Amplitude Information-Assisted Cognitive Radar Tracking Waveform Selection Method and System

Abstract

The present invention relates to the technical field of cognitive radar maneuvering target tracking and processing, and particularly relates to a method and system for selecting a cognitive radar tracking waveform assisted by amplitude information. By setting up a parametric radar waveform library and combining prior knowledge, the target measurement amplitude likelihood ratio is obtained by analyzing target fluctuations and environmental clutter background. In the state prediction stage, the tracking waveform parameters are selected based on the minimum information entropy criterion. In the state update stage, the target measurement amplitude likelihood ratio is used to correct the probability of the associated measurement in the unscented Kalman filter data probability data association, so that the weight of the probability of the associated measurement originating from the true target increases. The multi-model state estimation is updated using the corrected unscented Kalman filter probability data association, and the interactive multi-model state estimation is fused. The present invention uses target amplitude information to assist target tracking to reduce the cognitive radar maneuvering target tracking error and the loss-of-track rate, and improve the cognitive radar target tracking performance.

Description

Technical Field

[0001] The present invention relates to the technical field of cognitive radar target tracking processing, and particularly relates to a method and system for selecting a cognitive radar tracking waveform assisted by amplitude information. Background Art

[0002] Since the theory of cognitive radar was proposed, it has demonstrated performance advantages in various radar mission scenarios. By referring to the working mode of the echolocation system of bats in nature, a feedback closed-loop processing flow of radar system transmission and reception, namely the Perception-Action Cycle (PAC), was proposed. Among them, the transmission closed-loop feedback processing for tracking tasks can effectively improve the tracking performance of the radar system. The perception-action cycle of cognitive radar tracking is mainly used for radar transmission waveform selection, which can significantly improve the tracking performance of the radar system and reduce tracking errors compared with a radar system using a fixed transmission waveform. Kershaw and Evans proposed to approximate the observation covariance of the waveform using the Cramer-Rao Lower Bound (CRLB) of waveform parameter estimation, and derived a closed-form solution method for waveform parameters under the criteria of maximizing mutual information and minimum mean square error based on the recursive relationship of Kalman filtering under the linear observation relationship of the Gaussian model. And for the tracking problem in the scenario of dense measurement false alarms, a waveform adaptive probabilistic data association filter (WSPDAF) was given, which greatly improved the tracking performance.

[0003] However, in the existing basic theoretical framework for waveform selection in cognitive radar tracking, the relationship between the waveform and the observation noise covariance is mainly established based on the CRLB of parameter estimation, and then the waveform is selected based on the criterion by using the feedback of the state to minimize the tracking error. In the process of realizing waveform adaptability by the above method, only a part of the radar target information (such as time delay and Doppler information) is utilized for the feedback of the target state. These information directly represent the distance, radial velocity and azimuth observation values of the target. However, other useful information (target amplitude) contained in the target echo information is not effectively utilized. Summary of the Invention

[0004] Therefore, the present invention provides a method and system for selecting a cognitive radar tracking waveform assisted by amplitude information, which uses the target amplitude information to assist target tracking to improve and optimize the tracking performance of the radar for the target.

[0005] According to the design scheme provided by the present invention, a method for selecting a cognitive radar tracking waveform assisted by amplitude information is provided, including:

[0006] Set up a parameterized radar waveform library, and combine prior knowledge to obtain the likelihood ratio of the target measurement amplitude by analyzing target fluctuations and environmental clutter background;

[0007] In the state prediction stage of tracking, select the tracking waveform parameters for the next moment based on the minimum information entropy criterion;

[0008] In the state update stage, use the likelihood ratio of the target measurement amplitude to correct the probability of the associated measurement in the unscented Kalman filter data probability data association, so that the probability of the associated measurement from the true target increases; and use the corrected unscented Kalman filter probability data association to update the multi-model state estimation and fuse the interactive multi-model state estimation.

