CN109085571A - Hypersonic method for tracking target based on triple bayesian criterions - Google Patents
Hypersonic method for tracking target based on triple bayesian criterions Download PDFInfo
- Publication number
- CN109085571A CN109085571A CN201810950211.XA CN201810950211A CN109085571A CN 109085571 A CN109085571 A CN 109085571A CN 201810950211 A CN201810950211 A CN 201810950211A CN 109085571 A CN109085571 A CN 109085571A
- Authority
- CN
- China
- Prior art keywords
- weight
- model
- moment
- prior
- tracking
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/66—Radar-tracking systems; Analogous systems
- G01S13/72—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar Systems Or Details Thereof (AREA)
- Feedback Control In General (AREA)
Abstract
The invention belongs to hypersonic target tracking domains, and for the problem of hypersonic target following, the present invention proposes a kind of hypersonic method for tracking target based on triple bayesian criterions.Firstly, being directed to all possible forms of motion of hypersonic target, a kind of closure tracking channel based on sinusoidal model is designed, to realize that the matched jamming to hypersonic target is handled；Secondly, the approach to cooperation of different tracking channels, is designed as the function of Model Weight and prior model switching weight by the first heavy bayesian criterion of design, to improve the compatible degree of selected model and target real trace；Then, the second heavy bayesian criterion is designed, Model Weight is designed as the closed circuit with Radar Intercept information real-time update, to improve the tracking accuracy to hypersonic target；Finally, design third weight bayesian criterion, is the closed circuit with Radar Intercept information real-time update by prior model switching weight design, to further increase the sensitivity to hypersonic target following.
Description
Technical field
The present invention is under the jurisdiction of hypersonic target tracking domain, is suitable for solving hypersonic reconnaissance plane, hypersonic nothing
The tracking problem of the targets such as man-machine, hypersonic cruise missile.
Background technique
Hypersonic aircraft refers in endoatmosphere, the characteristics of using rarefied atmosphere, is done with 5 or more Mach number and is persistently flown
Capable aircraft.Such aircraft can both carry nuclear warheads in modern war, enhance the strategic deterrence ability of this country, again
Conventional weapon is carried to the property of can choose, Long-range precision strike is carried out to enemy's important goal, all kinds of active passives can also be carried
Detecting devices, quickly to be investigated to global sensitive target and early warning.In particular, under US Global formulation,
Hypersonic Weapon has become the preferred weapon that the U.S. implements Global Strike.For this purpose, how effectively to realize to hypersonic flight
The tracking interception of device has become a current critical issue urgently to be solved.
In the research intercepted to hypersonic aircraft tracking, the research of early stage regards target movement as some mostly
Tempo turn campaign or the snakelike motion of automobile in plane, and enough attention is not subject to the genuine property of target, it is
This, often has that tracking accuracy is not high.For this case, plurality of target trace model is widely used in high ultrasound
Fast target tracking domain.For example, target trajectory approximation to be seen to CV model and CA model for linear motion, emphasis is transported for circumference
Dynamic CT model, with the hypersonic highest Sine model of track S Trajectory Matching degree etc..But the shortcomings that these models is pair
The prior information of target has higher requirement, in particular, when selected model differs larger with target truth, unavoidably
Ground can introduce biggish model following error.This is difficult to realize that is, relying solely on single model to hypersonic target
Reliable tracking.
For this case, the method for multi mode tracking is effectively proposed.Such method by different models in a certain way
It combines, can solve the problems, such as single model following mismatch to a certain extent.But when selected model and target are true
When approach to cooperation between movement difference is larger or different models is unreasonable, often there is tracking accuracy decline, when serious even
Not as good as the problem of single mode tracking.
For this purpose, the present invention is directed to all possible forms of motion of target, selected by the optimization to different target trace model
It selects, the optimization design of approach to cooperation between different models, proposes a kind of hypersonic target following based on triple bayesian criterions
Method, to solve the problems, such as hypersonic target following.
Summary of the invention
It is an object of the invention to break through the limitation of conventional multi-mode tracking, in the case of solving target In Hypersonic Flow
Tracking problem, promote the ability that existing radar detection tracks hypersonic target, propose a kind of based on triple bayesian criterions
Hypersonic method for tracking target.Wherein to solve the problems, such as include:
1) it is unreasonable to have that model selects for existing hypersonic method for tracking target, can not match target institute
Possible forms of motion；
2) existing hypersonic method for tracking target has that model approach to cooperation is unreasonable, and then leads to multimode
Tracking accuracy is not high, tracking effect be not so good as single mode when the problem of；
3) existing hypersonic method for tracking target has that Model Weight updates not in time, and then there are models
The problem of mismatch, tracking accuracy declines；
4) existing hypersonic method for tracking target has that models switching weight does not update, and then exists because of mould
The model mismatch problem that type handoff error causes.
