CN105842687A - Detection tracking integrated method based on RCS prediction information - Google Patents

Abstract

The invention discloses a detection tracking integrated method based on radar cross section (RCS) prediction information, which is mainly used for solving the problems that, in the prior art, the target detection probability is low and target tracking distance is short. The detection tracking integrated method based on RCS prediction information comprises steps of determining a target prediction wave gate at the k moment according to a target state estimation value and a state estimation covariance matrix of the k-1 moment, calculating a detection threshold of each detection unit in the prediction wave gate, performing detection on an echo signal to obtain a primary measurement data set, determining an RCS prediction range of the k moment according to the RCS observation information of the k-1 moment before, screening out measurement data of output data of a wave detector which exceed the detection threshold from the primary measurement set as the measurement data set of the k moment, and calculating the object state estimation value and the state estimation covariance matrix of the k moment. The detection tracking integrated method disclosed by the invention improves the object detection probability and track continuity, can be used for improving the object detection probability under the radar object tracking state, and expands the object tracking distance.

Description

Detecting and tracking integral method based on RCS information of forecasting

Technical field

The invention belongs to Radar Technology field, a kind of based on Radar Cross Section RCS information of forecasting adjustment mesh The detecting and tracking method of mark prediction each detector unit false-alarm probability of Bo Mennei, can be used for improving under radar target tracking state mesh Mark detection probability, Extended target tracking range.

Background technology

Radar system comprises the big basic module of Object Detecting and Tracking two.The main task of target detection is to receive radar To echo-signal process, and judge the presence or absence of target, due to noise and the impact of interference, need to use CFAR side Method reduces the probability of erroneous judgement, it is ensured that Radar Signal Detection has CFAR characteristic, and conventional CFAR detection algorithm includes list Unit average CFAR, Generalized Likelihood Ratio, adaptive matched filter etc..Target following is the target location letter obtained based on detection Breath, follows the tracks of out the flight path of target continuously by filtering.In target tracking algorism, mainly there are Kalman filtering, expansion card Kalman Filtering, unscented kalman filter algorithm etc..

Target following is carried out on the basis of target detection, and high detection performance can ensure that quickly rising of targetpath Beginning, and the detection performance of difference can cause the end that targetpath is too early, therefore the detection performance of target directly affects target Tracking performance.Object Detecting and Tracking is generally regarded as two independent processes by conventional radar systems, first carries out Target detection estimating target motion parameter, send into radar tracking device and be predicted, associate, filter after obtaining measurement information Deng process, it is achieved the detect and track to target.When target echo signal to noise ratio is relatively low, target detection probability is relatively low, it will make Become the discontinuity of targetpath, be easily caused flight path and terminate prematurely, thus target following distance is shorter.

After targetpath is initial, it is possible to obtain target following information, this information is fed back to object detector and will assist in Improve target detection performance.Existing is to adjust target prediction based on following the tracks of information design radar detector method main policies Detection threshold in region, to obtain optimum tracking performance.The core concept of said method is to reduce in estimation range Detection door, thus improve target detection probability, the final lifting obtaining target tracking accuracy.But, when target suddenly disappears Time, still use the method to adjust detection threshold in target prediction region, it is impossible to ensure that flight path terminates rapidly, thus cause The generation of false track.

Summary of the invention

Present invention aims to the deficiency of above-mentioned prior art, propose a kind of detecting and tracking based on RCS information of forecasting Integral method, under conditions of ensureing not produce false track, to adjust the void of each detector unit of target prediction Bo Mennei Alarm probability, improves the detection probability of target, the tracking range of Extended target under tracking mode.

For achieving the above object, technical solution of the present invention includes the following:

1) initiation parameter: arranging false dismissal counter initial value is 0；Arranging cymoscope Initial output signal is 0, cymoscope Etc. signal to be received；By targetpath start algorithm could, obtain the Initial state estimation of targetpathAnd original state Estimate covariance matrix P₀；

2) target setting state transition equation and radar measurement equation, according to kth-1 moment Target state estimator valueWith K-1 moment state estimation covariance matrix P_k-1, calculate the predictive value that kth moment target measuresAnd kth moment mesh The prediction covariance matrix D that scalar is surveyed_k|k-1；

