Multi-target radar time resource allocation method based on dynamic priority evaluation

Abstract

The invention discloses a multi-target radar time resource allocation method based on dynamic priority evaluation, and the method comprises the steps: building a target motion model and a radar observation model, and enabling the target motion model to describe a target motion process through employing a plurality of models containing a possible motion mode; obtaining a radar measurement value according to the target motion model and the radar observation model, and constructing a priority function including target motion characteristics, spatial positions and historical observation information; embedding the priority function into a Bayesian Cramer-Rao lower bound, and establishing a radar resource allocation model which takes a minimum weighted Bayesian Cramer-Rao lower bound sum as a target function, takes a dwell time sum not exceeding a resource scheduling period and takes each target dwell time meeting a constraint range as conditions; and solving the radar resource allocation model through a gradient projection method to obtain a resource allocation result. Reasonable configuration of radar residence time resources can be realized, and the target tracking precision and reliability are effectively improved.

Description

Technical Field

[0001] The invention belongs to the field of radar multi-target tracking, and in particular relates to a multi-target radar time resource allocation method based on dynamic priority evaluation. Background Art

[0002] In multi-target tracking scenarios, radar dwell time is a limited resource, and its allocation efficiency directly impacts target state estimation accuracy and overall system performance. Existing dwell time optimization methods often assume equal target importance or rely solely on fixed priority settings. This makes it difficult to achieve differentiated dwell time allocation in multi-target environments and struggles to cope with real-time changes in target motion characteristics and observation requirements in dynamic scenarios. Summary of the Invention

[0003] The purpose of the present invention is to provide a multi-target radar time resource allocation method based on dynamic priority evaluation, a computer device, a computer-readable storage medium and a computer program product, which can achieve a reasonable configuration of radar dwell time resources and effectively improve the tracking accuracy and reliability of the target.

[0004] One aspect of the present invention provides a multi-target radar time resource allocation method based on dynamic priority evaluation, comprising:

[0005] Step S1, establishing a target motion model and a radar observation model, wherein the target motion model uses a multi-model including possible motion modes to describe the target motion process;

[0006] Step S2, obtaining radar measurement values ​​based on the target motion model and the radar observation model, and constructing a priority function that includes target motion characteristics, spatial position, and historical observation information;

[0007] Step S3, embedding the priority function into the Bayesian Cramer-Rao lower bound, and establishing a radar resource allocation model with minimizing the sum of the weighted Bayesian Cramer-Rao lower bound as the objective function, and with the conditions that the total residence time does not exceed the resource scheduling period and the residence time of each target meets the constraint range;

[0008] Step S4: solving the radar resource allocation model by using the gradient projection method to obtain the resource allocation result.

[0009] Another aspect of the present invention provides a computer device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0010] Another aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0011] Yet another aspect of the present invention provides a computer program product, comprising a computer program, which implements the steps of the above method when executed by a processor.

[0012] The multi-target radar time resource allocation method based on dynamic priority evaluation, computer equipment, computer-readable storage medium, and computer program product according to the above aspects of the present invention can achieve reasonable allocation of radar dwell time resources and effectively improve the tracking accuracy and reliability of targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the present invention, the following briefly introduces the drawings used in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts.

[0014] Figure 1 is a flow chart of a multi-target radar time resource allocation method based on dynamic priority evaluation according to an embodiment of the present invention;

[0015] Figure 2 This is a schematic diagram of a radar and target distribution scenario according to an embodiment of the present invention;

[0016] Figure 3 is a comparison chart of the total root mean square error using the traditional equal distribution strategy and the optimization strategy of the present invention;

[0017] Figure 4 is a priority change curve diagram of each target according to an embodiment of the present invention;

[0018] Figure 5 It is a structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0020] One embodiment of the present invention provides a multi-target radar time resource allocation method based on dynamic priority evaluation, such as Figure 1 As shown, the multi-target radar time resource allocation method according to an embodiment of the present invention includes steps S1 to S4.

[0021] Step S1: Establish a target motion model and a radar system observation model.