Hypersonic New Target Tracking of the present invention based on triple bayesian criterions, it is characterised in that including
Following technical measures:
Step 1: before being tracked to hypersonic target, using sinusoidal model, design and all possible fortune of target
The tracking channel that dynamic form matches；
Step 2: switch weight using posterior model, different tracking channels are designed as to the closed circuit that can be jumped in real time；
It is mould by the posterior model switching weight design of different tracking channels Step 3: building the first heavy bayesian criterion
The function of type weight and prior model switching weight；
Step 4: the second heavy bayesian criterion is built, before the Model Weight at different tracking channel current times is designed as
Model Weight, the prior model at one moment switch the function of weight and model likelihood；
Step 5: building third weight bayesian criterion, the prior model at different tracking channel current times is switched into weight
It is designed as the Model Weight of previous moment, the function of prior model switching weight and model likelihood；
Step 6: the model likelihood of different tracking channels is designed as radar and is cut using the filter tracking algorithm in each channel
Obtain the function of information.
The comparison prior art, the hypersonic New Target Tracking of the present invention based on triple bayesian criterions,
Beneficial effect is:
1) present invention is a kind of improvement to existing hypersonic target tracking algorism, in the matching all possible fortune of target
While dynamic form, moreover it is possible to optimize the approach to cooperation between different models；
2) approach to cooperation of different models is designed as the closed circuit that can be jumped in real time by the present invention, can be effectively improved selected
The compatible degree of model and target real trace；
3) present invention with the closed circuit of Radar Intercept information time-varying, can effectively improve the weight design of different models
To the tracking accuracy of hypersonic target；
4) the models switching weight design of the invention by different models is the closed circuit with Radar Intercept information time-varying, can
Effectively improve the sensitivity to hypersonic target following.
Detailed description of the invention
Fig. 1 is the hypersonic method for tracking target flow chart of steps based on triple bayesian criterions；
Fig. 2 is the close passage figure of the hypersonic target following based on sinusoidal model；
Fig. 3 is the closure update figure using the posterior model switching weight of the first weight bayesian criterion design；
Fig. 4 is the closure update figure using the Model Weight of the second weight bayesian criterion design；
Fig. 5 is the closure update figure using the prior model switching weight of third weight bayesian criterion design.
Specific implementation method
For the problem of hypersonic target following, the present invention proposes a kind of based on the hypersonic of triple bayesian criterions
Method for tracking target.Firstly, being directed to all possible forms of motion of hypersonic target, one kind is designed based on sinusoidal model
Target closed tracking channel, the matched jamming of hypersonic target is handled with realizing；Secondly, design the first weight Bayes is quasi-
Then, the approach to cooperation of different tracking channels is designed as to the function of Model Weight and prior model switching weight, selected by improving
The compatible degree of model and target real trace；Then, the second heavy bayesian criterion is designed, Model Weight is designed as cutting with radar
The closed circuit of information real-time update is obtained, to improve selected model to the tracking accuracy of hypersonic target；Finally, design third
Prior model switching weight design is the closed circuit with Radar Intercept information real-time update, with into one by weight bayesian criterion
Step improves selected sensitivity of the model to hypersonic target following.
The present invention is described in further detail below in conjunction with Figure of description.Referring to Fig.1, process flow of the invention
Divide following steps:
1) tracking channel to be matched using sinusoidal model design with all possible forms of motion of target
(1) the design of matched jamming Models Sets
In horizontal plane, selection has maximum angular frequency w1Sinusoidal model M1With there is minimum angular frequency w2Sine
Model M2, to match all possible forms of motion of the target in horizontal plane, then trace model of the target in horizontal plane
Collection may be designed as
MIt is horizontal={ M1,M2} (1)
In vertical plane, selection has maximum angular frequency w3Sinusoidal model M3With there is minimum angular frequency w4Sine
Model M4, to match all possible forms of motion of target in vertical plane, then the trace model of target in vertical plane
Collection may be designed as
MVertically={ M3,M4} (2)
In three-dimensional space, M is utilizedIt is horizontalAnd MVertically, it is by the matched jamming Models Sets comprehensive design of target
M={ MIt is horizontal,MVertically}={ M1,M2,M3,M4} (3)
To match target all possible forms of motion in three-dimensional space.
