CN111708982B - Association complexity measurement method in target tracking problem - Google Patents
Association complexity measurement method in target tracking problem Download PDFInfo
- Publication number
- CN111708982B CN111708982B CN202010560329.9A CN202010560329A CN111708982B CN 111708982 B CN111708982 B CN 111708982B CN 202010560329 A CN202010560329 A CN 202010560329A CN 111708982 B CN111708982 B CN 111708982B
- Authority
- CN
- China
- Prior art keywords
- complexity
- target tracking
- target
- time
- calculating
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention relates to a method for measuring association complexity in a target tracking problem,belonging to the field of target tracking data processing. In order to solve the problem that the complexity of stable and accurate tracking of a high maneuvering target cannot be described quantitatively, the target maneuvering complexity in the target tracking problem is defined as the allowable upper limit E of a filtering position estimation error not greater than a tracking estimation error at a certain moment 0 And the ratio of the theoretical maximum allowable sigma of the measured positioning error to satisfy this condition, i.e. E 0 And/sigma. Under a Kalman filtering framework, a calculation formula of the complexity is deduced by innovatively introducing a method of state updating error reduction factor lambda, and specific algorithm steps are given. The algorithm is utilized to provide a unified and standardized identification standard for the scene difficulty of the target tracking problem, quantitatively analyze the influence of the scene parameters on the target tracking problem, and support scientific and reasonable algorithm equivalent test experimental design.
Description
Technical Field
The invention belongs to the field of target tracking data processing, and particularly relates to a method for measuring correlation complexity in a target tracking problem.
Background
The filtering algorithm is an algorithm for inhibiting random noise of observation data by introducing a motion model, however, the motion model is difficult to be completely matched with actual motion when a target does complex high-mobility actions, the model error is large, and the estimation error is large, even the filtering is divergent. The stable and accurate tracking of the high maneuvering target is a difficult problem in the field of target tracking, and the complexity of the method is a comprehensive action result of coupling of factors such as maneuvering characteristics of the target, detection data rate of a sensor, measurement errors and the like. However, at present, there is no unified and standardized definition for the complexity of the maneuvering target tracking problem itself, so that different researches define maneuvering targets under the own speech system, the complexity of the researched problems cannot be described quantitatively, and the problem of knowing who researches is difficult to say is more complicated.
Disclosure of Invention
Technical problem to be solved
The invention aims to solve the technical problem of how to provide a method for measuring the association complexity in the target tracking problem so as to overcome the problem that the complexity of stably and accurately tracking a high maneuvering target cannot be described quantitatively.
(II) technical scheme
In order to solve the above technical problem, the present invention provides a method for measuring association complexity in a target tracking problem, which comprises the following steps:
step 3, calculating the target tracking data updating time T k-1 To T k Velocity mean change vectorWherein n is a group satisfying T k-1 ≤t j <T k T of the condition j Number of (c), v (t) k ) N is the motion velocity vector of the target;
wherein I is an identity matrix;
wherein H is (I) 3×3 0 3×3 ) T ;
And step 10, calculating each sampling moment in the analysis data time period, and stopping the calculation, otherwise, returning to the step 1.
Further, before the step 1, a step of inputting parameters is further included, where the inputted parameters include:
(1) true value vector x (t) of target motion track k ),k=1,...,m,t k Sampling time of true value data of the flight path;
(2) motion velocity vector v (t) of object k ),k=1,...,m;
(3) And updating the time step deltat of the target tracking data.
Further, E 0 The upper limit of position error that is allowed to be reached for tracking filtering.
Further, 0 < C prd (k) < 1, complexity C prd (k) Larger means more complex maneuvering target tracking.