[0009] Cognitive radar can improve the performance of the radar system through the feedback closed-loop processing flow of the Perception-Action Cycle (PAC). The closed-loop feedback processing flow of the cognitive radar for the tracking task adjusts the transmitted waveform for the next moment under the criterion of minimizing the cost (tracking performance) according to prior information and current observation data to improve the tracking accuracy. The cognitive radar of the traditional method only utilizes a part of the radar target information (such as time delay and Doppler information) for the feedback of the target state, but the measurement amplitude information contained in the target echo information is not effectively utilized. Therefore, in order to further improve the tracking performance of the cognitive radar, in this application, the amplitude measurement information-assisted tracking is combined with the perception-action cycle of the cognitive radar waveform selection, and the target amplitude information is used to assist target tracking to improve the radar's detection and tracking performance of the target.

[0010] As the method for selecting the tracking waveform of the cognitive radar assisted by amplitude information of the present invention, further, in obtaining the likelihood ratio of the target measurement amplitude by analyzing target fluctuations and environmental clutter background in combination with prior knowledge, first, set the size of the amplitude detection threshold; then, obtain the amplitude probability density function of the desired signal and the false alarm amplitude probability density function from the echo at the output of the matched receiver to construct the probability density functions of the presence and absence of the target above the amplitude detection threshold; then, use the probability density functions of the presence and absence of the target above the amplitude detection threshold to represent the likelihood ratio of the target measurement amplitude.

[0011] By analyzing the clutter background and the fluctuation type of the target, establish the amplitude likelihood ratio, assist tracking based on the measurement amplitude information, improve the problem that the existing cognitive radar waveform selection fails to fully utilize the target measurement information, enhance the tracking accuracy of the cognitive radar for maneuvering targets in the presence of clutter measurement scenarios, and reduce the loss of tracking rate.

[0012] Furthermore, the present invention also provides a cognitive radar tracking waveform selection system assisted by amplitude information, comprising: a data analysis module, a state prediction module, and a state update module, wherein,

[0013] The data analysis module is configured to obtain the target measurement amplitude likelihood ratio by setting a parameterized radar waveform library and analyzing target fluctuations and environmental clutter background in combination with prior knowledge;

[0014] The state prediction module is configured to select the tracking waveform parameters for the next moment based on the minimum information entropy criterion during the state prediction stage of tracking;

[0015] The state update module is configured to, during the state update stage, use the target measurement amplitude likelihood ratio to correct the probability of the associated measurement in the unscented Kalman filter data probability data association, so that the probability of the associated measurement from the true target increases; and use the corrected unscented Kalman filter probability data association to update the multi-model state estimation and fuse the interactive multi-model state estimation.

[0016] Advantages of the present invention:

[0017] The present invention makes full use of the prior knowledge of the target model in the cognitive radar maneuvering target tracking scenario, assists the probability data association through the amplitude measurement information, and combines the interactive multi-model unscented Kalman filter for waveform selection of maneuvering target tracking, reducing the cognitive radar tracking error and loss-of-lock rate, and improving the cognitive radar target tracking performance. Description of the Drawings

[0018] Figure 1 Schematic diagram of the cognitive radar tracking waveform selection process assisted by amplitude information in the embodiment;

[0019] Figure 2 Schematic diagram of the perception-action cycle of the cognitive radar tracking task in the embodiment;

[0020] Figure 3 Schematic diagram of the principle of the cognitive radar maneuvering target tracking waveform selection algorithm assisted by amplitude information in the embodiment;

[0021] Figure 4 Schematic diagram of the maneuvering target tracking trajectory in the embodiment;

[0022] Figure 5 Schematic diagram of the comparison of tracking distance errors in the embodiment;

[0023] Figure 6 Schematic diagram of the comparison of tracking speed errors in the embodiment. Detailed Embodiment

[0024] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the drawings and technical solutions.