(2) build and designed model M1、M2、M3And M4The 4 target tracking channels to match
Channel 1 is by model M1And its filter tracking algorithm composition, it is in the input function at k (1,2,3 ...) moment
X1input(k), output function X1output(k)；
Channel 2 is by model M2And its filter tracking algorithm composition, it is X in the input function at k moment2input(k), it exports
Function is X2output(k)；
Channel 3 is by model M3And its filter tracking algorithm composition, it is X in the input function at k moment3input(k), it exports
Function is X3output(k)；
Channel 4 is by model M4And its filter tracking algorithm composition, it is X in the input function at k moment4input(k), it exports
Function is X4output(k)。
2) switch weight using input function, output function and posterior model, different tracking channels are designed as can be real-time
The closed circuit of jump
(1) the design of 1~4 input function of tracking channel
The input function of k moment tracking channel 1 is designed as the output function and thereafter by k-1 moment tracking channel 1~4
The combination of models switching weight is tested, then the input function of k moment tracking channel 1 may be designed as
The input function of k moment tracking channel 2 is designed as the output function and thereafter by k-1 moment tracking channel 1~4
The combination of models switching weight is tested, then the input function of k moment tracking channel 2 may be designed as
The input function of k moment tracking channel 3 is designed as the output function and thereafter by k-1 moment tracking channel 1~4
The combination of models switching weight is tested, then the input function of k moment tracking channel 3 may be designed as
The input function of k moment tracking channel 4 is designed as the output function and thereafter by k-1 moment tracking channel 1~4
The combination of models switching weight is tested, then the input function of k moment tracking channel 4 may be designed as
Wherein, Pr [M1(k-1)|M1(k)]、Pr[M2(k-1)|M1(k)]、Pr[M3(k-1)|M1(k)]、Pr[M4(k-1)|M1
It (k)] is respectively to switch weight using the posterior model of the tracking channel 1 of the first weight bayesian criterion design；Pr[M1(k-1)|M2
(k)]、Pr[M2(k-1)|M2(k)]、Pr[M3(k-1)|M2(k)]、Pr[M4(k-1)|M2It (k)] is respectively to utilize the first heavy pattra leaves
The posterior model of the tracking channel 2 of this criterion design switches weight；Pr[M1(k-1)|M3(k)]、Pr[M2(k-1)|M3(k)]、Pr
[M3(k-1)|M3(k)]、Pr[M4(k-1)|M3(k)] after the tracking channel 3 for respectively utilizing the design of the first weight bayesian criterion
Test models switching weight；Pr[M1(k-1)|M4(k)]、Pr[M2(k-1)|M4(k)]、Pr[M3(k-1)|M4(k)]、Pr[M4(k-1)
|M4It (k)] is respectively to switch weight using the posterior model of the tracking channel 4 of the first weight bayesian criterion design.
(2) the acquisition of 1~4 output function of tracking channel
In the input function X for obtaining tracking channel 1~41input(k)、X2input(k)、X3input(k) and X4input(k) base
On plinth, the output function X of tracking channel 1~4 can be further obtained using the method that kalman is filtered1output(k)、X2output
(k)、X3output(k) and X4output(k), at this moment, will tracking channel 1~4 respectively construct formation one by
X1output(k-1)→X1input(k)→X1output(k) (8)
X2output(k-1)→X2input(k)→X2output(k) (9)
X3output(k-1)→X3input(k)→X3output(k) (10)
X4output(k-1)→X4input(k)→X4output(k) (11)
Closed circuit, it is specific as shown in Figure 2.