Further, the following steps are also included between step 6 and step 7: calculating a threshold value E meeting the position error requirement 0 Position covariance of
(III) advantageous effects
The invention provides an algorithm for measuring the tracking complexity of a maneuvering target in a target tracking problem so as to reflect the difficulty of the target tracking problem from measurement error, data frequency, observation geometry and maneuvering. The invention innovatively defines the target maneuvering complexity in the target tracking problem as the allowable upper limit E of the filtering position estimation error not greater than the tracking estimation error at a certain moment 0 And is fullMeasuring the ratio of the maximum allowable theoretical value sigma of the positioning error, i.e. E, for this condition 0 And/sigma. Under a Kalman filtering framework, a calculation formula of the complexity is deduced by innovatively introducing a method of state updating error reduction factor lambda, and specific algorithm steps are given.
The algorithm provides a unified and standardized scene difficulty identification standard of the target tracking problem, quantitatively analyzes the influence of scene parameters on the target tracking problem, and supports scientific and reasonable algorithm equivalent test experimental design. The algorithm can be applied to target tracking system index demonstration, and belongs to one of basic problems in target tracking algorithm test evaluation in a tracking algorithm test design process.
Drawings
FIG. 1 is a main body implementation step of the associative complexity metric algorithm of the present invention;
FIG. 2 shows traces 1 and 2 in example 1 of the present invention;
FIG. 3 is a graph showing the average variation of the target tracking data at the time of updating and the average variation of the target tracking data at the previous time;
FIG. 4 is a plot of the tracking scene complexity for track 1 at a sensor data rate of 0.1Hz in accordance with the present invention;
FIG. 5 is a complexity curve of four scenarios in embodiment 1 of the present invention;
FIG. 6(1) is a complex curve of trace 1 with respect to three thresholds in example 2 of the present invention;
FIG. 6(2) is a complex curve of trace 2 with respect to three thresholds in example 2 of the present invention;
FIG. 7 is a flow chart of an algorithm for a maneuvering target tracking complexity metric in the target tracking problem of the present invention.
Detailed Description
In order to make the objects, contents and advantages of the present invention clearer, the following detailed description of the embodiments of the present invention will be made in conjunction with the accompanying drawings and examples.
Symbol convention:
(1) thickening a body into a vector or matrix
(2) Italics without bolding to scalar quantity
(3) Default vector is column vector, and transpose is added to row vector
(4) I is an identity matrix, I' is a matrix with the last right diagonal of 1 and the rest of 0
Basic principle: (ii):
defining: for a measurement positioning system, when a Kalman filtering method is adopted to estimate a target position, a certain moment ensures that the estimation error of a filtering position is not more than the allowable upper limit E of the tracking estimation error 0 If the allowable theoretical maximum value of the positioning error measured at the moment is sigma, the complexity of the maneuvering target tracking at the moment is defined as
Description of the definitions: the definition starts from the working principle of a filtering algorithm, and the filtering position estimation is the maximum likelihood estimation under the condition of comprehensively considering model prediction errors and measurement data errors. By reducing the measured position error σ when the model error increases (target maneuvers), the filtered position estimate error E can be maintained 0 And is not changed. So the closer σ is to E 0 The larger the error of the model (the stronger the target maneuvering), the full failure of the model when the maneuvering complexity reaches the maximum, and the position estimation of the target can be provided only by measuring and positioning, i.e. sigma-E 0 It indicates that the target maneuver complexity is at a maximum at this time. From the above analysis, 0 < C mav <1,C mav Larger means more complex maneuvering target tracking.
