor group; The method for selecting the sensor includes: ， ， in, The detection accuracy of the sensor group. The relative distance between the sensor group and the target. The first in the sensor group The detection accuracy of each sensor, The first in the sensor group The detection range of each sensor, The first in the sensor group Sensor coordinates; These are the coordinates of the target; The first in the sensor group The relative distance between each sensor and the target, where N is the number of sensors in the sensor group.

2. The multi-sensor cross-caution target tracking method according to claim 1, characterized in that, When the first quantity is 0, it includes: Do not select the sensor to add to the sensor group; When the first quantity is 0 and the first quantity at the previous moment is not 0, the sensor working quantity threshold is selected and added to the sensor group according to the final target state information at the previous moment.

3. The multi-sensor cross-caution target tracking method according to claim 2, characterized in that, Selecting sensor groups based on comparison results also includes: When the first quantity is greater than the sensor working quantity threshold, the sensor corresponding to the sensor working quantity threshold is selected from the sensors in the working state and added to the sensor group.

4. The multi-sensor cross-caution target tracking method according to claim 3, characterized in that, Selecting sensor groups based on comparison results also includes: When the first quantity equals the threshold number of working sensors, all sensors in operation are added to the sensor group.

5. The multi-sensor cross-caution target tracking method according to claim 4, characterized in that, The fusion of all the target detection information includes: Calculate the weighting factor for each sensor in the sensor group; The target detection information is fused based on the weighting factor of each sensor.

6. The multi-sensor cross-caution target tracking method according to claim 5, characterized in that, The weighting factor is calculated as follows: ， in, Let be the weighting factor for the j-th sensor in the sensor group. Let be the observation noise covariance of the j-th sensor in the sensor group.

7. The multi-sensor cross-caution target tracking method according to claim 6, characterized in that, The final target state information determined based on the target fusion detection information includes: The target fusion detection information is used as observation information, and the target state information is estimated using the interactive multi-model unscented Kalman filter method to obtain the final target state information.

8. A multi-sensor cross-caution target tracking device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a multi-sensor cross-caution target tracking method as described in any one of claims 1-7.

Images