3) the first heavy bayesian criterion is designed, the approach to cooperation of different tracking channels is designed as Model Weight and priori mould
The function of type switching weight
In tracking channel 1 posterior model switching weight design
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, after in tracking channel 1
Testing models switching weight design is
Wherein, Pr [M1(k-1)]、Pr[M2(k-1)]、Pr[M3And Pr [M (k-1)]4It (k-1)] is respectively to track at the k-1 moment
The Model Weight in channel 1~4；Pr[M1(k)|M1(k-1)]、Pr[M1(k)|M2(k-1)]、Pr[M1(k)|M3And Pr [M (k-1)]1
(k)|M4(k-1)] be respectively k-1 moment tracking channel 1 prior model switching weight；
In tracking channel 2 posterior model switching weight design
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, after in tracking channel 2
Testing models switching weight design is
Wherein, Pr [M2(k)|M1(k-1)]、Pr[M2(k)|M2(k-1)]、Pr[M2(k)|M3And Pr [M (k-1)]2(k)|M4
(k-1)] be respectively k-1 moment tracking channel 2 prior model switching weight；
In tracking channel 3 posterior model switching weight design
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, after in tracking channel 3
Testing models switching weight design is
Wherein, Pr [M3(k)|M1(k-1)]、Pr[M3(k)|M2(k-1)]、Pr[M3(k)|M3And Pr [M (k-1)]3(k)|M4
(k-1)] be respectively k-1 moment tracking channel 3 prior model switching weight；
In tracking channel 4 posterior model switching weight design
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, after in tracking channel 4
Testing models switching weight design is
Wherein, Pr [M4(k)|M1(k-1)]、Pr[M4(k)|M2(k-1)]、Pr[M4(k)|M3And Pr [M (k-1)]4(k)|M4
(k-1)] be respectively k-1 moment tracking channel 4 prior model switching weight, it is specific as shown in Figure 3.
4) the second heavy bayesian criterion is designed, Model Weight is designed as returning with the closure of Radar Intercept information real-time update
Road
(1) model M1The design of weight
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Second heavy bayesian criterion, by k moment model M1Weight design be
Wherein, Pr [Z (k) | M1(k)]、Pr[Z(k)|M2(k)]、Pr[Z(k)|M3(k)] and Pr [Z (k) | M4(k)] respectively
For k moment model M1~M4Model likelihood；
(2) model M2The design of weight
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Second heavy bayesian criterion, by k moment model M2Weight design be
(3) model M3The design of weight
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Second heavy bayesian criterion, by k moment model M3Weight design be
(4) model M4The design of weight
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Second heavy bayesian criterion, by k moment model M4Weight design be
It is specific as shown in Figure 4.
5) prior model switching weight design is with Radar Intercept information real-time update by design third weight bayesian criterion
Closed circuit
(1) prior model switches weight Pr [M1(k)|M1(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment1(k+1)|M1(k)] it is designed as
(2) prior model switches weight Pr [M1(k)|M2(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment1(k+1)|M2(k)] it is designed as
(3) prior model switches weight Pr [M1(k)|M3(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment1(k+1)|M3(k)] it is designed as
(4) prior model switches weight Pr [M1(k)|M4(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment1(k+1)|M4(k)] it is designed as
(5) prior model switches weight Pr [M2(k)|M1(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment2(k)|M1(k-1)] it is designed as
(6) prior model switches weight Pr [M2(k)|M2(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment2(k)|M2(k-1)] it is designed as
(7) prior model switches weight Pr [M2(k)|M3(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment2(k)|M3(k-1)] it is designed as
(8) prior model switches weight Pr [M2(k)|M4(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment2(k)|M4(k-1)] it is designed as
(9) prior model switches weight Pr [M3(k)|M1(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment3(k)|M1(k-1)] it is designed as
(10) prior model switches weight Pr [M3(k)|M2(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment3(k)|M2(k-1)] it is designed as
(11) prior model switches weight Pr [M3(k)|M3(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment3(k)|M3(k-1)] it is designed as
(12) prior model switches weight Pr [M3(k)|M4(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment3(k)|M4(k-1)] it is designed as
(13) prior model switches weight Pr [M4(k)|M1(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment4(k)|M1(k-1)] it is designed as
(14) prior model switches weight Pr [M4(k)|M2(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment4(k)|M2(k-1)] it is designed as
(15) prior model switches weight Pr [M4(k)|M3(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment4(k)|M3(k-1)] it is designed as
(16) prior model switches weight Pr [M4(k)|M4(k-1)] design
It is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Third weight bayesian criterion, switches weight Pr [M for the prior model at k moment4(k)|M4(k-1)] it is designed as
It is specific as shown in Figure 5.