(iii) algorithmic derivation explanations
The following explains a specific derivation calculation method of the complexity of the maneuvering target tracking. Set forth the following tracking system
Equation of motion:
x k =F(x k-1 +A k-1 ) (19)
the measurement equation:
z k =Hx k +ε, (20)
for the state transition matrix, Δ t is the filter time step (sensor data update rate),
a target maneuver item is represented that is associated with the maneuver,measuring the position location for the detection system, H ═ I 3×3 0 3×3 ) T To measure the matrix,. epsilon.is the measurement error. A is to be k-1 And ε is considered a random variable, note cov (A) k-1 )=Q k-1 ,cov(ε)=R。
WhereinThe covariance is predicted for the state,the covariance is updated for the state. Are respectively provided withAndthe upper left corner 3 × 3 sub-matrix ofAndwhich are the covariances representing the position estimate. Order to
Wherein E 0 The upper bound of position error that is allowed to be reached for tracking filtering. The purpose of the equations (5,6) is to determine the ratio lambda of the position covariance to be reduced for the state update at time k based on the prediction covariance at time k and the desired upper bound of the filtering error at time k, and then estimate the coincidence E 0 Maximum position covariance matrix of upper bound of errorFrom this upper bound, the maximum measured covariance that can meet this requirement is calculated below. Because H ═ I, O]Substituting H into equation (4), take the upper left 3 × 3 sub-matrix, so
R k Is the measurement error at time k;
Is simplified to obtain
In addition, the updated covariance is observed as
The predicted covariance at the next time instant is
Therefore, the time of each moment can be obtained by iterative calculation according to the formulas (1) to (11)
Algorithm calculation step
Inputting:
(1) true value vector x (t) of target motion track k ),k=1,...,m,t k Sampling time of true value data of the flight path;
(2) motion velocity vector v (t) of object k ),k 1.. m, if not x (t) can pass k ) Obtaining difference;
(3) updating time step delta t of data tracked by the target;
(4) a matrix H, a measurement error covariance matrix R and a sensor deployment position S;
the metric algorithm comprises the following steps:
Step 2: computing state transition matrices
And step 3: calculating the target tracking data updating time T k-1 To T k Velocity mean change vectorWherein n is T k-1 ≤t j <T k T of the condition j Number of (2)
And 4, step 4: estimating a state transition error covariance matrix Q for a maneuver
Step 6: calculated to satisfy the position error E 0 The required position error reduction ratio at time k
And 7: calculating the satisfied position error E 0 Position covariance of
And 8: calculating a satisfied position error E 0 Measured positioning covariance of
And step 9: output T k Complexity of time of day
Step 10: updating state covariance matrix
Step 11:returning to the step 1, calculating the next detection time T k Of the system.
Example 1
The embodiment specifically describes a complexity comparison of an association complexity measurement algorithm in a target tracking problem, which is provided by the invention, for analyzing tracking difficulty in different tracking scenes, and a simulation test is performed.
The problems are as follows: under the conditions that the positioning error threshold is 2km and the sensor data rate is 0.1Hz and 1Hz, the complexity comparison analysis of the tracking problems corresponding to four scenes for tracking maneuvering tracks 1 and 2 with two different degrees shown in figure 2 is carried out, wherein the left side is track 1, and the right side is track 2.
Step 1: first, the sensor data rate is calculated to be 0.1Hz, and the tracking scene complexity for track 1 is calculated.
1.1, setting up
WhereinThe setting of the numerical value can be properly set according to the positioning error threshold value of 2km, and the difference of the setting only causes the scene complexity numerical value to be different in the early tracking period but is not related to the initial value soon.
1.2 setting Δ t to 10, then
1.3 calculating the velocity average change vector of the target tracking data at the updating time and the last time according to the data of the track 1As shown in fig. 3.
1.4 estimating State transition error covariance matrix for maneuvering
1.6 to satisfy the position error E 0 Required position error reduction ratio at time k of 2000
1.7: calculating a position covariance satisfying position error 2000
1.8: calculating a measurement positioning covariance that satisfies a position error of 2000
1.9: output T k Complexity of time of day
1.10: updating state covariance matrix
1.11: returning to the step 1, calculating the next detection time T k Of the system. Fig. 4 shows a tracking scene complexity curve for trace 1 at a sensor data rate of 0.1 Hz.
Step 2: and (3) calculating the data rate of the sensor to be 0.1Hz according to the method in the step 1, and tracking scene complexity of the track 2.
And step 3: and (3) calculating the data rate of the sensor to be 1Hz according to the method in the step 1, and tracking scene complexity of the track 1.
And 4, step 4: and (4) calculating the tracking scene complexity of the track 2 according to the method of the step 1, wherein the data rate of the sensor is 1 Hz.