[0022] Consider a self-transmitting and self-receiving radar located at the origin, with the tracking data rate and scheduling period both set to T0. At time t k , the status of a single target is where [x k ,y k ]and and Represent position, velocity, and acceleration respectively.

[0023] In the embodiment of the present invention, a multi-model including possible motion modes is used to describe the motion process of a maneuvering target. Model variable M k It is represented by a discrete random process, which takes one of m possible models. The target motion model is

[0024]

[0025] Among them, M k represents the model variables that affect time k-1 to k, M k The corresponding state transition matrix, represents zero-mean process noise, and its covariance matrix is The initial probability of the model is μ n,k＝1 =Pr{M k＝1 =n} and μ n,1 ≥0, as well as The evolution of the model sequence is represented as a first-order homogeneous Markov chain containing m states, and its transition probability is

[0026] π mn =Pr{M k =i|M k-1 =j}(2)

[0027] Where m,n∈M. The order of the model transition probability matrix M is i×j, and its (i,j)th element [M] j,i =π j,i satisfy π j,i ≥0,

[0028] Assuming that the radar can measure the radial distance and azimuth of the target, its measurement model is:

[0029] z k =h(x k )+u k (3)

[0030] where h(x k) represents the measurement function of active radar, including ranging and azimuth measurement

[0031]

[0032] u k Represents measurement error, which has a mean of zero and a covariance of ∑ k Gaussian distribution

[0033]

[0034] and The distance R k and azimuth angle θ k Cramer-Rao bound for the maximum likelihood estimate of :

[0035]

[0036] Where c is the speed of light, B is the effective bandwidth of the transmitted signal, and θ 3dB is the effective beam width, κ k is the target scattering characteristic, and SNR is the signal-to-noise ratio.

[0037] Step S2: Obtain radar measurement values ​​based on the target motion model and radar observation model, and construct a priority quantization model that includes target motion characteristics, spatial position, and historical observation information.

[0038] The core of priority sorting is to establish a quantitative index that reflects the real-time importance of the target. In an embodiment of the present invention, a quantitative model including target motion characteristics, spatial position and historical observation information is constructed.

[0039] 1) Motion characteristics: The target's speed and acceleration are selected as key parameters. High-speed targets or highly maneuverable targets are more difficult to track and require higher priority, so the calculation is defined as:

[0040]

[0041] 2) Spatial position: Considering the azimuth of the target and the radar, the target with an azimuth angle in the close-range surveillance area should be tracked first. The quantitative calculation is:

[0042]

[0043] 3) Historical observation information: The revisit interval is introduced to reflect the urgency of target observation. Targets with longer intervals accumulate uncertainty in state estimation and require priority resource allocation:

[0044] Δt=t k -t k-1 (10)

[0045] A differentiated weight allocation strategy is adopted. Considering that the change of the target motion state directly affects the tracking difficulty and information timeliness, the motion characteristics are given the highest weight {ω v ,ω a} to ensure the algorithm's responsiveness to highly maneuverable targets. Spatial position is second only to ω θ , historical observation information is mainly used to evaluate the continuity of target observation and plays an auxiliary adjustment role in the comprehensive priority calculation. Therefore, the weight of historical observation time ω t Minimum, satisfying ∑ω i = 1. The final priority function is defined as:

[0046] P q,k =ω v Δv+ω a Δa+ω θ Δθ+ω t Δt (11)

[0047] ω v is the weight of the target speed, ω a is the weight of the target acceleration, ω θ is the weight of the target azimuth, ω t is the weight of the revisit time interval, Δv=‖v k ‖-‖v k-1 ‖, Δa=‖a k ‖-||a k-1 ||, Δθ=θ k -θ k-1 , Δt=t k -t k-1 , v k 、a k ,θ k , t k are the velocity, acceleration, azimuth and revisit time interval of the target at time k, respectively, v k-1 、a k-1 ,θ k-1 , t k-1 They represent the velocity, acceleration, azimuth and revisit time interval of the target at time k-1 respectively.