Claims (6)
1. the hypersonic method for tracking target based on triple bayesian criterions, which comprises the following steps:
Step 1: before being tracked to hypersonic target, using sinusoidal model, building and all possible movement shape of target
The tracking channel that formula matches；
Step 2: switching weight using input function, output function and posterior model, different tracking channels are configured to can be real-time
The closed circuit of jump；
Step 3: building the first heavy bayesian criterion, the posterior model switching weight of different tracking channels is configured to model power
The function of weight and prior model switching weight；
Step 4: the second heavy bayesian criterion is built, when the Model Weight at different tracking channel current times is configured to previous
Model Weight, the prior model at quarter switch the function of weight and model likelihood；
Step 5: third weight bayesian criterion is built, by the prior model switching weight building at different tracking channel current times
Switch the function of weight and model likelihood for Model Weight, the prior model of previous moment；
Step 6: the model likelihood of different tracking channels is configured to Radar Intercept letter using the filter tracking algorithm in each channel
The function of breath.
2. the hypersonic method for tracking target according to claim 1 based on triple bayesian criterions, which is characterized in that
The construction method of tracking channel in step 1 specifically:
1) building of matched jamming Models Sets
In horizontal plane, selection has maximum angular frequency w1Sinusoidal model M1With there is minimum angular frequency w2Sinusoidal model
M2, to match all possible forms of motion of the target in horizontal plane, then trace model collection of the target in horizontal plane can
It is configured to
MIt is horizontal={ M1,M2}
In vertical plane, selection has maximum angular frequency w3Sinusoidal model M3With there is minimum angular frequency w4Sinusoidal model
M4, to match all possible forms of motion of target in vertical plane, then the trace model collection of target in vertical plane can
It is configured to
MVertically={ M3,M4}
In three-dimensional space, M is utilizedIt is horizontalAnd MVertically, the matched jamming Models Sets synthesis of target is configured to
M={ MIt is horizontal,MVertically}={ M1,M2,M3,M4}
To match target all possible forms of motion in three-dimensional space；
2) it builds and constructed model M1、M2、M3And M4The 4 target tracking channels to match
Channel 1 is by model M1And its filter tracking algorithm composition, it is X in the input function at k (1,2,3 ...) moment1input
(k), output function X1output(k),
Channel 2 is by model M2And its filter tracking algorithm composition, it is X in the input function at k moment2input(k), output function is
X2output(k),
Channel 3 is by model M3And its filter tracking algorithm composition, it is X in the input function at k moment3input(k), output function is
X3output(k),
Channel 4 is by model M4And its filter tracking algorithm composition, it is X in the input function at k moment4input(k), output function is
X4output(k)。
3. the hypersonic method for tracking target according to claim 2 based on triple bayesian criterions, which is characterized in that
Switch weight using input function, output function and posterior model in step 2, different tracking channels are configured to jump in real time
The method of the closed circuit of change are as follows:
1) building of 1~4 input function of tracking channel
The input function of k moment tracking channel 1 is configured to the output function and its posteriority mould by k-1 moment tracking channel 1~4
Type switches the combination of weight, then the input function of k moment tracking channel 1 can be configured such that
X1input(k)=X1output(k-1)Pr[M1(k-1)|M1(k)]+X2output(k-1)Pr[M2(k-1)|M1(k)]+X3output
(k-1)Pr[M3(k-1)|M1(k)]+X4output(k-1)Pr[M4(k-1)|M1(k)]
The input function of k moment tracking channel 2 is configured to the output function and its posteriority mould by k-1 moment tracking channel 1~4
Type switches the combination of weight, then the input function of k moment tracking channel 2 can be configured such that