[0025] For the feedback of the target state by the cognitive radar of the traditional method, only a part of the radar target information (such as time delay and Doppler information) is utilized. However, the measured amplitude information contained in the target echo information is not effectively utilized. To further improve the tracking performance of the cognitive radar, the embodiments of the present invention, see Figure 1 as shown, provide a method for selecting a tracking waveform of a cognitive radar assisted by amplitude information, including:

[0026] S101. Set a parameterized radar waveform library, and obtain the target measurement amplitude likelihood ratio by analyzing the target fluctuation and the environmental clutter background in combination with prior knowledge;

[0027] S102. In the state prediction stage of tracking, select the tracking waveform parameters for the next moment based on the minimum information entropy criterion;

[0028] S103. In the state update stage, use the target measurement amplitude likelihood ratio to correct the probability of the associated measurement in the unscented Kalman filter data probability data association, so that the probability of the associated measurement from the real target increases in weight;

[0029] S104. Use the corrected unscented Kalman filter probability data association to update the multi-model state estimation and fuse the interactive multi-model state estimation.

[0030] See Figure 2 as shown, for the PAC cycle process of typical cognitive radar tracking waveform selection, according to the prior information and the information feedback of the current observation data, select the transmission waveform for the next moment under the criterion of minimizing the cost (tracking performance). The transmitted waveform interacts with the target and the environment to generate an echo that is received. After radar signal processing, new observation information is extracted for tracking, and an information feedback for the next moment is formed.

[0031] Suppose the radar emits a narrowband single-pulse signal:

[0032]

[0033] where is the baseband envelope signal and for transmission waveforms with different parameters, there is Re(·) is the real part operation, E T is the signal energy, f c is the signal carrier frequency. When selecting a linear frequency modulation signal with a Gaussian envelope, there is:

[0034]

[0035] where λ is the effective duration of the signal and b is the frequency modulation slope. Then, for a Swerling Ι type fluctuating target, the echo signal can be expressed as:

[0036]

[0037] where τ is the target time delay, υ is the Doppler frequency shift, is a complex Gaussian random process with a mean of 0 and a variance of N0 / 2. is the complex amplitude of the echo. Let be the measured amplitude information, and the echo energy is

[0038] Kershaw pointed out that the inverse of the Fisher information matrix is the CRLB of the measurement noise covariance of the unbiased state estimate. The Fisher information matrix J is defined as:

[0039]

[0040] where η = 2E R / N0 is the signal-to-noise ratio, and the elements in the matrix J are the second-order derivatives of the ambiguity function of the baseband transmitted signal s(t) at the positions in the time-delay and Doppler frequency-shift plane.

[0041] For the tracking of maneuvering targets in a two-dimensional plane, the target state at time k is representing the position, velocity, and acceleration states of the two coordinates respectively. Then the discretized state equation:

[0042]

[0043] where: F is the state transition matrix, U is the acceleration input matrix, is the mean acceleration, w k-1 is white noise following a Gaussian distribution with a noise covariance of Q k-1 , w k-1 ~N(0, Q k-1 ).

[0044] The observation equation of the target is

[0045] z k = h(X k|k ) + v(θ k-1 ) (6)

[0046] where the observed values are the distance, velocity, and azimuth angle, h(·) is a non-linear observation function, and the following relationships hold for the observed values:

[0047]

[0048] v(θ k-1 ) is a waveform parameter with θ k-1 = [τ, b] TThe observation noise that follows a Gaussian distribution, i.e., v(θ k-1 ) ~ N(0, R(θ k-1 ))), it is currently considered that the ambiguity function AF(τ, υ) is the likelihood estimate of the waveform time delay and frequency shift (τ, υ), and the inverse J -1 of its information matrix is the Cramer-Rao lower bound of parameter estimation:

[0049]

[0050] where is the signal-to-noise ratio, and the measurement error covariance converted to distance and speed is:

[0051]

[0052] where diag(·) is a diagonal matrix, and c is the speed of electromagnetic waves. Then there is:

[0053]

[0054] The observation noise for the angle is related to the signal-to-noise ratio and the beam width:

[0055]

[0056] where is a constant related to the beam width, and the joint observation covariance of distance, speed, and azimuth angle is:

