A radar low probability of intercept tracking method based on TBD

Abstract

The present invention discloses a radar low probability of intercept tracking method based on TBD, comprising the following steps: (1) In the radar target search stage, the PF-TBD algorithm is used to detect the target and obtain the target motion trajectory, and the target motion model transition probability matrix is provided; (2) Based on the target motion model transition probability matrix, the IMM-EKF algorithm is used for target tracking; (3) In the target tracking stage, by designing a radar radiation control model, the radar intermittent radiation is realized, and under the premise of meeting the radar target tracking accuracy, the number of radar radiation times is reduced, so as to achieve the radar low probability of intercept tracking. The present invention can reduce the tracking error, reduce the number of radar radiation times required for correcting the trajectory, and achieve the low probability of intercept tracking.

Description

Technical Field

[0001] The present invention relates to a radar low probability of intercept target tracking algorithm based on tracking before detection, and belongs to the field of radar radio frequency stealth. Background Technique

[0002] With the continuous development of modern science and technology, radars have been widely used, and the target tracking technology of radars in military is becoming increasingly important. During air combat, fighter jets often use radars to detect suspicious targets, and obtain parameters such as the position information and motion state of the targets according to the echoes of electromagnetic waves after encountering signals, and can lock and track the suspicious targets, estimate the motion state and threat level. Traditional radar target tracking algorithms do not have prior information on the state model and can only be applied to the tracking of some low-maneuver targets under relatively ideal conditions. However, in practical applications, there are often some situations with strong maneuverability during radar target tracking. Due to the lack of prior information on the state model, it is difficult to determine the state of the target being tracked. If traditional radar target tracking algorithms are used, it will lead to large errors, a large number of radar radiation times, and is easily intercepted by enemy aircraft, with poor safety.

[0003] The radar intermittent radiation model based on the prior state model is used to solve the problems of large errors and a large number of radar radiation times during the tracking of maneuvering targets without prior information on the state model.

[0004] In the air combat in the new era, due to the lack of prior information on the target state model, traditional target tracking algorithms cannot achieve the tracking of maneuvering target models in a short time, and the prediction covariance of the algorithm during the tracking process is large, the required number of radar radiation times is large, and it is easily intercepted by enemy aircraft. Summary of the Invention

[0005] Object of the Invention: Aiming at the above problems, the radar intermittent radiation model based on the prior state model proposed in this paper estimates the trajectory of the maneuvering target through the tracking before detection algorithm, obtains the state model transition probability matrix required for the target tracking filtering algorithm, and then uses the interactive multiple model extended Kalman filter with prior information on the state model to track the maneuvering target, and controls the radiation of the radar by comparing the prediction covariance of the filtering algorithm during the target tracking process with a preset threshold. The present invention can reduce the tracking error, reduce the number of radar radiation times required for correcting the trajectory, and achieve low probability of intercept tracking.

[0006] Technical Solution: A method for radar low probability of intercept tracking based on TBD includes the following steps:

[0007] (1) In the radar target search stage, use the PF-TBD algorithm to detect the target and obtain the target motion trajectory, and provide the target motion model transition probability matrix;

[0008] (2) Based on the target motion model transition probability matrix, use the IMM-EKF (Interactive Multiple Model-Extended Kalman Filter) algorithm for target tracking;

[0009] (3) In the target tracking stage, by designing a radar radiation control model, realize intermittent radar radiation, reduce the number of radar radiation times on the premise of meeting the radar target tracking accuracy, so as to achieve low probability of intercept tracking of the radar.

[0010] Further, the algorithm flow of the PF-TBD in step (1) is as follows:

[0011] (1) According to the prior probability distribution q(·) of the target state, generate N b target newborn particles,

[0012]

[0013] where, X k represents the target state at time k, Z k represents the measurement value at time k, E k represents the variable of the target existence state, E k = 0 indicates that the target disappears, E k = 1 indicates that the target appears;

[0014] (2) Calculate the weights of the newborn particles and normalize them. The weights of the unnormalized newborn particles are:

[0015]

[0016] where, represents the likelihood ratio, represents the birth probability, and normalize the newborn particles:

[0017]

[0018] (3) According to the target state transition equation, estimate N c continuing particles that continue to exist:

[0019]

[0020] (4) Calculate the weights of the continuing particles and normalize them. The weights of the unnormalized continuing particles are:

[0021]

[0022] where, represents the likelihood ratio, and normalize the continuing particles,

[0023]

[0024] (5) Calculate the mixture probability using the unnormalized weights;

[0025] (6) Target detection probability estimation, calculate the probability that the target exists at time k:

[0026]

[0027] (7) Combine the states of the newborn particles and the continuing particles in a weighted manner:

[0028]

[0029]

[0030] Put the two particle sets together to form a complete particle set

[0031]

[0032] (8) Resampling;

[0033] (9) Target state update and detection probability estimation update, after completing resampling, obtain the estimated value of the target's motion state

[0034] Furthermore, step (8) resampling is to resample N b +N c particles to N c particles.