X2input(k)=X1output(k-1)Pr[M1(k-1)|M2(k)]+X2output(k-1)Pr[M2(k-1)|M2(k)]+X3output
(k-1)Pr[M3(k-1)|M2(k)]+X4output(k-1)Pr[M4(k-1)|M2(k)]
The input function of k moment tracking channel 3 is configured to the output function and its posteriority mould by k-1 moment tracking channel 1~4
Type switches the combination of weight, then the input function of k moment tracking channel 3 can be configured such that
X3input(k)=X1output(k-1)Pr[M1(k-1)|M3(k)]+X2output(k-1)Pr[M2(k-1)|M3(k)]+X3output
(k-1)Pr[M3(k-1)|M3(k)]+X4output(k-1)Pr[M4(k-1)|M3(k)]
The input function of k moment tracking channel 4 is configured to the output function and its posteriority mould by k-1 moment tracking channel 1~4
Type switches the combination of weight, then the input function of k moment tracking channel 4 can be configured such that
X4input(k)=X1output(k-1)Pr[M1(k-1)|M4(k)]+X2output(k-1)Pr[M2(k-1)|M4(k)]+X3output
(k-1)Pr[M3(k-1)|M4(k)]+X4output(k-1)Pr[M4(k-1)|M4(k)]
Wherein, Pr [M1(k-1)|M1(k)]、Pr[M2(k-1)|M1(k)]、Pr[M3(k-1)|M1(k)]、Pr[M4(k-1)|M1(k)]
Respectively model M in tracking channel 11~M4Posterior model switch weight；Pr[M1(k-1)|M2(k)]、Pr[M2(k-1)|M2
(k)]、Pr[M3(k-1)|M2(k)]、Pr[M4(k-1)|M2It (k)] is respectively model M in tracking channel 21~M4Posterior model cut
Change weight；Pr[M1(k-1)|M3(k)]、Pr[M2(k-1)|M3(k)]、Pr[M3(k-1)|M3(k)]、Pr[M4(k-1)|M3(k)] divide
It Wei not model M in tracking channel 31~M4Posterior model switch weight；Pr[M1(k-1)|M4(k)]、Pr[M2(k-1)|M4
(k)]、Pr[M3(k-1)|M4(k)]、Pr[M4(k-1)|M4It (k)] is respectively model M in tracking channel 41~M4Posterior model cut
Change weight；
2) acquisition of 1~4 output function of tracking channel
In the input function X for obtaining tracking channel 1~41input(k)、X2input(k)、X3input(k) and X4input(k) on the basis of,
The output function X of tracking channel 1~4 can be further obtained using the method that kalman is filtered1output(k)、X2output(k)、
X3output(k) and X4output(k), at this moment, will tracking channel 1~4 respectively construct formation one by
X1output(k-1)→X1input(k)→X1output(k)
X2output(k-1)→X2input(k)→X2output(k)
X3output(k-1)→X3input(k)→X3output(k)
X4output(k-1)→X4input(k)→X4output(k)
Closed circuit.
4. the hypersonic method for tracking target according to claim 3 based on triple bayesian criterions, which is characterized in that
The first heavy bayesian criterion is utilized in step 3, by the posterior models of different tracking channels switching weight be configured to Model Weight and
The method that prior model switches the function of weight are as follows:
1) in tracking channel 1 posterior model switching weight building
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, by the posteriority mould in tracking channel 1
Type switching weight is configured to
Wherein, Pr [M1(k-1)]、Pr[M2(k-1)]、Pr[M3And Pr [M (k-1)]4It (k-1)] is respectively k-1 moment model M1~
M4Weight；Pr[M1(k)|M1(k-1)]、Pr[M1(k)|M2(k-1)]、Pr[M1(k)|M3And Pr [M (k-1)]1(k)|M4(k-
1)] be respectively k-1 moment tracking channel 1 prior model switching weight；
2) in tracking channel 2 posterior model switching weight building
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, by the posteriority mould in tracking channel 2
Type switching weight is configured to
Wherein, Pr [M2(k)|M1(k-1)]、Pr[M2(k)|M2(k-1)]、Pr[M2(k)|M3And Pr [M (k-1)]2(k)|M4(k-
1)] be respectively k-1 moment tracking channel 2 prior model switching weight；
3) in tracking channel 3 posterior model switching weight building
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, by the posteriority mould in tracking channel 3
Type switching weight is configured to
Wherein, Pr [M3(k)|M1(k-1)]、Pr[M3(k)|M2(k-1)]、Pr[M3(k)|M3And Pr [M (k-1)]3(k)|M4(k-
1)] be respectively k-1 moment tracking channel 3 prior model switching weight；
4) in tracking channel 4 posterior model switching weight building
The first heavy bayesian criterion is built using Model Weight and prior model switching weight, by the posteriority mould in tracking channel 4
Type switching weight is configured to
Wherein, Pr [M4(k)|M1(k-1)]、Pr[M4(k)|M2(k-1)]、Pr[M4(k)|M3And Pr [M (k-1)]4(k)|M4(k-
1)] be respectively k-1 moment tracking channel 4 prior model switching weight.