3) predictive value measured according to kth moment targetThe prediction covariance matrix D measured with kth moment target_k|k-1, Determine the target prediction ripple door O in kth moment_k；

4) probability P that after target setting suddenly disappears, flight path correctly terminates_E, utilize equation below to calculate in kth moment target Prediction ripple door O_kInterior false-alarm probability P_Z:

(1-P_Z)^M=P_E,

Wherein, M represents the continuous false dismissal number of times terminated needed for flight path；

5) weight w (i of kth moment prediction Bo Mennei i-th detector unit is set；K), when utilizing equation below group to calculate kth Carve prediction ripple door O_kThe false-alarm probability of interior each detector unit:

$\{\begin{array}{c}\underset{i=1}{\overset{N_k}{\mathrm{\Π}}}\[1-P_f(i;k)\]=1-P_Z\\ P_f(i;k)=w(i;k)P_a,i=1,2,\mathrm{...},N_k\end{array},$

Wherein, P_f(i；K) kth moment prediction ripple door O is represented_kThe false-alarm probability of interior i-th detector unit, i=1,2 ..., N_k, N_k Represent kth moment prediction ripple door O_kThe number of interior detector unit, P_aVariable is solved for centre；

6) according to false-alarm probability P of each detector unit_f(i；K), algorithm of target detection is utilized to calculate kth moment prediction ripple door O_kDetection threshold T of interior each detector unit_i, by the output data of cymoscope and prediction ripple door O_kInterior arbitrary detector unit Detection threshold T_iCompare, if cymoscope output data are higher than detection threshold T of this detector unit_i, then this detector unit is used The metric data of corresponding estimated spatial position target, and will prediction ripple door O_kThe all metric data inside obtained are as preliminary amount Survey data acquisition system Z₁(k)；Otherwise, any process is not made；

7) according to the Radar Cross Section rcs measurement value in front k-1 moment, the predictive value of kth moment RCS is calculatedWith The prediction variance of kth moment RCSAnd according to this predictive valueWith prediction varianceDetermine kth moment RCS's Estimation range B_k；

8) according to estimation range B of kth moment RCS_kIn minimum RCS numerical value λ_min, calculate kth moment minimum noise Ratio numerical value ρ_min, and to utilize detection algorithm to calculate be ρ in signal to noise ratio_min, lowest detection probability beUnder conditions of required High false-alarm probability P_f,max；

9) according to the highest false-alarm probability P_f,maxCalculate corresponding detection threshold T_min, filter out preliminary metric data set Z₁(k) Middle cymoscope output data are higher than detection threshold T_minMetric data, as metric data set Z (k) in kth moment；

10) from kth moment metric data set Z (k), the metric data the highest with track association degree is chosen, by the amount chosen Survey data separate track algorithm and calculate kth moment Target state estimator valueWith state estimation covariance matrix P_k；

11) judge whether kth moment metric data set Z (k) is empty: if it is empty, then set to 0 by false dismissal enumerator, otherwise False dismissal enumerator originally numerically add 1；

12) value of false dismissal enumerator is judged: if the value of enumerator is more than or equal to the continuous false dismissal number of times M terminated needed for flight path, Then object tracking process terminates；Otherwise, step 2 is returned).

Advantages of the present invention is as follows:

The present invention is after targetpath is initial, owing to can obtain target following information, and this information feeds back to target inspection Survey device, so being favorably improved target detection performance；

The present invention, by adjusting the detection threshold in target prediction region, can obtain the tracking performance of optimum；

The present invention is not under conditions of producing false track, general by adjusting the false-alarm of each detector unit of target prediction Bo Mennei Rate, improves the detection probability of target under tracking mode, extends the tracking range of target.

Accompanying drawing explanation

Fig. 1 is the flowchart of the present invention；

Fig. 2 is with present invention signal to noise ratio variation diagram in object tracking process；

Fig. 3 is the detection probability comparison diagram of the present invention and traditional detection tracking.

Detailed description of the invention

With reference to Fig. 1, the present invention to realize step as follows:

Step 1, initiation parameter.