[0057] The waveform parameter θ k-1 transmitted by the cognitive radar acts on the environment, and the echo generated is received at time k and used to update and obtain The process of waveform selection is exactly to select the optimal waveform parameter θ k-1 based on the cost function to minimize the tracking error at time k. Usually, the cost function J(X k|k , θ k-1 ) is defined according to the mean square error in the statistical sense:

[0058]

[0059] where E k [·] is the expectation operation. It can be found that is the posterior estimate of the state. Due to the non-linear mapping relationship between the observation and the state, it is difficult to directly calculate Equation (12). Then, the information entropy of the filtering covariance can be selected as an approximation of the cost function:

[0060]

[0061] Then, the minimum information entropy is used as the criterion for waveform selection:

[0062]

[0063] where Γ is the filter structure, det(·) is the determinant, and Θ is the feasible region of waveform parameters, i.e.:

[0064] Θ = {τ ∈ [τ min :Δτ:τ max , b ∈ [b min :Δb:b max} (15)

[0065] The above is the basic theoretical framework of waveform selection for existing cognitive radar tracking. It mainly establishes the connection between the waveform and the observation noise covariance based on the CRLB of parameter estimation, and then uses the feedback of the state to select the waveform based on the criterion to minimize the tracking error.

[0066] In the embodiment of this case, using the target amplitude measurement information to correct the probability data association and combining with the cognitive radar waveform selection can effectively improve the tracking performance of the cognitive radar and reduce the loss-of-lock rate.

[0067] As a preferred embodiment, further, when obtaining the target measurement amplitude likelihood ratio by analyzing the target fluctuation and the environmental clutter background in combination with prior knowledge, first, set the size of the amplitude detection threshold; then, obtain the amplitude probability density function of the desired signal and the false alarm amplitude probability density function from the echo at the output of the matched receiver to construct the probability density functions of the presence and absence of the target above the amplitude detection threshold; then, use the probability density functions of the presence and absence of the target above the amplitude detection threshold to represent the target measurement amplitude likelihood ratio.

[0068] Such as Figure 3The Amplitude Information-aided Waveform Selection for Cognitive Radar Tracking (AIWSCRT) algorithm shown can be generally summarized as follows: 1. Establish a parameterized radar waveform library; 2. Use prior knowledge to analyze target fluctuations and environmental clutter background, and establish a target measurement amplitude likelihood ratio model; 3. Receive echoes to extract target observations, and perform tracking track initiation and interactive multiple model filtering initialization according to the three-point method; 4. According to one-step state prediction, obtain the comprehensive one-step prediction and one-step prediction covariance, calculate the predicted innovation covariance and gain matrix by means of the Unscented Transformation (UT), and search for the best waveform parameters in the waveform library based on the minimum information entropy criterion; 5. Transmit the best waveform, obtain new measurement values, generate an association gating based on the comprehensive innovation covariance and observation prediction, and perform unified measurement confirmation; 6. Update the state estimate by the Amplitude Information-aided Unscented Kalman Probability Data Association Filtering (UKFPDAF-AI) of multiple models; 7. Perform multi-model state estimate fusion and update the model probability.

[0069] The amplitude information (AI) of the target measurement, that is, the echo amplitude information a, can be obtained from the output end of the matched reception. The measurement can be confirmed by setting the measurement threshold parameter of the amplitude for comparison. Generally, the echo amplitude of the real target obtained by matched reception is larger than that of the false alarm.

[0070] Assume that the amplitude detection threshold size is set to τ a , p1(a) represents the amplitude probability density function of the desired signal in the echo, p0(a) represents the amplitude probability density function of the false alarm, and let and represent the probability density functions of the target presence and absence above the detection threshold respectively, then there are:

[0071]

[0072] In the formula, the detection probability The false alarm probability Then the amplitude likelihood ratio (ALR) of the i-th measurement at time k can be expressed as:

[0073]

[0074] If the \(i\)-th measurement \(a\) at time \(k\) k (i) comes from the target, and its probability distribution is \(p_1(a\) k (i)|\(\eta,q\)), otherwise it is \(p_0(a\) k (i)|\(q\)), where \(\eta\) is the signal-to-noise ratio and \(q\) is the clutter background parameter.