[0048] The model generates a priority sequence in real time by dynamically updating target feature data. Targets are sorted in descending order based on this sequence, high-priority targets are tracked first, and differentiated weights are provided for subsequent residence time allocation.

[0049] Step S3: Priority weights are embedded into the Bayesian Cramér-Rao Lower Bound (BCRLB), and an optimization model (radar resource allocation model) is established with the objective function of minimizing the sum of weighted BCRLBs, with the conditions that the total residence time does not exceed the resource scheduling period and the residence time of each target meets the constraint range.

[0050] BCRLB provides an error lower bound for any unbiased estimator and is therefore widely used in resource allocation performance evaluation. At the kth moment, the FIM (Fisher Information Matrix) J of the target q is q (x q,k ,τ q,k ) is a function of the target state and the dwell time

[0051]

[0052] Where x q,k represents the state of the target q at the kth moment, τ q,k represents the radar's dwell time on target q at the kth moment, Y q,k is the residual variable of the radar measurement covariance, that is Represents the measurement function for the target q state x q,k The Jacobian matrix of T. q,k represents the radar's revisit time interval for target q. BCRLB is the inverse matrix of FIM and can be written as

[0053] In physical terms, the shorter the revisit time interval, the longer the dwell time, and the higher the tracking accuracy. On the one hand, a revisit time interval that is too short will significantly increase the time resource consumption of the radar, especially in a multi-target environment, which may cause beam scheduling conflicts and affect the overall performance of the system. In addition, an excessively long dwell time will cause the radar beam to stay on a single target or area for a longer period, reducing the coverage efficiency of the airspace and the multi-target tracking capability. Based on the above two considerations, the embodiment of the present invention establishes the following radar resource allocation model

[0054]

[0055] Where tr(·) represents the trace, τ=[τ1,τ2,...,τ Q ] T Assign the dwell time vector, τ total is the total residence time, τ max and τ min are the upper and lower limits of the residence time, T max and T minare the upper and lower limits of the revisit time interval respectively. The objective function f(x q,k ,τ q,k ) is nonlinear and monotonically decreasing with respect to the residence time, so Equation (13) is a nonlinear programming problem.

[0056] Step S4: According to the characteristics of the optimization model, a multi-objective residence time allocation optimization process based on target priority is designed to obtain the resource allocation result.

[0057] When the target revisit sequence is given, model (13) degenerates into a convex problem of single target residence time optimization, which can be solved by the gradient projection method as follows.

[0058] Step 1: Initialization phase: given the target state, residence time and target priority at time k.

[0059] Step 2: Priority calculation stage: Calculate the target priority through equations (7)-(11) and sort the targets in descending order according to priority.

[0060] Step 3: Resource scheduling phase: based on the current residence time τ q,k and priority P q,k , calculate the weighted BCRLB sum as the objective function; under the condition that the constraints are met, use the gradient descent method to update the residence time allocation vector; calculate the relative change of the objective function between two adjacent iterations, if it is less than the threshold ε, jump out of the loop; otherwise, continue to iterate.

[0061] Step 4: Measurement phase: According to the target order, use the optimal parameters to measure, and use the Interacting Multiple Model (IMM) algorithm to predict and update the target state x k .

[0062] The following describes the technical effects of the multi-target radar time resource allocation method according to an embodiment of the present invention with reference to specific experimental examples.

[0063] like Figure 2 As shown, the radar position is set as the coordinate origin, tracking Q = 6 targets, of which 4 targets are maneuvering targets and the remaining targets are non-maneuvering targets. The sampling interval T0 = 2s, the initial state acceleration of each target is 0m / s 2 , and the rest of the parameters are shown in Table 1.