[0035] Furthermore, step (9) obtain the estimated value of the target's motion state for approximating the target posterior probability distribution and the target appearance probability

[0036] Furthermore, the process of obtaining the target motion model transition probability matrix in step (1) is as follows:

[0037] In the radar target search stage, that is, from time k = 1 to time k = K, the estimated value of the target's motion state can be obtained through the PF-TBD algorithm According to the estimated value of the target's motion state, estimate the target motion state transition probability matrix Π = [π ij M×N , where π ij is the state transition probability from the i-th motion model to the j-th motion model.

[0038] Furthermore, if it is estimated that the target is moving in a uniform straight line, assign a relatively large transition probability π il > 0.9 to the constant velocity model, l is the constant velocity model, and the remaining elements in the target motion state transition probability matrix Π ​If it is estimated that the target is making a uniform turning motion, a relatively large transition probability π is assigned to the cooperative turning model. ir > 0.9, r is the uniform turning model, and the remaining elements in the target motion state transition probability matrix Π

[0039] Furthermore, the process of target tracking using the IMM-EKF algorithm in step (2) is as follows:

[0040] (a) Calculate the mixture probability of the target motion model:

[0041]

[0042] where, π ij is the state transition probability from the i-th model to the j-th model, represents the probability of the j-th model at the k-th moment, is the normalization factor;

[0043] (b) Mixture state estimation of the input of each filter and covariance matrix estimation

[0044]

[0045]

[0046] where, and respectively represent the filtered estimation value of the target motion state and the error covariance estimation matrix of the i-th model filter at the (k - 1)-th moment;

[0047] (c) Filtering of each model filter:

[0048] According to the mixture state estimation value and the error covariance estimation matrix of each model, as well as the observation value Z k , the extended Kalman filter (EKF) algorithm is used to filter the target motion state, and the target motion state estimation values and P j (k|k) of each filtering model are obtained;

[0049] (d) Calculate the likelihood function:

[0050]

[0051] where, is the adjoint covariance matrix, The calculation formula of

[0052]

[0053] (e) Calculate the updated values of the probabilities of each model:

[0054]

[0055] (f) Calculate the estimated value of the final target motion state and the estimated value of the error covariance matrix:

[0056]

[0057]

[0058] Further, the radar radiation control model designed in step (3) is: According to the predicted covariance matrix P(k|k - 1) obtained by the IMM - EKF algorithm and the comparison with the preset threshold P th , control the radiation of the radar. When the predicted covariance matrix P(k|k - 1) of the IMM - EKF is less than the preset threshold P th , that is, P(k|k - 1) < P th , the radar does not radiate energy externally, and uses the passive sensor EMS to provide the observation data and the parameters required by the filtering algorithm; when the predicted covariance matrix P(k|k - 1) of the IMM - EKF is greater than the preset threshold P th , the radar radiates energy externally, and uses the observation data provided by the radar and the parameters required by the filtering algorithm to perform a new round of target tracking.

[0059] Beneficial effects: Using the tracking - before - detection algorithm to provide the prior information of the state model, it solves the problems of large errors in the radar target tracking algorithm and more radar radiation times when there is no prior information of the state model. The verification and analysis results show that compared with the traditional radar intermittent radiation model without prior information of the state model, this scheme reduces the error of the filtering and tracking algorithm while solving the problem of more radar radiation times, and can achieve low - probability - of - intercept tracking. Description of the Drawings

[0060] Figure 1 is the flow chart of the radar low - probability - of - intercept tracking method based on TBD

[0061] Figure 2 is the flow chart of IMM - EKF

[0062] Figure 3 is the sensor management model diagram

[0063] Figure 4 is the radar radiation diagram without prior information of the state model in the CV model

[0064] Figure 5 is the diagram of the prior value of the state model provided by PF - TBD in the CV model

[0065] Figure 6 The IMM-EKF target tracking diagram for obtaining the state model transition probability matrix in the CV model

[0066] Figure 7 The IMM-EKF estimation error diagram in the CV model

[0067] Figure 8 The radar radiation diagram of the radar intermittent radiation model based on the state model prior in the CV model

[0068] Figure 9 The radar radiation diagram in the CT model without state model prior information

[0069] Figure 10 The state model prior value diagram provided by PF-TBD in the CT model

[0070] Figure 11 The IMM-EKF target tracking diagram for obtaining the state model transition probability matrix in the CT model

[0071] Figure 12 The IMM-EKF estimation error diagram in the CT model

[0072] Figure 13 The radar radiation diagram of the radar intermittent radiation model based on the state model prior in the CT model Detailed implementation manner

[0073] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0074] As Figure 1 shown, a radar low intercept probability tracking method based on TBD includes the following steps:

[0075] (1) In the radar target search stage, use the PF-TBD algorithm to detect the target and obtain the target motion trajectory, and provide the target motion model transition probability matrix;

[0076] (2) Based on the target motion model transition probability matrix, as Figure 2 shown, use the IMM-EKF (interactive multiple model - extended Kalman filter) algorithm for target tracking;

[0077] (3) In the target tracking stage, by designing a radar radiation control model, realize radar intermittent radiation, and reduce the radar radiation times on the premise of meeting the radar target tracking accuracy, so as to realize radar low intercept probability tracking.