The initial value arranging false dismissal enumerator is 0；The signals to be received such as arranging cymoscope Initial output signal is 0, cymoscope； By targetpath start algorithm could, obtain the Initial state estimation value of targetpathAnd Initial state estimation covariance square Battle array P₀。

Step 2, target setting state transition equation and radar measurement equation, according to kth-1 moment Target state estimator value With kth-1 moment state estimation covariance matrix P_k-1, calculate the predictive value that kth moment target measures

2a) target setting state transition equation is:

x_k=F_k|k-1x_k-1+v_k|k-1,

Wherein, x_kRepresent the dbjective state in kth moment, F_k|k-1Represent the dbjective state transfer matrix in kth moment to kth-1 moment, x_k-1Represent the dbjective state in kth-1 moment, v_k|k-1Represent the process noise in kth moment to kth-1 moment, in this example F_k|k-1Use following form:

$F_{k|k-1}=\left[\begin{array}{cccc}1& \mathrm{\Δ}T& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& \mathrm{\Δ}T\\ 0& 0& 0& 1\end{array}\right],$

Wherein, Δ T represents and takes Δ T=10s in sweep spacing, this example.

2b) set radar measurement equation as:

z_k=h_k(x_k)+w_k,

Wherein, z_kRepresent the aim parameter measured value in kth moment, h_k() represents that the target in kth moment measures function, w_kWhen representing kth The measurement noise carved；

2c) according to kth-1 moment Target state estimator valueCalculate kth moment dbjective state predictive value

${\hat{x}}_{k|k-1}=F_{k-1}{\hat{x}}_{k-1};$

2d) according to kth moment dbjective state predictive valueCalculate the predictive value that kth moment target measures

${\hat{z}}_{k|k-1}=h_k\left({\hat{x}}_{k|k-1}\right).$

Step 3, according to kth-1 moment state estimation covariance matrix P_k-1With kth moment dbjective state predictive value Calculate the prediction covariance matrix D that kth moment target measures_k|k-1。

3a) according to kth moment dbjective state predictive valueCalculate kth moment target and measure the Jacobian matrix H of function_k:

$H_k={\▿}_x\left(h_k^T\right(x\left)\right){|}_{x={\hat{x}}_{k|k-1}},$

Wherein,Represent vector x derivation, ()^TRepresent transposition computing,Representative function?The functional value at place；

3b) according to kth-1 moment Target state estimator covariance matrix P_k-1With the Jacobi that kth moment target measures function Matrix H_k, calculate the prediction covariance matrix D that kth moment target measures_k|k-1:

$D_{k|k-1}=H_k\[F_{k-1}{P}_{k-1}{F}_{k-1}^T+Q_{k-1}\]H_k^T,$

Wherein, Q_k|k-1Represent the process noise covariance matrix in kth moment to kth-1 moment, this example use following form:

$Q_{k|k-1}={{\mathrm{\σ}}_p}^2\left[\begin{array}{cccc}{\mathrm{\ΔT}}^3/3& {\mathrm{\ΔT}}^2/2& 0& 0\\ {\mathrm{\ΔT}}^2/2& \mathrm{\Δ}T& 0& 0\\ 0& 0& {\mathrm{\ΔT}}^3/3& {\mathrm{\ΔT}}^2/2\\ 0& 0& {\mathrm{\ΔT}}^2/2& \mathrm{\Δ}T\end{array}\right],$

Wherein, σ_pRepresent process noise standard deviation, take σ_p=0.1.

Step 4, the predictive value measured according to kth moment targetThe prediction covariance matrix measured with kth moment target D_k|k-1, determine that the kth moment predicts ripple door O_k:

4a) target setting falls into the probability of prediction ripple doorThis example is taken as

4b) determine prediction ripple door coefficient gamma by lookup chi-square distribution table₁, to ensure that degree of freedom measures the card side of dimension for target Distribution variables is more than prediction ripple door coefficient gamma₁Probability beDuring wherein chi-square distribution table is theory of probability, the distribution of card side is random The distribution function table of variable；

4c) according to step 4b) in the prediction ripple door coefficient gamma that obtains₁, determine that the kth moment predicts ripple door O as follows_k:

$O_k=\left\{y_k\right|{(y_k-{\hat{z}}_{k|k-1})}^T{D}_{k|k-1}^{-1}(y_k-{\hat{z}}_{k|k-1})\≤{\mathrm{\γ}}_1\},$

Wherein, D_k|k-1Represent the prediction covariance matrix that kth moment target measures, γ₁Represent ripple door coefficient,When representing kth Carve the predictive value that target measures, y_kRepresent the position that kth moment target is likely to occur, | representing conditional code, the symbol left side is Set element, the right is the condition that element meets.