[0075] The Amplitude Information-aided Unscented Kalman Filtering Probabilistic Data Association Filtering (UKFPDAF-AI) uses the amplitude likelihood ratio to correct the data association probability \(\beta\) on the basis of probabilistic data association k (i). The amplitude likelihood ratio is related to the target amplitude size. Since the measurement amplitude from the target is large, the weight of the measurement from the true target in the corrected association probability will increase, making the comprehensive state update value more accurate.

[0076] The unscented Kalman filter can well handle the tracking problem under the nonlinear observation of Equation (6). The current one-step predicted state is generated according to the state equation as The one-step predicted covariance is \(P\) k|k-1 :

[0077]

[0078] The Unscented Transformation (UT) generates a sigma point set through proportional sampling and the corresponding weights

[0079]

[0080] where \(n\) x is the state dimension, \(i = 0,1,2,\cdots,2n\) x is the index of the sigma point set and the weights, and \(\beta\) is a constant factor that adjusts the distance between sampling points. The observed predicted sigma point set is calculated through the observation function Then the state prediction mean and the observed prediction mean are:

[0081]

[0082] Then the state-observation cross-covariance and the self-covariance of the observed predicted value are:

[0083]

[0084] Waveform θ k-1 The corresponding innovation covariance and gain matrix are respectively:

[0085]

[0086] For data association, the observation dimension is n z , and the measurement z that meets the following requirements according to the ellipsoidal wave gate rule k,i is a valid measurement:

[0087]

[0088] where g is the wave gate parameter, and it is called the "σ number" of the ellipsoidal wave gate. At this time, the volume of the ellipsoidal wave gate is:

[0089]

[0090]

[0091] Assume that the false alarm density inside the wave gate is ρ, and the number of false alarms follows a Poisson distribution with a mean of ρV k , then the false alarms are generated:

[0092]

[0093] where m k is the number of observations, then the measurement set including amplitude information is The measurement innovation and the combined innovation are:

[0094]

[0095] where the event association probability is:

[0096]

[0097]

[0098] where P G is the probability that the measurement value falls into the association wave gate. When the association probability after adding the amplitude likelihood ratio correction is:

[0099]

[0100] Then the state update based on the combined innovation is:

[0101]

[0102] where is the one-step state prediction value, and K is the filtering gain matrix. The filtering covariance P k|k is updated to:

[0103]

[0104] where P k|k-1 is the state one-step prediction covariance.

[0105] When selecting the cognitive radar waveform, the association probability of the observation cannot be calculated, so the prediction of the filtering covariance cannot be obtained according to Equation (34). The degradation factor q2(ρV k , P D ) is introduced to approximate the filtering covariance after probability data association:

[0106]

[0107] The expression of the degradation factor is computationally complex due to the inclusion of high-dimensional integrals. Therefore, when predicting the filtering covariance generated by each waveform, the approximate calculation of the degradation factor can meet the requirements. Usually, when the observation dimension is 3 and the associated gate parameter takes 4 "σ" numbers, the degradation factor can be approximately calculated:

[0108]

[0109] Analyzing the above equation and the probability data association process, UKFPDAF-AI only modifies the association probability using the measurement amplitude likelihood ratio and does not affect the parameters ρV k and P D in Equation (36). Therefore, for UKFPDAF-AI, the filtering covariance can still be predicted by combining Equations (35 - 36), and the waveform selection is performed according to the minimum information entropy criterion.

[0110] (3) Interactive multiple model filtering. The number of models in the multiple model set for maneuvering tracking is N. The filtering state of model i at time k - 1 is The filtering covariance is The probability of model i is The transition probability from model i to j is p ij i, j = 1, 2,..., N.

[0111] Multiple model interaction:

[0112]

[0113] Multiple model one-step prediction:

[0114]

[0115] Multiple model update and fusion:

[0116] Model likelihood probability

[0117] Probability update

[0118] State fusion:

[0119]

[0120] Make full use of the prior knowledge of the target model, assist probability data association through amplitude measurement information, and combine interactive multiple model unscented Kalman filtering for waveform selection of maneuvering target tracking, effectively reducing the tracking error and loss-of-lock rate of cognitive radar.