[0064] Table 1 Initial motion state of each target

[0065]

[0066] The effective bandwidth of the transmitted signal is B = 1MHz, and the effective beam width is θ 3dB =0.0524rad, the total scheduling period is τtotal =0.3s, maximum dwell time τ max =τ total , minimum residence time τ min =0.02s, the noise intensity during the movement is η=0.1m 2 / s 3 , the target scattering characteristic is κ q,k =1m 2 , the radar reference distance is R ref =8×10 9 km, the reference signal-to-noise ratio is SNR ref =30dB, learning step size α = 0.5, and the iteration termination condition is that the relative change of the objective function value between two consecutive times is less than ε = 0.01 or the maximum number of iterations maxit = 200 is reached. A total of 90 frames of measurement data are used in this experiment. The motion status of the target at each stage is shown in Table 2.

[0067] Table 2 Target motion status at each stage

[0068]

[0069] In Table 2, g = 9.8 m / s 2 , ω is the turning angle, a x ,a y They represent the acceleration in the x direction and the acceleration in the y direction respectively.

[0070] The number of target dynamic models M=3, covering characteristic scenarios such as uniform speed, uniform acceleration, and coordinated turning.

[0071] Model 1: The first is the CV model, whose state transfer matrix and process noise covariance matrix are as follows

[0072]

[0073] Model 2: The second is the CA motion model, whose state transfer matrix and process noise covariance matrix are

[0074]

[0075] Model 3: The third model is the CT motion model, whose process noise covariance matrix is ​​the same as that of the CV model, that is, The state transfer matrix of CT is

[0076]

[0077] The initial model probability is μ1 = [0.7 0.1 0.2] T , the model transition probability matrix is

[0078]

[0079] from Figure 3 It can be seen that the optimization strategy of the embodiment of the present invention can make the total RMSE (Root Mean Square Error) of all targets lower than the traditional fixed allocation (equal allocation) strategy in the entire time domain, and can be reduced by about 25% on average. Between the 70th and 80th frames, the motion models of some targets are switched, resulting in a mismatch between the tracking filter and the actual state, and the RMSE of a single target increases instantly and increases the total error. The method of the embodiment of the present invention can re-evaluate the priority and adjust the dwell time in a timely manner, which can make the error fall back quickly compared with the fixed allocation strategy.

[0080] Depend on Figure 4 It can be seen that in the initial stage, the strategy of the embodiment of the present invention quickly searches for a feasible solution space for resource allocation by adjusting the weights of each target over a large range, resulting in large fluctuations. As the information acquired by the radar accumulates, it gradually stabilizes. For non-maneuvering targets 1-2, the optimization algorithm reduces the dwell time allocation while ensuring basic tracking accuracy, and the remaining resources are used for highly maneuverable targets. In the coordinated turning stage 46-75 frames, the priority of targets 4-6 is increased, driving an increase in dwell time. In the uniform acceleration stage 76-90 frames, targets 4-6 are still allocated a large amount of dwell time. At the same time, target 3 also receives more dwell time due to its increased priority. Therefore, the proposed strategy can implement differentiated resource scheduling based on the maneuverability characteristics of the target, proving the effectiveness of the proposed strategy.

[0081] In summary, the method of the embodiment of the present invention is aimed at multi-target tracking scenarios, takes into account the influence of target motion parameters, spatial position and historical observation information on target tracking, constructs a multidimensional priority evaluation model, generates dynamic priority indicators through quantitative analysis, and on this basis, proposes a dwell time optimization model that integrates dynamic priority evaluation. The Bayesian Cramer-Rao lower bound with priority weights and the interactive multi-model are combined to achieve optimal allocation of radar dwell time and target tracking, thereby reducing the total target tracking error and implementing differentiated resource allocation based on the target's maneuverability characteristics.

[0082] The embodiment of the present invention further provides a computer device, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 5As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store operating parameter data of each framework. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the steps of the method of the embodiment of the present invention are implemented.

[0083] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0084] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method of the embodiment of the present invention are implemented.

[0085] An embodiment of the present invention further provides a computer program product, including a computer program, which implements the steps of the method of the embodiment of the present invention when executed by a processor.