[0078] The radar radiation control model designed in step (3) is: according to the predicted covariance matrix P(k|k - 1) obtained by the IMM-EKF algorithm and the preset threshold P thComparison is made to control the radar radiation. When the predicted covariance matrix P(k|k - 1) of IMM - EKF is less than the preset threshold P th i.e., P(k|k - 1) < P th the radar does not radiate energy externally and uses the passive sensor EMS to provide the observation data and the parameters required by the filtering algorithm. When the predicted covariance matrix P(k|k - 1) of IMM - EKF is greater than the preset threshold P th the radar radiates energy externally and uses the observation data provided by the radar and the parameters required by the filtering algorithm to perform a new round of target tracking.

[0079] Figure 3 It is the management model diagram of the passive sensor

[0080] To verify the performance of the radar intermittent radiation model based on the prior of the state model proposed by the present invention, the number of radar radiation times of the improved method and the traditional radar intermittent radiation model without the prior of the state model under different motion models are analyzed and verified here. The experimental parameter settings are as follows:

[0081] Spatial coordinate system: two - dimensional coordinate system

[0082] Initial state model of the tracking target: (30.0 0.15 100.0 0.26)

[0083] Coordinate of the fighter plane: (0, 0)

[0084] Standard deviation of the process noise: 0.04

[0085] Sampling interval: 1 time point

[0086] Initial velocity: 300 m / s

[0087] The specific implementation is carried out according to the following steps:

[0088] Step 1: In the CV (constant velocity model) environment, the target moves in a uniform straight line in space. The radar measures the traces at 85 moments of the tracking target, and sets the threshold for target positioning and tracking to 0.05.

[0089] Step 2: First, without using the PF - TBD estimation model trajectory to provide prior conditions, only rely on IMM - EKF itself to perform target tracking. The number of radar radiation times is as Figure 4 shown

[0090] Step 3: Then use PF - TBD to provide the prior value of the state model, and use IMM - EKF that has obtained the state model transition probability matrix to perform the next - step tracking, as Figure 5 and 6 shown.

[0091] Step 4: Set the CT (Coordinated Turning Model) to start with a uniform linear motion, but at the 15th time point, let the target turn. Measure 85 random traces within the tracking time using radar, and set the threshold for target positioning and tracking to 0.05.

[0092] Step 5: Do not use the PF-TBD estimation model trajectory to provide prior conditions first, and only rely on the IMM-EKF itself for target tracking. The number of radar radiation times is as Figure 9 shown.

[0093] Step 6: Then use the PF-TBD to provide the prior value of the state model, and perform the next step of tracking with the IMM-EKF that has obtained the state model transition probability matrix, as Figure 10 and 11 shown.

[0094] In the CV model, the radar radiation pattern obtained from the radar intermittent radiation model based on the prior of the state model is as Figure 8 shown. Compared with the radar radiation times diagram without prior information of the state model, that is Figure 4 shown, the number of radiation times is significantly reduced, about Figure 4 one-third of the radar radiation times in Figure 7 . The radar intermittent radiation model designed in the present invention based on the prior of the state model is significantly superior to the traditional radar intermittent radiation model without prior information of the state model. The IMM-EKF estimation error is as

[0095] shown. It can be seen from the figure that the error is small and can meet the requirements. Figure 13 In the CT model, the radar radiation pattern obtained from the radar intermittent radiation model based on the prior of the state model is as Figure 9 shown. Compared with the radar radiation times diagram without prior information of the state model, that is Figure 9 shown, the number of radiation times is significantly reduced, about Figure 12 one-third of the radar radiation times in

[0096] In summary, by using the tracking-before-detection algorithm to provide prior information of the state model, the problems of large error in the radar target tracking algorithm and large number of radar radiation times without prior information of the state model are solved. The verification and analysis results show that compared with the traditional radar intermittent radiation model without prior information of the state model, this scheme reduces the error of the filtering and tracking algorithm while solving the problem of large number of radar radiation times, and can achieve low probability of intercept tracking.

