Jump-gliding trajectory target tracking method and system based on Jerk model

Abstract

The present invention belongs to the field of target tracking technology, and particularly relates to a jump-glide trajectory target tracking method and system based on the Jerk model, which constructs a maneuvering target tracking model based on the Jerk model; uses the deviation of position estimation to represent the acceleration disturbance and obtains the jerk variance representation, and positively correlates the jerk variance with the jerk frequency to obtain the jerk frequency representation; and sets the maneuvering target tracking model parameters according to the obtained jerk variance representation and jerk frequency representation, and obtains the process noise covariance matrix of the maneuvering target tracking model in the target tracking process; obtains the target state equation and observation equation according to the maneuvering target tracking model and the radar observation model, and performs jump-glide trajectory target tracking based on the target state equation and observation equation. The present invention adaptively adjusts and sets the maneuvering target tracking model parameters through the deviation of position estimation, so that the motion state estimation converges faster, effectively avoids the tracking error caused by unreasonable parameter setting, and improves the tracking effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of target tracking in high-end equipment manufacturing industry, and in particular relates to a jumping gliding trajectory target tracking method and system based on a Jerk model. Background Art

[0002] The trajectory of a hypersonic glide vehicle in the glide phase mainly includes two forms: balanced glide and jump glide. The balanced glide trajectory has a smooth flight trajectory and generally does not have maneuverable trajectory changes, but its requirements for the angle of attack control are relatively strict, and the engineering implementation is relatively difficult. The jump glide trajectory has an unfixed flight trajectory, and can maneuver and change trajectory multiple times during flight. The trajectory is difficult to predict, which greatly enhances the evasion and penetration capabilities. At the same time, the jump glide trajectory has a low requirement for the angle of attack control and is easy to implement in engineering. Therefore, it is the main flight trajectory used by hypersonic glide vehicles and has received extensive attention and research in recent years. Due to the characteristics of the jump glide trajectory, which has fast maneuvering changes and complex and unpredictable trajectories, it is of great research value to accurately track it. This problem is a typical high-maneuverability target tracking problem, and its research mainly focuses on two aspects: nonlinear filtering and maneuvering target modeling.

[0003] For highly maneuverable ballistic targets such as jump-glide trajectories, the most suitable maneuvering models are the "current" statistics (CS) model and the Jerk model. The Jerk model is the highest-order maneuvering model and can broaden the maneuvering range of the target by estimating the target acceleration change rate in real time. In theory, it is more suitable for tracking highly maneuverable targets. However, when the Jerk model is applied, the "jerk" frequency and acceleration variance need to be set in advance, which will make the model's accuracy and maneuverability adaptability unable to reach the optimal level, thus affecting the target tracking effect. Summary of the invention

[0004] In response to the problem of radar jump-glide trajectory tracking, the present invention provides a jump-glide trajectory target tracking method and system based on the Jerk model, which adaptively adjusts and sets the parameters of the maneuvering target tracking model through the deviation of position estimation, so that the motion state estimation converges faster, effectively avoids the tracking error caused by unreasonable parameter setting, improves the tracking effect, and facilitates application in actual scenarios.

[0005] According to the design scheme provided by the present invention, a jumping and gliding trajectory target tracking method based on the Jerk model is provided, which includes the following contents:

[0006] Construct a maneuvering target tracking model based on the Jerk model;

[0007] The deviation of the position estimation is used to represent the acceleration disturbance and obtain the acceleration variance representation, and the jerk variance is positively correlated with the jerk frequency to obtain the jerk frequency representation; and the parameters of the maneuvering target tracking model are set according to the obtained jerk variance representation and jerk frequency representation, and the process noise covariance matrix of the maneuvering target tracking model in the target tracking process is obtained;

[0008] The target state equation and observation equation are obtained according to the maneuvering target tracking model and radar observation model, and jumping and gliding trajectory target tracking is performed based on the target state equation and observation equation.

[0009] As a jumping and gliding trajectory target tracking method based on the Jerk model of the present invention, further, the maneuvering target tracking model is expressed as: Among them, w(t) represents the variance Zero-mean Gaussian white noise, is the model parameter representing the target jerk variance, and α is the model parameter representing the jerk frequency.