[0086] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. A multi-target radar time resource allocation method based on dynamic priority evaluation, characterized in that: include: Step S1, establishing a target motion model and a radar observation model, wherein the target motion model uses a multi-model including possible motion modes to describe the target motion process; Step S2, obtaining radar measurement values ​​based on the target motion model and the radar observation model, and constructing a priority function that includes target motion characteristics, spatial position, and historical observation information; Step S3, embedding the priority function into the Bayesian Cramer-Rao lower bound, and establishing a radar resource allocation model with minimizing the sum of the weighted Bayesian Cramer-Rao lower bound as the objective function, and with the conditions that the total residence time does not exceed the resource scheduling period and the residence time of each target meets the constraint range; Step S4: solving the radar resource allocation model by using the gradient projection method to obtain the resource allocation result.

2. The method according to claim 1, wherein In step S3, the radar resource allocation model is: Where τ=[τ1,τ2,...,τ Q ] T Assign the dwell time vector, T q,k represents the revisit time interval of the radar to the target q at time k, τ q,k represents the radar's dwell time on target q at time k, f(x q,k ,τ q,k ) is the objective function, Q is the total number of targets, P q,k is the priority function, J q (x q,k ,τ q,k ) is the Fisher information matrix of the target q at time k, and the inverse matrix of the Fisher information matrix is is the Bayesian Clamer-Rao lower bound, tr(·) represents the trace, τ total is the total residence time, τ max and τ min are the upper and lower limits of the residence time, T max and T min are the upper and lower limits of the revisit time interval respectively.

3. The method according to claim 2, wherein In step S2, the target motion characteristics are the target's velocity and acceleration, the spatial position is the target's azimuth, and the historical observation information is the revisit time interval; the priority function is defined as: P q,k =ω v Δv+ω a Δa+ω θ Δθ+ω t Δt, Among them, ω v is the weight of the target speed, ω a is the weight of the target acceleration, ω θ is the weight of the target azimuth, ω t is the weight of the revisit time interval, Δv=||v k ||-||v k-1 ||, Δa=||a k ||-||a k-1 ||, Δθ=θ k -θ k-1 , Δt=t k -t k-1 , v k 、a k ,θ k , t k are the velocity, acceleration, azimuth and revisit time interval of the target at time k, respectively, v k-1 、a k-1 ,θ k-1 , t k-1 They represent the velocity, acceleration, azimuth and revisit time interval of the target at time k-1 respectively.

4. The method according to claim 3, wherein In step S1, the target motion model is: Among them, x k represents the target state, M k represents the model variables that affect time k-1 to k, is the model variable M k The corresponding state transition matrix, represents zero-mean process noise, and its covariance matrix is The radar observation model is: z k =h(x k )+u k , where h(x k ) represents the radar measurement function, u k Represents measurement error, which has a mean of zero and a covariance of ∑ k Gaussian distribution, in, and The distance R to the target k and azimuth angle θ k Cramer-Rao bound for the maximum likelihood estimate of : Where c is the speed of light, B is the effective bandwidth of the transmitted signal, and θ 3dB is the effective beam width, κ k is the target scattering characteristic, and SNR is the signal-to-noise ratio.

5. The method according to claim 4, wherein The initial probability of the multi-model is μ n,k＝1 =Pr{M k＝1 =n} and as well as Transition probability π of multiple models mn for: π mn =Pr{M k =i|M k-1 =j} Among them, m,n∈M, M is the number of multiple models.

6. The method according to claim 5, wherein The Fisher information matrix of the target q at time k is: in, Denotes the measurement function h(x k ) for the state x of the target q q,k The Jacobian matrix, Y q,k is the residual variable of the radar measurement covariance, 7. The method according to claim 6, wherein Step S4 includes: Given the target state, residence time and target priority at the initial moment; Calculate the target priority according to the priority function and sort the targets in descending order of priority; Based on the current residence time and priority, the objective function is calculated. Under the condition that the constraints are met, the residence time allocation vector is updated using the gradient descent method. The relative change of the objective function between two consecutive iterations is calculated until the relative change of the objective function is less than the specified threshold. According to the target sequence, the optimal parameters are used for measurement, and the interacting multiple model algorithm is used to predict and update the target state.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images