[0097] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A radar low probability of intercept tracking method based on TBD, characterized in that, It includes the following steps: (1) In the radar target search stage, use the PF-TBD algorithm to detect the target and obtain the target motion trajectory, and provide the target motion model transition probability matrix; (2) Based on the target motion model transition probability matrix, use the IMM-EKF algorithm for target tracking; (3) In the target tracking stage, by designing the radar radiation control model, realize the intermittent radiation of the radar. On the premise of meeting the radar target tracking accuracy, reduce the number of radar radiation times, so as to achieve low probability of intercept tracking of the radar.

2. The radar low probability of intercept tracking method based on TBD according to claim 1, characterized in that The algorithm flow of PF-TBD described in step (1) is as follows: (1) Generate N target newborn particles according to the prior probability distribution q(·) of the target state b target newborn particles Among them, X k represents the target state at time k, Z k represents the measurement value at time k, E k represents the variable of the target existence state, E k = 0 indicates that the target disappears, E k = 1 indicates that the target appears; (2) Calculate the weights of the newborn particles and normalize them. The weights of the unnormalized newborn particles are: Among them, represents the likelihood ratio, represents the birth probability, and normalizes the newborn particles: (3) Estimate N c continuing particles that continue to exist according to the target state transition equation: (4) Calculate the weights of the continued particles and normalize them. The weights of the unnormalized continued particles are: Among them, represents the likelihood ratio and will continue with particle normalization, (5) Calculate the mixing probability using the unnormalized weights; (6) Target detection probability estimation, calculate the probability of the target existing at time k: (7) Combine the states of the newborn particles and the continued particles by weighting: Put the two particle sets together to form a complete particle set (8) Resampling; (9) Target state update and detection probability estimation update. After resampling is completed, the estimated value of the target's motion state is obtained 3. A radar low probability of intercept tracking method based on TBD according to claim 2, characterized in that, Step (8) resampling is to resample N b + N c particles to N c particles.

4. A radar low probability of intercept tracking method based on TBD according to claim 2, characterized in that Step (9) obtains the estimated value of the motion state of the target For approximating the target posterior probability distribution and the target appearance probability 5. A radar low probability of intercept tracking method based on TBD according to claim 1, characterized in that, The obtaining process of the target motion model transition probability matrix described in step (1) is as follows: In the radar target search stage, that is, from time k = 1 to time k = K, the estimated value of the target's motion state is obtained through the PF-TBD algorithm. According to the estimated value of the target's motion state, estimate the target motion state transition probability matrix Π = [π ij M×N , where π ij is the state transition probability from the i-th motion model to the j-th motion model.​ 6. The radar low probability of intercept tracking method based on TBD according to claim 5, characterized in that If it is estimated that the target is moving in a uniform straight line, a relatively large transition probability π is assigned to the constant velocity model il > 0.9, where l is the constant velocity model, and the remaining elements in the target motion state transition probability matrix Π If it is estimated that the target is moving in a uniform turn, a relatively large transition probability π is assigned to the coordinated turn model ir > 0.9, where r is the uniform turn model, and the remaining elements in the target motion state transition probability matrix Π 7. A radar low probability of intercept tracking method based on TBD according to claim 1, characterized in that The process of using the IMM-EKF algorithm for target tracking described in step (2) is as follows: (a) Calculate the target motion model mixing probability: where, π ij is the state transition probability from the i-th model to the j-th model, represents the probability of the j-th model at the k-th moment, is the normalization factor; (b) Hybrid state estimation for each filter input and covariance matrix estimation Among them, and respectively represent the filtered estimated value of the target motion state and the error covariance estimation matrix of the i-th model filter at the (k-1)-th moment. (c) Filtering of each model filter: According to the estimated values of the mixing states of each model and the error covariance estimation matrix as well as the observation value Z k , the extended Kalman filtering algorithm is used to filter the target motion state, and the estimated values of the target motion state of each filtering model and P j (k|k); (d) Calculate the likelihood function: Among them, is the adjoint covariance matrix, The calculation formula of (e) Calculate the updated values of the probabilities of each model: (f) Calculate the estimated value of the final target motion state and the estimated value of the error covariance matrix:

8. A radar low probability of intercept tracking method based on TBD according to claim 1, characterized in that, The radar radiation control model designed in step (3) is as follows: Based on the predicted covariance matrix P(k|k-1) obtained by the IMM-EKF algorithm and the preset threshold P th for comparison, control the radar radiation. When the predicted covariance matrix P(k|k-1) of the IMM-EKF is less than the preset threshold P th , that is, P(k|k-1) < P th , the radar does not radiate energy externally, and uses the passive sensor EMS to provide the observation data and the parameters required by the filtering algorithm; when the predicted covariance matrix P(k|k-1) of the IMM-EKF is greater than the preset threshold P th , the radar radiates energy externally, and uses the observation data provided by the radar and the parameters required by the filtering algorithm to perform a new round of target tracking.

Images