Step 5, target setting suddenly disappear after the probability P that correctly terminates of flight path_E, utilize equation below to calculate the kth moment pre- Survey ripple door O_kThe probability P of false-alarm inside occurs_Z:

(1-P_Z)^M=P_E,

Wherein, M represents the continuous false dismissal number of times terminated needed for flight path, takes M=3.

Step 6, sets the weight w (i of kth moment prediction Bo Mennei i-th detector unit；K), this example takes w(i；K)=1/N_k, i=1,2 ..., N_k, utilize equation below group to calculate the false-alarm of the kth moment prediction each detector unit of Bo Mennei Probability:

$\left\{\begin{array}{c}\underset{i=1}{\overset{N_k}{\Π}}\[1-P_f(i;k)\]=1-P_Z\\ P_f(i;k)=w(i;k)P_a,i=1,2,...,N_k\end{array}\right.,$

Wherein, P_f(i；K) false-alarm probability of kth moment prediction Bo Mennei i-th detector unit, N are represented_kRepresent the prediction of kth moment The number of Bo Mennei detector unit, P_aVariable is solved for centre.

Step 7, the kth moment is from prediction ripple door O_kInside filter out preliminary metric data set Z₁(k)。

7a) according to false-alarm probability P of the kth moment prediction each detector unit of Bo Mennei_f(i；K), i=1,2 ..., N_k, utilize Algorithm of target detection calculates detection threshold T of the kth moment prediction each detector unit of Bo Mennei_i；

7b) by the output data of cymoscope and prediction ripple door O_kDetection threshold T of interior arbitrary detector unit_iCompare, if inspection Ripple device output data are higher than detection threshold T of this detector unit_i, then by estimated spatial position target corresponding to this detector unit Metric data, otherwise, does not make any process；

7c) will prediction ripple door O_kThe all metric data inside obtained are as preliminary metric data set Z₁(k)。

The detection form of described radar cymoscope includes, square law detection, linear detection etc., and this example is selected but is not limited to put down Side's rate cymoscope.

Described algorithm of target detection includes, matched filtering detection algorithm, CA-CFAR, order statistic CFAR, Generalized Likelihood Ratio etc., this example is selected but is not limited to matched filtering detection algorithm.

Step 8, according to the rcs measurement value in front k-1 moment, utilizes RCS Forecasting Methodology to calculate the prediction of kth moment RCS ValuePrediction variance with kth moment RCS

Described RCS Forecasting Methodology includes, RCS Forecasting Methodology based on dependency, RCS prediction based on probability density transfer Method etc..This example is selected but is not limited to RCS Forecasting Methodology based on dependency, and its step is as follows:

8a) set RCS predictor exponent number as L, take L=3 in this example；

8b) calculate the correlation coefficient C in kth-i moment and kth-j moment target RCS as follows_k(i, j):

C_k(i, j)=[2 J₁(πu)/πu]²+ 1,

Whereinθ_i、θ_jRepresenting the radar observation angle in i Yu j moment, d represents observed object size, C represents the light velocity, F_cRepresent carrier frequency, J₁() represents Bessel function of the first kind；

8c) according to RCS dependency, obtain RCS predictor weights by solving equation below group:

$C_k\left[\begin{array}{c}1\\ w_{k-1}\\ .\\ .\\ .\\ w_{k-L}\end{array}\right]=\left[\begin{array}{c}{\mathrm{\σ}}_{\mathrm{\λ},p}^2\\ 0\\ .\\ .\\ .\\ 0\end{array}\right],$

Wherein, w_k-lRepresent the RCS predictor weights in the kth-l moment；C_kRepresent RCS correlation matrix, by matrix element, The correlation coefficient C of i.e. RCS_k(i j) is constituted；

8d) according to RCS predictor at the weights in kth-1 moment to kth-L moment, calculate RCS predictive value by following formula:

${\widehat{\mathrm{\λ}}}_{k,p}=\underset{l=1}{\overset{L}{\Σ}}{w}_{k-l}{\mathrm{\λ}}_{k-l}$

Wherein, λ_k-lRepresent kth-l moment target RCS observation.

Step 9, according to the predictive value of the kth moment RCS obtained in step 8Prediction variance with kth moment RCSDetermine estimation range B of kth moment RCS_k。

9a) the true RCS of target setting falls into the probability of RCS estimation rangeThis example is taken as

9b) determine RCS estimation range B by lookup Gauss distribution table_kIn minimum RCS numerical value λ_min, to ensure that variance isGaussian distributed random variable more than λ_minProbability beDuring wherein Gauss distribution table is theory of probability, Gauss distribution is random The distribution function table of variable；

9c) according to step 9b) in RCS estimation range B that obtains_kIn minimum RCS numerical value λ_min, determine RCS as the following formula Estimation range B_k:

B_k=[λ_min,+∞)。

Step 10, according to estimation range B of the kth moment RCS obtained in step 9_kIn minimum RCS numerical value λ_min, meter The signal to noise ratio numerical value ρ that the calculation kth moment is minimum_min, and to utilize algorithm of target detection to calculate be ρ in signal to noise ratio_min, detection probability ForThe highest false-alarm probability P needed under conditions of_f,max。

Described algorithm of target detection includes, matched filtering detection algorithm, CA-CFAR, order statistic CFAR, Generalized Likelihood Ratio etc., this example is selected but is not limited to matched filtering detection algorithm.

Step 11, is ρ according to obtain in step 10 in signal to noise ratio_min, detection probability beUnder conditions of required the highest False-alarm probability P_f,max, utilize algorithm of target detection to calculate this false-alarm probability P_f,maxCorresponding detection threshold T_min, and screen Go out the preliminary metric data set Z obtained in step 7₁K in (), cymoscope output data are higher than detection threshold T_minMeasurement number According to, as metric data set Z (k) in kth moment.

Described algorithm of target detection includes, matched filtering detection algorithm, CA-CFAR, order statistic CFAR, Generalized Likelihood Ratio etc., this example is selected but is not limited to matched filtering detection algorithm.

Step 12, utilizes association algorithm to choose the amount the highest with track association degree from kth moment metric data set Z (k) Survey data, the metric data chosen utilize track algorithm to calculate kth moment Target state estimator valueWith state estimation association side Difference matrix P_k。

Described association algorithm includes, nearest-neighbor algorithm, Probabilistic Data Association Algorithm, optimum Bayes's association algorithm etc., this Example is selected but is not limited to nearest-neighbor algorithm.

Step 13, it is judged that whether kth moment metric data set Z (k) is empty: if it is empty, then set to 0 by false dismissal enumerator, Otherwise false dismissal enumerator originally numerically add 1.

Step 14, it is judged that the value of false dismissal enumerator: if the value of enumerator is more than or equal to the continuous false dismissal number of times terminated needed for flight path M, then object tracking process terminates；Otherwise, step 2 is returned).

The effect of the present invention is further illustrated by the test of following simulation comparison:

1. experiment scene: use a 2D radar being positioned at zero, if carrier frequency f_c=600MHz, launches signal band Wide B=1MHz, sample frequency is F_s=1.5MHz, beam angle is 1 degree, and sweep spacing is Δ T=0.5s, and radar is surveyed Amount parameter is the distance and bearing angle of target；Target is the circular target of a diameter of 10m, and RCS scattering properties obeys scattering point Model, simulated scatter point number is taken as N_s=1500, the scattering coefficient obedience average of each scattering point is 0, variance is 1/N_s Multiple Gauss distribution, and the scattering coefficient of different scattering point is separate；If initial time target is 0km in X-axis position, Being 150km in Y-axis position, and fly at a constant speed away from radar station, the velocity component of X-axis is-200m/s, the speed of Y-axis Component is 0m/s；If initial time target RCS is 1, signal power is 15 with the ratio of noise power.Traditional detection In track algorithm, the false-alarm probability of target detection is 10^-6；Flight path termination rule is: if continuous three frames are not detected by target, then Flight path terminates, and object tracking process terminates.

2. emulation content:

Using above experiment scene, target signal to noise ratio is with the such as Fig. 2 of change the most in the same time；Utilize traditional detecting and tracking method With the detecting and tracking method of the present invention, the detection performance of radar is carried out simulation comparison, result such as Fig. 3.

3. interpretation:

By Fig. 2 and Fig. 3 it can be seen that either high s/n ratio situation or low signal-to-noise ratio situation, the present invention and tradition side Method is compared, and all can obtain higher detection probability.As a example by the 22nd moment, now target signal to noise ratio is 10 decibels of left sides The right side, the target detection probability of traditional method is 0.2593, and the target detection probability of the present invention can improve to 0.9123, phase 0.6530 is improve than traditional method.

Summary emulation experiment it can be seen that the present invention is relative to traditional detecting and tracking method, due to considered with Track device, to the information of forecasting of target and RCS information of forecasting, can ensure under conditions of not producing false track, improves The detection performance of target and flight path seriality.

Claims (6)

1. a detecting and tracking integral method based on RCS information of forecasting, comprises the steps:

1) initiation parameter: arranging false dismissal counter initial value is 0；The signals to be received such as arranging cymoscope Initial output signal is 0, cymoscope；By targetpath start algorithm could, obtain the Initial state estimation of targetpathAnd Initial state estimation covariance matrix P₀；

2) target setting state transition equation and radar measurement equation, according to kth-1 moment Target state estimator valueWith kth-1 moment state estimation covariance matrix P_k-1, calculate the predictive value that kth moment target measuresAnd the prediction covariance matrix D that kth moment target measures_k|k-1；

3) predictive value measured according to kth moment targetThe prediction covariance matrix D measured with kth moment target_k|k-1, determine the target prediction ripple door O in kth moment_k；

4) probability P that after target setting suddenly disappears, flight path correctly terminates_E, utilize equation below to calculate at kth moment target prediction ripple door O_kInterior false-alarm probability P_Z:

(1-P_Z)^M=P_E,

Wherein, M represents the continuous false dismissal number of times terminated needed for flight path；

5) weight w (i of kth moment prediction Bo Mennei i-th detector unit is set；K), equation below group is utilized to calculate kth moment prediction ripple door O_kThe false-alarm probability of interior each detector unit:

Wherein, P_f(i；K) kth moment prediction ripple door O is represented_kThe false-alarm probability of interior i-th detector unit, i=1,2 ..., N_k, N_kRepresent kth moment prediction ripple door O_kThe number of interior detector unit, P_aVariable is solved for centre；

6) according to false-alarm probability P of each detector unit_f(i；K), algorithm of target detection is utilized to calculate kth moment prediction ripple door O_kDetection threshold T of interior each detector unit_i, by the output data of cymoscope and prediction ripple door O_kDetection threshold T of interior arbitrary detector unit_iCompare, if cymoscope output data are higher than detection threshold T of this detector unit_i, then with the metric data of estimated spatial position target corresponding to this detector unit, and will prediction ripple door O_kThe all metric data inside obtained are as preliminary metric data set Z₁(k)；Otherwise, any process is not made；

7) according to the Radar Cross Section rcs measurement value in front k-1 moment, the predictive value of kth moment RCS is calculatedPrediction variance with kth moment RCSAnd according to this predictive valueWith prediction varianceDetermine estimation range B of kth moment RCS_k；

8) according to estimation range B of kth moment RCS_kIn minimum RCS numerical value λ_min, calculate kth moment minimum signal to noise ratio numerical value ρ_min, and to utilize detection algorithm to calculate be ρ in signal to noise ratio_min, lowest detection probability beThe highest false-alarm probability P needed under conditions of_f,max；

9) according to the highest false-alarm probability P_f,maxCalculate corresponding detection threshold T_min, filter out preliminary metric data set Z₁K in (), cymoscope output data are higher than detection threshold T_minMetric data, as metric data set Z (k) in kth moment；

10) from kth moment metric data set Z (k), choose the metric data the highest with track association degree, the metric data chosen utilize track algorithm to calculate kth moment Target state estimator valueWith state estimation covariance matrix P_k；

11) judge that whether kth moment metric data set Z (k) is empty: if it is empty, then set to 0 by false dismissal enumerator, on the contrary false dismissal enumerator originally numerically add 1；

12) value of false dismissal enumerator is judged: if the value of enumerator is more than or equal to the continuous false dismissal number of times M terminated needed for flight path, then object tracking process terminates；Otherwise, step 2 is returned).

Detecting and tracking integral method based on RCS information of forecasting the most according to claim 1, wherein step 2) in determine the predictive value that kth moment target measuresComprise the steps:

2a) target setting state transition equation is:

x_k=F_k|k-1x_k-1+v_k|k-1,

Wherein, x_kRepresent the dbjective state in kth moment, x_k-1Represent the dbjective state in kth-1 moment, v_k|k-1Represent the process noise in kth moment to kth-1 moment, F_k|k-1Represent the dbjective state transfer matrix in kth moment to kth-1 moment,

Wherein, Δ T represents sweep spacing；

2b) set radar measurement equation as:

z_k=h_k(x_k)+w_k,

Wherein, z_kRepresent the aim parameter measured value in kth moment, h_k() represents that the target in kth moment measures function, w_kRepresent the measurement noise in kth moment；

2c) according to kth-1 moment Target state estimator valueCalculate kth moment dbjective state predictive value

2d) according to kth moment dbjective state predictive valueCalculate the predictive value that kth moment target measures

Detecting and tracking integral method based on RCS information of forecasting the most according to claim 1, wherein step 2) in determine the prediction covariance matrix D that kth moment target measures_k|k-1, comprise the steps:

2e) according to kth moment dbjective state predictive valueCalculate kth moment target and measure the Jacobian matrix H of function_k:

Wherein,Represent vector x derivation, ()^TRepresent transposition computing,Representative function?The functional value at place；

2f) according to kth-1 moment Target state estimator covariance matrix P_k-1With the Jacobian matrix H that kth moment target measures function_k, calculate the prediction covariance matrix D that kth moment target measures_k|k-1:

Wherein, Q_k|k-1Represent the process noise covariance matrix in kth moment to kth-1 moment.

Detecting and tracking integral method based on RCS information of forecasting the most according to claim 1, wherein step 3) in determine the prediction ripple door O of kth moment target_k, comprise the steps:

3a) target setting falls into the probability of prediction ripple door

3b) determine prediction ripple door coefficient gamma by lookup chi-square distribution table₁, to ensure that degree of freedom measures card side's distribution variables of dimension more than γ for target₁Probability beThe distribution function table of card side's distribution variables during wherein chi-square distribution table is theory of probability；

3c) according to prediction ripple door coefficient gamma₁, determine prediction ripple door O as follows_k:

Wherein, D_k|k-1Represent the prediction covariance matrix that kth moment target measures, γ₁Represent ripple door coefficient,Represent the predictive value that kth moment target measures, y_kRepresent the position that kth moment target is likely to occur, | representing conditional code, the symbol left side is set element, and the right is the condition that element meets.

Detecting and tracking integral method based on RCS information of forecasting the most according to claim 1, wherein step 7) in determine the predictive value of kth moment RCS, comprise the steps:

7a) set RCS predictor exponent number L；

7b) calculate the correlation coefficient C in kth-i moment and kth-j moment target RCS as follows_k(i, j):

C_k(i, j)=[2 J₁(πu)/πu]²+ 1,

Whereinθ_i、θ_jRepresenting the radar observation angle in i Yu j moment, d represents observed object size, and c represents the light velocity, F_cRepresent carrier frequency, J₁() represents Bessel function of the first kind；

7c) according to RCS dependency, obtain RCS predictor weights by solving equation below group:

Wherein, w_k-lRepresent the RCS predictor weights in the kth-l moment；C_kRepresent RCS correlation matrix, C_kBy the correlation coefficient C of matrix element, i.e. RCS_k(i j) is constituted；

7d) according to RCS predictor at the weights in kth-1 moment to kth-L moment, calculate RCS predictive value by following formula:

Wherein, λ_k-lRepresent kth-l moment target RCS observation.

Detecting and tracking integral method based on RCS information of forecasting the most according to claim 1, wherein step 7) in determine estimation range B of kth moment RCS_k, comprise the steps:

7d) the true RCS of target setting falls into the probability of RCS estimation range

7e) determine RCS estimation range B by lookup Gauss distribution table_kIn minimum RCS numerical value λ_min, to ensure that RCS prediction variance isGaussian distributed random variable more than λ_minProbability beThe distribution function table of Gaussian distributed random variable during wherein Gauss distribution table is theory of probability；

7f) by RCS estimation range B_kIn minimum RCS numerical value λ_min, determine RCS estimation range B as the following formula_k:

B_k=[λ_min,+∞)。