And 5: the four scene complexity curves calculated in steps 1 to 4 are plotted on a unified scale and compared for analysis, as shown in fig. 5. The complexity of the maneuvering motion of the track 1 is higher than that of the track 2, under the condition that the data rate of the sensor is 0.1Hz, the tracking complexity of the track 1 is kept above 0.65, the tracking complexity of the track 2 near a corner is similar to that of the track 1, and the tracking complexity of other time periods is below 0.35. It is shown that at a sensor data rate of 0.1Hz, the tracking complexity for trace 1 is significantly greater than the tracking complexity for trace 2. At a sensor data rate of 1Hz, although trajectory 1 maneuver complexity is greater than trajectory 2, the two tracking complexity measures are not as far apart, with trajectory 1 tracking complexity being slightly greater than trajectory 2 tracking complexity. In addition, comparing the tracking complexity of trace 1 with the sensor 0.1Hz and 1Hz data, the former is significantly larger than the latter, while for trace 2, the tracking complexity with the sensor 0.1Hz and 1Hz data, the former is significantly larger than the latter only in the turn segment, while the other segments are identical. Therefore, the measurement algorithm of the invention can carry out unified measurement on scenes matched with different sensors and target tracks under the given positioning error requirement threshold value, thereby carrying out unified comparison.
Example 2
For two tracks 1 and 2 with different maneuvering strengths, a sensor with a data rate of 0.1Hz is adopted for observation, and when the positioning error removing requirement threshold values are respectively 500m, 1000m and 2000m, the complexity of a scene on a tracking algorithm is analyzed.
The method of embodiment 1 is adopted to calculate the complexity curves of six scenes respectively, three types corresponding to the track 1 are drawn in one graph, and three types corresponding to the track 2 are drawn in one graph for analysis. As can be seen from fig. 6(1), for track 1, as the positioning error requirement threshold is decreased (i.e., the requirement for positioning accuracy is higher), the scene complexity curve is consistently raised, which indicates that track 1 is consistently changed along with the positioning error requirement threshold in the whole course. As can be seen from fig. 6(2), for track 2, only the turning maneuver section has the characteristics of track 1, while in the non-maneuver section, the scene is not sensitive to the complexity of target tracking with respect to the positioning error requirement threshold, which indicates that the tracking of the non-maneuver target is a simple problem.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (5)
1. A method for measuring correlation complexity in a target tracking problem is characterized by comprising the following steps:
Step 2, calculating a state transition matrix, wherein delta t is a data updating time step length of target tracking;
step 3, calculating the updating time T of the target tracking data k-1 To T k Velocity averagingVariation vectorWherein n is T k-1 ≤t j <T k T of the condition j Number of (c), v (t) k ) N is a motion velocity vector of the target;
step 4, estimating a state transition error covariance matrix Q of the maneuvering motion;
Step 6, calculating to meet the position error demand threshold value E 0 The required position error reduction ratio at the time k;
step 7, calculating a threshold value E meeting the position error requirement 0 The measured positioning covariance of (a);
wherein I is an identity matrix;
step 8, outputting T k Complexity of the time;
step 9, updating a state estimation covariance matrix;
wherein H is (I) 3×3 0 3×3 ) T ;
And step 10, calculating each sampling moment in the analysis data time period, and stopping the calculation, otherwise, returning to the step 1.
2. The method of correlating complexity metrics in a target tracking problem as claimed in claim 1 further comprising the step of inputting parameters prior to said step 1, said inputted parameters comprising:
(1) true value vector x (t) of target motion track k ),k=1,...,m,t k Sampling time of true track data;
(2) motion velocity vector v (t) of object k ),k=1,...,m;
(3) And updating the time step deltat of the target tracking data.
3. As claimed inThe method for measuring the association complexity in the target tracking problem of 1 is characterized in that E 0 The upper limit of position error that is allowed to be reached for tracking filtering.
4. The method of correlating complexity metrics in an object tracking problem as recited in claim 1 in which 0 < C prd (k) < 1, complexity C prd (k) Larger means more complex maneuvering target tracking.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010560329.9A CN111708982B (en) | 2020-06-18 | 2020-06-18 | Association complexity measurement method in target tracking problem |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010560329.9A CN111708982B (en) | 2020-06-18 | 2020-06-18 | Association complexity measurement method in target tracking problem |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111708982A CN111708982A (en) | 2020-09-25 |
CN111708982B true CN111708982B (en) | 2022-09-30 |
Family
ID=72541666
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010560329.9A Active CN111708982B (en) | 2020-06-18 | 2020-06-18 | Association complexity measurement method in target tracking problem |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111708982B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106933106A (en) * | 2016-05-26 | 2017-07-07 | 哈尔滨工程大学 | A kind of method for tracking target based on fuzzy control Multiple Models Algorithm |
CN108414995A (en) * | 2018-03-12 | 2018-08-17 | 中国人民解放军国防科技大学 | Gauss-Newton filtering based target tracking method under nonlinear model |
-
2020
- 2020-06-18 CN CN202010560329.9A patent/CN111708982B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106933106A (en) * | 2016-05-26 | 2017-07-07 | 哈尔滨工程大学 | A kind of method for tracking target based on fuzzy control Multiple Models Algorithm |
CN108414995A (en) * | 2018-03-12 | 2018-08-17 | 中国人民解放军国防科技大学 | Gauss-Newton filtering based target tracking method under nonlinear model |
Non-Patent Citations (1)
Title |
---|
基于SR-CKF的移动机器人动态目标跟踪算法;李朕阳等;《电子测量与仪器学报》;20160815(第08期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN111708982A (en) | 2020-09-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107728138B (en) | Maneuvering target tracking method based on current statistical model | |
CN103217175B (en) | A kind of self-adaptation volume kalman filter method | |
CN108319570B (en) | Asynchronous multi-sensor space-time deviation joint estimation and compensation method and device | |
CN106772524B (en) | A kind of agricultural robot integrated navigation information fusion method based on order filtering | |
CN112668104B (en) | Online identification method for pneumatic parameters of hypersonic aircraft | |
CN105136145A (en) | Kalman filtering based quadrotor unmanned aerial vehicle attitude data fusion method | |
CN110503071A (en) | Multi-object tracking method based on the more Bernoulli Jacob's Additive Models of variation Bayes's label | |
Deng et al. | Adaptive unscented Kalman filter for parameter and state estimation of nonlinear high-speed objects | |
CN112683269A (en) | MARG attitude calculation method with motion acceleration compensation | |
CN110209180A (en) | A kind of UAV navigation method for tracking target based on HuberM-Cubature Kalman filtering | |
Narasimhappa et al. | An innovation based random weighting estimation mechanism for denoising fiber optic gyro drift signal | |
CN110677140B (en) | Random system filter containing unknown input and non-Gaussian measurement noise | |
Yan et al. | An adaptive nonlinear filter for integrated navigation systems using deep neural networks | |
CN109447122B (en) | Strong tracking fading factor calculation method in distributed fusion structure | |
CN112906213B (en) | Self-adaptive prediction method for residual life of airborne electronic equipment | |
CN112986977B (en) | Method for overcoming radar extended Kalman track filtering divergence | |
CN105701292B (en) | A kind of parsing discrimination method of maneuvering target turning rate | |
CN111708982B (en) | Association complexity measurement method in target tracking problem | |
CN114462293A (en) | Method for predicting medium and long-term trajectory of hypersonic target | |
Li et al. | A new adaptive unscented Kalman filter based on covariance matching technique | |
CN111340853A (en) | Multi-sensor GMPHD self-adaptive fusion method based on OSPA iteration | |
CN109188422B (en) | Kalman filtering target tracking method based on LU decomposition | |
CN114637956B (en) | Method for realizing target position prediction based on double Kalman filters | |
CN113280821B (en) | Underwater multi-target tracking method based on slope constraint and backtracking search | |
CN104268597A (en) | AHCIF-based centralized measured value dimension-expansion fusion method |
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 |