5. according to claim 4, the hypersonic method for tracking target based on triple bayesian criterions, feature exists
In, in step 4 utilize the second heavy bayesian criterion, Model Weight is configured to the closure with Radar Intercept information real-time update
The method in circuit are as follows:
1) model M1The building of weight
Second is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Weight bayesian criterion, by k moment model M1Weight be configured to
Wherein, Pr [Z (k) | M1(k)]、Pr[Z(k)|M2(k)]、Pr[Z(k)|M3(k)] and Pr [Z (k) | M4(k)] when being respectively k
Carve model M1~M4Model likelihood；
2) model M2The building of weight
Second is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Weight bayesian criterion, by k moment model M2Weight be configured to
3) model M3The building of weight
Second is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Weight bayesian criterion, by k moment model M3Weight be configured to
4) model M4The building of weight
Second is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
Weight bayesian criterion, by k moment model M4Weight be configured to
6. according to claim 5, the hypersonic method for tracking target based on triple bayesian criterions, feature exists
In, using third weight bayesian criterion in step 5, the method that building prior model switches weight are as follows:
1) prior model switches weight Pr [M1(k)|M1(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion1(k+1)|M1(k)] it is configured to
2) prior model switches weight Pr [M1(k)|M2(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion1(k+1)|M2(k)] it is configured to
3) prior model switches weight Pr [M1(k)|M3(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion1(k+1)|M3(k)] it is configured to
4) prior model switches weight Pr [M1(k)|M4(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion1(k+1)|M4(k)] it is configured to
5) prior model switches weight Pr [M2(k)|M1(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion2(k)|M1(k-1)] it is configured to
6) prior model switches weight Pr [M2(k)|M2(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion2(k)|M2(k-1)] it is configured to
7) prior model switches weight Pr [M2(k)|M3(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion2(k)|M3(k-1)] it is configured to
8) prior model switches weight Pr [M2(k)|M4(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion2(k)|M4(k-1)] it is configured to
9) prior model switches weight Pr [M3(k)|M1(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion3(k)|M1(k-1)] it is configured to
10) prior model switches weight Pr [M3(k)|M2(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion3(k)|M2(k-1)] it is configured to
11) prior model switches weight Pr [M3(k)|M3(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion3(k)|M3(k-1)] it is configured to
12) prior model switches weight Pr [M3(k)|M4(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion3(k)|M4(k-1)] it is configured to
13) prior model switches weight Pr [M4(k)|M1(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion4(k)|M1(k-1)] it is configured to
14) prior model switches weight Pr [M4(k)|M2(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion4(k)|M2(k-1)] it is configured to
15) prior model switches weight Pr [M4(k)|M3(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion4(k)|M3(k-1)] it is configured to
16) prior model switches weight Pr [M4(k)|M4(k-1)] building
Third is built using the model likelihood of the Model Weight at k-1 moment, the prior model at k-1 moment switching weight and k moment
The prior model at k moment is switched weight Pr [M by weight bayesian criterion4(k)|M4(k-1)] it is configured to
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810950211.XA CN109085571B (en) | 2018-08-20 | 2018-08-20 | Hypersonic target tracking method based on triple Bayesian criterion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810950211.XA CN109085571B (en) | 2018-08-20 | 2018-08-20 | Hypersonic target tracking method based on triple Bayesian criterion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109085571A true CN109085571A (en) | 2018-12-25 |
| CN109085571B CN109085571B (en) | 2020-04-21 |
Family
ID=64793840
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810950211.XA Active CN109085571B (en) | 2018-08-20 | 2018-08-20 | Hypersonic target tracking method based on triple Bayesian criterion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109085571B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117269951A (en) * | 2023-08-21 | 2023-12-22 | 中国电子科技集团公司第五十四研究所 | Target tracking method for air-ground multi-view information enhancement |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005043333A (en) * | 2003-07-19 | 2005-02-17 | National Institute Of Information & Communication Technology | Method and system for reverse estimation of ocean wave direction spectrum from radar image |