[0010] As a jumping and gliding trajectory target tracking method based on the Jerk model of the present invention, further, in obtaining the acceleration variance representation, the target displacement prediction estimate and the target displacement estimate are obtained according to the target motion state vector and the changes in jerk, acceleration and velocity within the sampling period, the difference between the target displacement prediction estimate and the target displacement estimate is used as the deviation of the position estimation, and the acceleration variance representation is obtained in combination with the jerk change and the velocity change; the process noise covariance is obtained based on the zero-mean Gaussian white noise.

[0011] As a jumping and gliding trajectory target tracking method based on the Jerk model of the present invention, further, the acceleration variance is expressed as: Where N is the acceleration variance adaptive coefficient, T is the sampling period, and x(k+1|k+1) and x(k+1|k) represent the estimated value and predicted estimated value of the target displacement at time k+1, respectively.

[0012] As a jumping and gliding trajectory target tracking method based on the Jerk model of the present invention, further, the jerk frequency at time k obtained when the jerk variance is positively correlated with the jerk frequency is expressed as: Among them, α min and α max They represent the minimum and maximum values ​​in the jerk frequency range, σ(k) represents the jerk variance at time k, and σ max and σ min They represent the maximum and minimum values ​​of all jerk variances at time k respectively.

[0013] As a jump-glide trajectory target tracking method based on the Jerk model of the present invention, jump-glide trajectory target tracking is further performed based on the target state equation and the observation equation, which specifically includes the following contents: obtaining the state prediction and covariance prediction at the next moment based on the state estimation and covariance matrix at the current moment, compensating the covariance prediction by using the process noise covariance matrix, and updating the state estimation and covariance matrix at the next moment by using the observation data at the next moment.

[0014] As a jumping and gliding trajectory target tracking method based on the Jerk model of the present invention, further, the unscented Kalman filter UKF algorithm is used to realize jumping and gliding trajectory target tracking, which specifically includes the following contents: using and To predict the target state at time k+1 and covariance P k+1 / k , where Φ k+1,k is the state transfer matrix at time k+1, Q k is the process noise covariance; Sigma points and corresponding weights are constructed based on the predicted state and covariance, and the target state estimate and covariance estimate at time k+1 are predicted based on the solution of the nonlinear propagation process of the observation equation.

[0015] Furthermore, the present invention also provides a jumping and gliding trajectory target tracking system based on the Jerk model, comprising: a model building module, a parameter setting module and a target tracking module, wherein:

[0016] Model building module, used to build a maneuvering target tracking model based on the Jerk model;

[0017] A parameter setting module is used to use the deviation of the position estimation to represent the acceleration disturbance and obtain the acceleration variance representation, and to positively correlate the acceleration variance with the acceleration frequency to obtain the acceleration frequency representation; and to set the maneuvering target tracking model parameters according to the obtained acceleration variance representation and jerk frequency representation and to obtain the process noise covariance matrix of the maneuvering target tracking model in the target tracking process;

[0018] The target tracking module is used to obtain the target state equation and observation equation according to the maneuvering target tracking model and the radar observation model, and to track the jumping and gliding trajectory target based on the target state equation and observation equation.

[0019] Beneficial effects of the present invention:

[0020] In order to solve the problem of estimation error introduced by the conventional Jerk algorithm due to the need to manually pre-set the acceleration variance and the "jerk" frequency, the present invention adaptively sets the acceleration variance through the deviation of position estimation, and associates the "jerk" frequency with the acceleration variance, so as to realize adaptive adjustment of model parameters while estimating the target motion state; further, the improved Jerk model is combined with the UKF algorithm, and verified by experimental data, the motion state estimation of this scheme has a relatively fast convergence speed, and the root mean square error and cumulative error results are small, which has a better application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The flowchart of the jumping and gliding trajectory target tracking method based on the Jerk model in the embodiment is shown;

[0022] Figure 2 It is a schematic diagram of the jump-gliding trajectory target tracking algorithm flow in the Jerk model in the embodiment;

[0023] Figure 3 It is a schematic diagram of a three-dimensional ballistic curve and a radar detection curve in the embodiment;

[0024] Figure 4 It is a schematic diagram of the velocity curve in the earth-fixed coordinate system in the embodiment;

[0025] Figure 5 Schematic diagram of the comparison of the root mean square errors of the X, Y and Z directions in the embodiment;

[0026] Figure 6 The comparison of the root mean square error of the speed in the X, Y and Z directions in the embodiment is shown;

[0027] Figure 7 The comparison of the cumulative errors of the X, Y and Z directions in the embodiment is shown;

[0028] Figure 8 Schematic diagram of the comparison of the accumulated errors of the speeds in the X, Y and Z directions in the embodiment. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention clearer and more understandable, the present invention is further described in detail below in conjunction with the accompanying drawings and technical solutions.

[0030] The jump-glide trajectory is one of the typical trajectories of hypersonic gliding vehicles. It has the characteristics of high speed, strong maneuverability, complex trajectory, and high prediction difficulty. It is one of the important targets of in-depth research in the field of target detection and tracking in recent years. In order to solve the problem of radar tracking of jump-glide trajectories, the embodiment of the present invention provides a jump-glide trajectory target tracking method based on the Jerk model, see Figure 1 As shown, it contains the following contents:

[0031] S101. Construct a maneuvering target tracking model based on the Jerk model;

[0032] S102, using the deviation of the position estimation to represent the acceleration disturbance and obtain the jerk variance representation, and positively correlating the jerk variance with the jerk frequency to obtain the jerk frequency representation; and setting the maneuvering target tracking model parameters according to the obtained jerk variance representation and jerk frequency representation, and obtaining the process noise covariance matrix of the maneuvering target tracking model in the target tracking process;

[0033] S103, obtaining a target state equation and an observation equation according to a maneuvering target tracking model and a radar observation model, and performing jump-glide trajectory target tracking based on the target state equation and the observation equation.

[0034] To address the estimation errors caused by the need to manually pre-set the jerk variance and "jerk" frequency in the conventional Jerk algorithm, the jerk variance is adaptively set through the deviation of position estimation, and the "jerk" frequency is associated with the jerk variance. This allows adaptive adjustment of model parameters while estimating the target motion state, thereby improving the practicality of the maneuvering target tracking model.

[0035] The Jerk model expands the acceleration dimension based on the Singer model and uses a zero-mean colored noise process to describe the target's acceleration. Its exponential autocorrelation function satisfies

[0036]

[0037] In the formula, α is the unknown parameter of the model, is the target jerk variance, α is the jerk frequency, i.e. the “jerk” frequency, which is the inverse of the jerk time constant. Whitening equation (1) yields:

[0038]

[0039] Where: w(t) is the variance Zero-mean Gaussian white noise.

[0040] The state vector of the system is represented as The continuous-time state equation of the target can be expressed as

[0041]

[0042] in, B=[0 0 0 1] T .

[0043] The corresponding discrete state equation is

[0044] X(k+1)=F(k)X(k)+W(k)

[0045] (4)

[0046]

[0047] The process noise covariance is

[0048]

[0049] Where:

[0050]

[0051] It can be seen that the estimation of target jerk is added to the state vector of the Jerk model, which can improve the accuracy of tracking the state information of the maneuvering target and the tracking performance of the target maneuvering moment. However, when the Jerk model is used for maneuvering target tracking, the tracking performance is affected by the process noise covariance matrix of the maneuvering model, and the matrix contains the "jerk" frequency α, the acceleration variance and sampling period T, where T depends on the performance of the sensor used to detect the target, and α and Depending on the maneuvering characteristics of the target, in actual application, the latter two parameters need to be pre-set based on human experience, but the "jerk" frequency and acceleration of the maneuvering target change in real time, which will introduce human estimation errors.

[0052] In the embodiment of the present case, further, in obtaining the jerk variance representation, the target displacement prediction estimate and the target displacement estimate are obtained based on the target motion state vector and the changes in jerk, acceleration and velocity during the sampling period, the difference between the target displacement prediction estimate and the target displacement estimate is used as the deviation of the position estimation, and the jerk variance representation is obtained in combination with the jerk change and the velocity change; the process noise covariance is obtained based on zero-mean Gaussian white noise.

[0053] In the present embodiment, a ground-based radar is used to detect and track the jump-glide trajectory. Usually, the position of the radar station is described in the geodetic coordinate system, the position of the detected target is described in the earth-fixed system (ECEF), and the radar detection of the target position is described in the radar sphere coordinate system. In order to describe the target position in a unified coordinate system, it is necessary to convert between coordinate systems. Without considering noise, the conversion process is as follows:

[0054] (1) Convert the radar station position (L, B, H) to ground-fixed coordinates (r R ) ECEF

[0055]

[0056] In the formula a is the equatorial radius of the earth, e is the eccentricity of the earth, L is the geographical longitude of the radar station, B is the geographical latitude of the radar station, and H is the geodetic height of the radar station.

[0057] (2) Convert the target coordinates (X, Y, Z) in the ground-fixed system to the coordinates (x, y, z) in the radar rectangular coordinate system (ENU)

[0058]

[0059] In the formula, is the transformation matrix between the ground fixed system and the radar rectangular coordinate system.

[0060]

[0061] (3) Convert the target coordinates (x, y, z) in the radar rectangular coordinate system into the radar spherical coordinate system (R, A, E)

[0062]

[0063] Assume that the motion state vector of the target at time k is In the sampling period T from time k to time k+1, the change in jerk is Δj, the change in acceleration is Δa, and the change in velocity is Δv. Due to the change in jerk, the motion in the sampling period cannot be simply approximated as uniformly accelerated linear motion. According to the literature, the predicted estimated value of the target displacement at time k+1 is

[0064]

[0065] In the formula, N0 represents the proportional constant of non-uniformly accelerated motion. But in fact, the acceleration of the target at time k+1 is Speed Therefore, after correction, the estimated value of the target displacement at time k+1 is

[0066]

[0067] The difference between the estimated value and the predicted value at time k+1 is Δd=x k+1 / k+1 -x k+1 / k , combined with Δv=ΔaT, Δa=ΔjT, then

[0068]

[0069] Therefore, a new adaptive adjustment formula for jerk variance can be obtained:

[0070]

[0071] Where N is the adaptive coefficient and T is the sampling period.

[0072] Therefore, we can see that the basic idea of ​​jerk variance adaptation is the predicted estimated value of the target displacement from time k to time k+1. The acceleration disturbance increment between time k and time k+1 is not considered, and the estimated value at time k+1 is The jerk disturbance effect from time k to time k+1 is included, so the jerk disturbance can be directly expressed by the deviation of the position estimate. When the target is in a weak maneuvering state, the state estimation value of the target at time k is slightly different from the state prediction value, and the jerk variance is small. When the target is in a strong maneuvering state, the state estimation value at time k is significantly different from the state prediction value at time k, and the jerk variance is large.

[0073] Since the "jerk" frequency is compatible with the jerk variance, that is, when the target undergoes a large maneuver, the deviation of the position estimate becomes larger, its jerk variance also becomes larger, and the corresponding "jerk" frequency should also change; and when the target maneuvers are small, the deviation of the position estimate becomes smaller, its jerk variance also becomes smaller, and the corresponding "jerk" frequency should also become smaller. In order to obtain a "jerk" frequency that is compatible with the jerk variance, the "jerk" frequency is positively correlated with the jerk variance. The "jerk" frequency at time k is defined as follows

[0074]

[0075] In the formula, α min and α max are the minimum and maximum values ​​that the "jerk" frequency can take respectively; σ max is the maximum value of all acceleration variances at time k, that is, σ max =max{σ1,σ2…σ k}. It can be seen that when σ(k)=σ max When α(k)=α max ; When σ(k)=σ min When α(k)=α min In this way, the acceleration variance and the "jerk frequency" are adaptively positively correlated.

[0076] Furthermore, jump-glide trajectory target tracking is performed based on the target state equation and observation equation, which specifically includes the following contents: obtaining the state prediction and covariance prediction at the next moment based on the state estimation and covariance matrix at the current moment, compensating the covariance prediction using the process noise covariance matrix, and updating the state estimation and covariance matrix at the next moment using the observation data at the next moment.

[0077] Initialized by 4 sets of measurement data before time k, the initial estimate at time k is obtained and covariance P k / k , the algorithm flow is as follows Figure 2 As shown, it can be described as follows:

[0078] (1) Calculate the acceleration variance using formula (15): Use equation (16) to calculate the "jerk" frequency α.

[0079] (2) Use equation (6) to calculate the process noise variance Q k .

[0080] (3) State prediction

[0081]

[0082] (4) Covariance prediction

[0083]

[0084] (5) Construct Sigma points and corresponding weights based on state prediction values

[0085]

[0086] (6) Nonlinear propagation based on observation equations

[0087] Z k+1 / k (i) = h(χ k+1 / k (i)),i=0,1,···,2n (20)

[0088] (7) Calculate the estimated value and covariance matrix of the observed value

[0089]

[0090] (8) Calculate the innovation vector and innovation covariance matrix

[0091]

[0092] (9) State Estimation and Covariance Update

[0093]

[0094] For the improved Jerk model, four sets of observation data are required for filter initialization. Therefore, the four sets of observation data before time k can be taken, and the initial estimate and covariance at time k are calculated using the unscented transformation (UT); then the process noise variance matrix at time k is constructed; the state and covariance prediction at time k+1 are calculated based on the state estimate and covariance matrix at time k, and the predicted covariance is compensated with the process noise variance matrix; finally, the state estimate and covariance matrix at time k+1 are updated using the observation data at time k+1.

[0095] Furthermore, based on the above method, an embodiment of the present invention also provides a jumping and gliding trajectory target tracking system based on the Jerk model, comprising: a model building module, a parameter setting module and a target tracking module, wherein:

[0096] Model building module, used to build a maneuvering target tracking model based on the Jerk model;

[0097] A parameter setting module is used to use the deviation of the position estimation to represent the acceleration disturbance and obtain the acceleration variance representation, and to positively correlate the acceleration variance with the acceleration frequency to obtain the acceleration frequency representation; and to set the maneuvering target tracking model parameters according to the obtained acceleration variance representation and jerk frequency representation and to obtain the process noise covariance matrix of the maneuvering target tracking model in the target tracking process;

[0098] The target tracking module is used to obtain the target state equation and observation equation according to the maneuvering target tracking model and the radar observation model, and to track the jumping and gliding trajectory target based on the target state equation and observation equation.

[0099] In order to verify the effectiveness of this solution, the following is a further explanation based on experimental data:

[0100] According to the dynamic equations of the hypersonic gliding vehicle and the numerical integration algorithm, a jump-gliding trajectory is optimized. The three-dimensional trajectory curve is as follows: Figure 3 As shown in the figure, the radar detection curve is the bold part, and its velocity curve in the ground-fixed coordinate system is as follows Figure 4 As shown, it can be seen that the target is in a maneuvering state. The actual position of the radar is 65.6°E, 2°N, and the elevation is 0m. Its detection range is 115°~245° in azimuth and 1°~85° in elevation, with a maximum range of 2000 kilometers. The radar found the target and started tracking after 1266s, and ended tracking at 1459.2s. The tracking time was 193.2s, and the data rate was T=0.1s.

[0101] The experimental platform is as follows: a mobile workstation with Windows 10 operating system, Intel Core(TM) i7-9850 (2.60GHz) CPU, 32GB memory, NVIDIA Quadro RTX 3000 graphics card, and MatlabR2020b programming environment.

[0102] The improved Jerk model in this case is compared with the existing common Jerk model, and the filtering algorithm adopts UKF. The parameters of the improved Jerk model are as follows: the value range of the "jerk" frequency is 0.1~1, and the adaptive coefficient N=10 -5 The parameters of the Jerk model are as follows: the jerk frequency is 0.5, and the jerk variance is 0.04. The observation data of the first four moments of the radar are selected for initialization to obtain the initial values ​​of the state and covariance estimation. The standard deviation of the observation angle is 0.1°, and the standard deviation of the distance is 100m.

[0103] Based on the above simulation scenarios and model parameters, Matlab R2020b was used to perform Monte Carlo simulation 100 times. In order to verify the effectiveness of the algorithm in this case, two evaluation indicators can be used, namely the root mean square error RMSE and the cumulative error AE, where the cumulative error AE represents the overall error in the entire tracking process. The definitions are as follows:

[0104]

[0105] In the formula, x i (k) and x i (k|k) represents the true value and estimated value at the kth moment in the i-th simulation, N is the number of Monte Carlo simulations, and M represents the number of observation points. The simulation results are shown in Figure 4 , Figure 5 As shown. Figure 4 From the position RMSE curves in the X, Y, and Z directions, it can be seen that at the beginning of tracking, the position error of the target for both maneuvering models is relatively large, about 1000m. After about 10s, the error decreases rapidly and reaches stability. Before 1300s, the target maneuvering speed is relatively small, and the tracking accuracy of the two maneuvering models is comparable. In the time period of 1300s to 1400s, the target maneuvering speed is relatively large, and the tracking accuracy of the two in the Z direction is comparable, but in the X and Y directions, the improved Jerk model has a smaller root mean square error than the original model and is more suitable for the maneuvering characteristics of the target. It can be clearly observed that the root mean square errors of the two maneuvering models fluctuate up and down, but the root mean square error of the improved Jerk model is smaller than that of the original model as a whole, and the accuracy is higher. From Figure 5 From the velocity RMSE curves in the X, Y, and Z directions, it can be seen that the performance of the improved Jerk model is better than that of the original model as a whole, and the improved Jerk model converges slightly faster than the original model.

[0106] In order to more intuitively compare the performance of the improved Jerk model with the original model, the cumulative error AE mentioned above is used for analysis. Figure 7 It can be seen that the position estimation errors in all directions always exist during the tracking process. The position cumulative errors of the improved Jerk model and the original model are increasing, but the position cumulative error of the improved Jerk model is smaller than that of the original model. Figure 8 It can be seen that the velocity accumulation error in each direction increases rapidly at the beginning of the tracking process, and gradually becomes flat after the tracking is stable, indicating that the velocity estimation error is small, but the velocity accumulation error of the improved Jerk model is also smaller than that of the original model. This verifies the effectiveness of the improved Jerk model algorithm in this case from another perspective.

[0107] The above experimental results further prove that the original model algorithm uses artificially set acceleration variance and "jerk" frequency, which cannot achieve adaptive adjustment of model parameters. Therefore, it cannot fully adapt to the maneuvering changes that occur during target tracking, and there will be large root mean square errors and cumulative errors. However, the improved algorithm proposed in this case optimizes the shortcomings of the original algorithm and can adaptively adjust the model parameters during the tracking process, making the tracking algorithm more sensitive to the maneuvering changes of the target and achieving better tracking effects.

[0108] Table 1 Comparison of algorithm running time

[0109]

[0110] Since the improved Jerk model adds adaptive calculations for acceleration variance and "jerk" frequency, it will take longer to run than the original model. Through 100 model Carlo simulations, as shown in Table 1, the running time of the improved Jerk model algorithm is 16.691s, and the running time of the original model is 16.255s. It can be seen that in terms of algorithm time consumption, the improved Jerk model only increases by 0.446s, and the operation time is at the same order of magnitude, indicating that the algorithm proposed in this case has certain engineering application value.

[0111] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0112] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A jumping and gliding trajectory target tracking method based on the Jerk model, characterized in that: Contains the following: Construct a maneuvering target tracking model based on the Jerk model; The deviation of position estimation is used to represent the jerk disturbance and obtain the jerk variance representation. The jerk variance is positively correlated with the jerk frequency to obtain the jerk frequency representation. The parameters of the maneuvering target tracking model are set according to the obtained jerk variance representation and jerk frequency representation, and the process noise covariance matrix of the maneuvering target tracking model target tracking process is obtained. Among them, in obtaining the jerk variance representation, the target displacement prediction estimate and the target displacement estimate are obtained according to the target motion state vector and the changes of jerk, acceleration and velocity within the sampling period. The difference between the target displacement prediction estimate and the target displacement estimate is used as the deviation of the position estimation, and the jerk variance representation is obtained in combination with the jerk change and the velocity change. The process noise covariance is obtained according to zero-mean Gaussian white noise. The jerk variance is expressed as: N is the adaptive coefficient of the jerk variance, T is the sampling period, x(k+1|k+1) and x(k+1|k) represent the estimated value and predicted estimated value of the target displacement at time k+1 respectively; the jerk frequency at time k obtained when the jerk variance is positively correlated with the jerk frequency is expressed as: α min and α max They represent the minimum and maximum values ​​in the jerk frequency range, σ(k) represents the jerk variance at time k, and σ max and σ min They represent the maximum and minimum values ​​of all jerk variances at time k respectively; The target state equation and observation equation are obtained according to the maneuvering target tracking model and radar observation model, and jumping and gliding trajectory target tracking is performed based on the target state equation and observation equation.

2. The jumping and gliding trajectory target tracking method based on the Jerk model according to claim 1 is characterized in that: The maneuvering target tracking model is expressed as: Among them, w(t) represents the variance Zero-mean Gaussian white noise, is the model parameter representing the target jerk variance, and α is the model parameter representing the jerk frequency.

3. The jumping and gliding trajectory target tracking method based on the Jerk model according to claim 1 is characterized in that: Jump-gliding trajectory target tracking is performed based on the target state equation and observation equation, which specifically includes the following contents: obtaining the state prediction and covariance prediction at the next moment based on the state estimation and covariance matrix at the current moment, compensating the covariance prediction using the process noise covariance matrix, and updating the state estimation and covariance matrix at the next moment using the observation data at the next moment.

4. The jumping and gliding trajectory target tracking method based on the Jerk model according to claim 1 or 3, characterized in that: The unscented Kalman filter (UKF) algorithm is used to realize the tracking of jumping and gliding ballistic targets, which specifically includes the following contents: and To predict the target state at time k+1 and covariance P k+1 / k , where Φ k+1,k is the state transfer matrix at time k+1, Q k is the process noise covariance; Sigma points and corresponding weights are constructed based on the predicted state and covariance, and the target state estimate and covariance estimate at time k+1 are predicted based on the solution of the nonlinear propagation process of the observation equation.

5. A jumping and gliding trajectory target tracking system based on the Jerk model, characterized in that: Contains: model building module, parameter setting module and target tracking module, among which, Model building module, used to build a maneuvering target tracking model based on the Jerk model; The parameter setting module is used to use the deviation of the position estimation to represent the acceleration disturbance and obtain the acceleration variance representation, and to positively correlate the acceleration variance with the acceleration frequency to obtain the acceleration frequency representation; and to set the maneuvering target tracking model parameters according to the obtained acceleration variance representation and the acceleration frequency representation and to obtain the process noise covariance matrix in the target tracking process of the maneuvering target tracking model; wherein, in obtaining the acceleration variance representation, the target displacement prediction estimate and the target displacement estimate are obtained according to the target motion state vector and the changes of the jerk, acceleration and velocity within the sampling period, and the difference between the target displacement prediction estimate and the target displacement estimate is used as the deviation of the position estimation, and the acceleration variance representation is obtained in combination with the jerk change and the velocity change; the process noise covariance is obtained according to the zero-mean Gaussian white noise; the jerk variance is expressed as: N is the adaptive coefficient of the jerk variance, T is the sampling period, x(k+1|k+1) and x(k+1|k) represent the estimated value and predicted estimated value of the target displacement at time k+1 respectively; the jerk frequency at time k obtained when the jerk variance is positively correlated with the jerk frequency is expressed as: α min and α max They represent the minimum and maximum values ​​in the jerk frequency range, σ(k) represents the jerk variance at time k, and σ max and σ min They represent the maximum and minimum values ​​of all jerk variances at time k respectively; The target tracking module is used to obtain the target state equation and observation equation according to the maneuvering target tracking model and the radar observation model, and to track the jumping and gliding trajectory target based on the target state equation and observation equation.

Images