[0121] Furthermore, based on the above method, the embodiment of the present invention also provides a cognitive radar tracking waveform selection system assisted by amplitude information, including: a data analysis module, a state prediction module, and a state update module, where,

[0122] The data analysis module is used to set a parameterized radar waveform library and obtain the target measurement amplitude likelihood ratio by analyzing target fluctuations and environmental clutter background in combination with prior knowledge;

[0123] The state prediction module is used to select the tracking waveform parameters at the next moment based on the minimum information entropy criterion during the state prediction stage of tracking;

[0124] The state update module is used to correct the probability of the associated measurement in the unscented Kalman filter data probability data association by using the target measurement amplitude likelihood ratio during the state update stage, so that the probability of the associated measurement from the true target increases; and use the corrected unscented Kalman filter probability data association to update the multiple model state estimation and fuse the interactive multiple model state estimation.

[0125] To verify the effectiveness of the solution of this case, the following is a further explanation in combination with test data:

[0126] Taking the tracking of a Swerling Ι type maneuvering target based on Rayleigh clutter background as an example, set the multiple model set as the constant velocity model (CV) and the current statistical model (CS), the model initial probability μ0 = [0.5, 0.5], and the transition matrix Initial state X 0|0 = [15000, -100, 0, 10000, -50, 0] T , and the initial state covariance matrix of tracking is P 0|0= diag([10000, 100, 1, 10000, 100, 1]), moving in a uniform straight line from 1 to 20 seconds, 41 to 60 seconds, and 81 to 100 seconds, making a left turn from 21 to 40 seconds with a turning rate of 6 degrees per second, and making a right turn from 61 to 80 seconds with a turning rate of 6 degrees per second. The following three tracking strategies are set for comparison with the algorithm of this case: Fixed Waveform Tracking (FWT), Amplitude Information aided Fixed Waveform Tracking (AIFWT), and Waveform Selection for Cognitive Radar Tracking (WSCRT). Under the above tracking strategies, the same false alarm rate and false alarm density are set, and at the same time, with h(X k|k ) as the center, false alarm measurements are generated within the comprehensive association gate area. The amplitude information of the measurements in the tracking process can be generated using random numbers. The measurement of the i-th amplitude threshold τ a at time k is:

[0127]

[0128] where rand is a random number. The signal-to-noise ratio is calculated as η = (r0 / r) 4 , and r0 is the reference distance at 0 dB.

[0129] The example simulation is implemented by programming in the MATLAB2022b environment. The simulation parameter settings are shown in Table 1:

[0130] Table 1 Simulation Parameter Settings

[0131]

[0132] Figure 4 The result graph of maneuvering target tracking is shown. The tracking trajectory of the cognitive radar with waveform selection is closer to the true trajectory, and the tracking process accuracy with the addition of amplitude information aided waveform selection is significantly better, especially in terms of smooth trajectory and high accuracy at the maneuvering turns. Figure 5 and 6 show the comparison of the tracking accuracy between the algorithm of this case and several maneuvering target strategies. It can be seen from the figure that the tracking distance error and speed error of the algorithm of this case are better than those of the three comparison algorithms.

[0133] 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.

[0134] In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method section.

[0135] The units and method steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A person of ordinary skill in the art can use different methods to implement the described functions for each specific application, but such implementation is not considered to exceed the scope of the present invention.

[0136] A person of ordinary skill in the art can understand that all or part of the steps in the above methods can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a magnetic disk, or an optical disc, etc. Optionally, all or part of the steps of the above embodiments can also be implemented using one or more integrated circuits. Correspondingly, the various modules / units in the above embodiments can be implemented in the form of hardware or in the form of software function modules. The present invention is not limited to any specific form of the combination of hardware and software.

[0137] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present invention, used to illustrate the technical solutions of the present invention, and are not intended to limit them. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, a person of ordinary skill in the art should understand that: any person familiar with the technical field of the present invention can still modify the technical solutions described in the foregoing embodiments, or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes, or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An amplitude information-assisted cognitive radar tracking waveform selection method, characterized in that Comprising: Setting a parameterized radar waveform library, setting the magnitude detection threshold, obtaining the amplitude probability density function of the desired signal and the false alarm amplitude probability density function from the echo at the output of the matched receiver to construct the probability density functions of the presence and absence of the target above the magnitude detection threshold, and combining prior knowledge to utilize the probability density functions of the presence and absence of the target above the magnitude detection threshold to obtain the likelihood ratio of the target measurement amplitude by analyzing target fluctuations and environmental clutter background; In the state prediction stage of tracking, selecting the tracking waveform parameters for the next moment based on the minimum information entropy criterion; In the state update stage, using the target measurement amplitude likelihood ratio to correct the probability of the associated measurement in the unscented Kalman filter data probability data association, increasing the weight of the probability of the associated measurement from the true target; and using the corrected unscented Kalman filter probability data association to update the multi-model state estimation and fuse the interactive multi-model state estimation, wherein the correction of the unscented Kalman filter data association probability using the target measurement amplitude likelihood ratio is expressed as: m k is the number of observations, P G is the probability that the measurement value falls into the associated gate, z k,i is the i-th measurement at time k, is the predicted mean of the observation, ρ is the false alarm density within the gate, S k is the waveform θ k-1 corresponds to the innovation covariance, P D is the detection probability, represents the amplitude likelihood ratio of the i-th associated measurement.

2. The amplitude information-assisted cognitive radar tracking waveform selection method according to claim 1, characterized in that For the target measurement amplitude likelihood ratio, the i-th measurement amplitude likelihood ratio at time k is expressed as: where and respectively represent the probability density functions of the presence and absence of the target above the amplitude detection threshold, a represents the echo amplitude information, and τ a represents the magnitude of the amplitude detection threshold.

3. The amplitude information-assisted cognitive radar tracking waveform selection method according to claim 1, characterized in that In the process of updating and fusing the multi-model state estimation using the corrected data association probability of the unscented Kalman filter, the state and filter covariance update and fusion process is expressed as: Where, is the one-step predicted value of the state from time k-1 to time k, K is the filter gain matrix, N is the number of maneuvering tracking multi-model sets, j is the model label, and z k represents the true observation with noise, represents the predicted value of the state observation.

4. The amplitude information-assisted cognitive radar tracking waveform selection method according to claim 3, characterized in that In the filtering covariance update, a degradation factor is introduced to approximate the filtering covariance after probability data association, where the approximation process is expressed as: q2(ρV k ,P D ) is the degradation factor, and V k is the volume of the ellipsoidal wave gate.

5. The amplitude information-assisted cognitive radar tracking waveform selection method according to claim 1 or 4, characterized in that When selecting the waveform according to the minimum cost criterion, the target waveform is selected using the criterion based on minimizing the information entropy of the prediction filter covariance.

6. An amplitude information-assisted cognitive radar tracking waveform selection system, characterized in that The method according to claim 1, comprising: a data analysis module, a state prediction module, and a state update module, wherein, The data analysis module is configured to obtain the likelihood ratio of the target measurement amplitude by setting a parameterized radar waveform library and combining prior knowledge to analyze target fluctuations and environmental clutter background; The state prediction module is configured to select the tracking waveform parameters for the next moment based on the minimum information entropy criterion in the state prediction stage of tracking; The state update module is configured to, in the state update stage, use the target measurement amplitude likelihood ratio to correct the probability of the associated measurement in the unscented Kalman filter data probability data association, increasing the weight of the probability of the associated measurement from the true target; and using the corrected unscented Kalman filter probability data association to update the multi-model state estimation and fuse the interactive multi-model state estimation.

7. An electronic device, characterized in that Including a memory and a processor, the processor and the memory complete communication with each other through a bus; the memory stores program instructions executable by the processor, and the processor can execute the method steps described in any one of claims 1 to 5 by calling the program instructions.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 5 are implemented.

Images