| CN104833964A (en) * | 2015-04-28 | 2015-08-12 | 南京邮电大学 | Object detection tracking integrated method and device aiming at radar/sonar system |
| CN105842686A (en) * | 2016-03-21 | 2016-08-10 | 西安电子科技大学 | Fast TBD detection method based on particle smoothness |
| CN107315172A (en) * | 2017-07-10 | 2017-11-03 | 中国人民解放军海军航空工程学院 | The strong maneuvering target tracking method of three dimensions based on intelligent sub-band filter |
| CN107783111A (en) * | 2017-09-25 | 2018-03-09 | 西安电子科技大学 | A kind of radar foresight super-resolution imaging method based on maximum entropy criterion |
-
2018
- 2018-08-20 CN CN201810950211.XA patent/CN109085571B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005043333A (en) * | 2003-07-19 | 2005-02-17 | National Institute Of Information & Communication Technology | Method and system for reverse estimation of ocean wave direction spectrum from radar image |
| CN104833964A (en) * | 2015-04-28 | 2015-08-12 | 南京邮电大学 | Object detection tracking integrated method and device aiming at radar/sonar system |
| CN105842686A (en) * | 2016-03-21 | 2016-08-10 | 西安电子科技大学 | Fast TBD detection method based on particle smoothness |
| CN107315172A (en) * | 2017-07-10 | 2017-11-03 | 中国人民解放军海军航空工程学院 | The strong maneuvering target tracking method of three dimensions based on intelligent sub-band filter |
| CN107783111A (en) * | 2017-09-25 | 2018-03-09 | 西安电子科技大学 | A kind of radar foresight super-resolution imaging method based on maximum entropy criterion |
Non-Patent Citations (3)
| Title |
|---|
| LI LIN 等: ""A TBD Algorithm for Near Space Hypersonic Target under the Condition of Range Ambiguity"", 《2016 2ND IEEE INTERNATIONAL CONFERENCE ON COMPUTER AND COMMUNICATIONS》 * |
| 任伟建 等: ""基于粒子滤波和贝叶斯估计的目标跟踪"", 《大庆石油学院学报》 * |
| 徐涛 等: ""基于贝叶斯推理的目标跟踪"", 《计算机仿真》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117269951A (en) * | 2023-08-21 | 2023-12-22 | 中国电子科技集团公司第五十四研究所 | Target tracking method for air-ground multi-view information enhancement |
| CN117269951B (en) * | 2023-08-21 | 2024-03-26 | 中国电子科技集团公司第五十四研究所 | Target tracking method for air-ground multi-view information enhancement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109085571B (en) | 2020-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104237879B (en) | A kind of multi-object tracking method in radar system | |
| CN103439697B (en) | Target detection method based on dynamic programming | |
| CN104077601A (en) | Method for carrying out synthetic target recognition through information of different types | |
| CN104021292B (en) | Dim target detection and tracking method based on formation active networking | |
| CN108107423A (en) | Distributed networked Radar Targets'Detection Tracking Integrative processing method | |
| CN105093215A (en) | Doppler information based method for tracking low-altitude low-speed small target through radar | |
| CN114859339A (en) | Multi-target tracking method based on millimeter wave radar | |
| CN106526559A (en) | Maneuvering target passive coherent location method based on BCD-VSMM | |
| CN104680002B (en) | A kind of distributed fusion method theoretical based on random set | |
| CN105044710A (en) | Tracking initiation method including radial velocity information for a circular motion target | |
| RU2570111C1 (en) | Apparatus for radar recognition of aerospace objects | |
| CN108919254A (en) | The CS-PHD method of the motor-driven small and weak multi-target detection tracking of Gao Zhongying radar | |
| CN107238835B (en) | Formation target point navigation anti-crossing association method | |
| CN113702940B (en) | Spatial cluster target resolution method based on multi-element characteristic information hierarchical fusion and application | |
| CN109085571A (en) | Hypersonic method for tracking target based on triple bayesian criterions | |
| Zhang et al. | Precession parameter estimation of warhead with fins based on micro-Doppler effect and radar network | |
| CN110488273A (en) | A kind of vehicle tracking detection method and device based on radar | |
| CN105548985A (en) | Maneuvering target tracking method based on RAV-Jerk model | |
| CN108535704A (en) | A kind of signal Pre-sorting method based on self-adaption two-dimensional cluster | |
| Frencl et al. | Tracking with range rate measurements: turn rate estimation and particle filtering | |
| CN109188420A (en) | Narrow-band Radar method for tracking target based on depth shot and long term memory network | |
| CN109031278A (en) | It is a kind of based on amplitude to extra large Radar Multi Target fusion tracking method | |
| Yang et al. | Particle swarm optimization algorithm for passive multi-target tracking | |
| CN104237880B (en) | Structure changes Joint Probabilistic Data Association formation target tracking method | |
| Wang et al. | Research on multi-maneuvering target tracking JPDA algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |