CN104977022B - Multiple-target system Performance Evaluation emulation mode - Google Patents
Multiple-target system Performance Evaluation emulation mode Download PDFInfo
- Publication number
- CN104977022B CN104977022B CN201410134856.8A CN201410134856A CN104977022B CN 104977022 B CN104977022 B CN 104977022B CN 201410134856 A CN201410134856 A CN 201410134856A CN 104977022 B CN104977022 B CN 104977022B
- Authority
- CN
- China
- Prior art keywords
- algorithm
- flight path
- target
- probability
- track
- 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
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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/40—Means for monitoring or calibrating
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
A kind of multiple-target system Performance Evaluation emulation mode, employ a variety of different approach loaded targets flight paths, the diversity in targetpath path is realized, by self-defined targetpath, meets demand of the user to diversity flight path type configuration to greatest extent；Radar and infrared two kinds of sensing data sources are employed, realizes the Performance Evaluation to different multiple-target systems；Employ server/customer end pattern, evaluated algorithm may operate in client, realize algorithm secrecy, employ the mode that single is combined with multiple Monte Carlo method, the statistical property of evaluated Image Tracking Algorithms Performance can be obtained, and can realizes the debugging to being evaluated algorithm, finds the discrete point of algorithm.
Description
Technical field
The present invention relates to a kind of performance estimating method, relate more specifically to, to the algorithm performance in multiple-target system
The method that index is assessed.
Background technology
In multiple-target system, with the development of the key technologies such as data correlation and Multi-sensor Fusion, from
Single sensor multi target tracking progressively develops to multisensor, and kind of sensor is also developing into thunder from past single radar
Reach, infrared multiple species.
Found by the retrieval to prior art literature, most of is some link progress to multiple target tracking algorithm
Analysis and assessment, seldom carry out Performance Evaluation to whole algorithm, and performance improvement and optimization are such as carried out to data associated section, this
Assessment mode, evaluation index is limited, lacks third-party assessment system, not objective.Unique one is divided whole system
The article of analysis, it is that Zhu exists from modest and amiable Zhang Shifang《Fire control radar technology》Deliver on volume 32 within 2003《Multiple target tracking (MTT)
Performance evaluation system》, this article proposed to Multi -- Target Radar Tracking System performance estimating method, and it is disadvantageous in that only thunder
Up to a kind of sensor, target source track limitednumber, load mode is single, some typical flight paths is only limited to, in progressive
Multi-objective Algorithm is run while assessing, is unfavorable for the confidentiality of algorithm, not to the single-step debug work(of evaluated algorithm
Energy.And with the development of multi-sensor fusion technology, kind of sensor is also developing into radar, infrared from past single radar
Also there is Multiple target tracking system in multiple species, multiple-target system.
The content of the invention
The present invention is achieved by the following technical solutions, and concrete implementation method of the present invention is as follows：
Step 1：Targetpath configures
Configuration of the present invention to targetpath is adapted to different system using radar, infrared two kinds of input data sources
It is required that targetpath configuration can be carried out by three kinds of modes, the first is that the typical flight path of design is loaded directly into, and typical flight path is
The track documents previously generated, it is made up of various types of target formation flight paths, typical flight path includes horizontal linear, level is met
Fly, flat bank, horizontal snakelike, horizontal cross, level are spiraled, vertical snakelike, steep dive, are vertically risen to；Second is to lead
Enter outside track data file, outside track data file is the targetpath data that user obtains from other places or actual recording
Flight track, it is necessary to which track data is preserved data into corresponding data file according to the form of the track documents of regulation；
The third is self-defined targetpath, the targetpath for meeting to require by configuring target component generation, is carrying out self-defined mesh
When marking flight path, by the way of piecewise combination, a complete object flight path divide into three sections, the flight path of each effectively section is entered
Row synthesis, specific method are as follows：
(a) first paragraph initial track information, that is, multiple target initial position message by user configuration, it is necessary to configuration ginseng
Number is multiple target geographical coordinate (x, y, z), initial yaw angle (heading when target starts).By one section of user's setting
Flight path 2 is transferred to after time t1 (s)；
(b) initial information of flight path 2 is the position and course drift angle of the last frame of multi-target traces 1, flight path 2 be
Moved on the basis of the position and course drift angle of the last frame of flight path 1, it is motor-driven initial inclined if flight path 2 will carry out maneuvering flight
Boat angle is the yaw angle of the last frame of flight path 1, and flight path 3 is transferred to after a period of time t2 (s) by user's setting；
(c) as the combination of the combinatorial principle of flight path 3 and flight path 2, the relevant information of the last frame of flight path 2 is first obtained,
As the initial information of flight path 3, the 3rd section of flight path is then obtained according to the information of flight path section 3.
A unified track documents are made into targetpath generation, and the data to being deposited in track documents carry out unification
Regulation, the form of track documents is as shown in table 1, as long as the generation of other flight paths substitutes the data file can loading mesh specified
Track data file is marked, the definition of targetpath data file can ensure the diversity in targetpath source.
The track documents data of table 1
In table：
Obshu --- ----number of targets
T --- ----simulation time
The resolving step-length of T --- ----flight path generation
The x-axis data matrix of x1 --- ----target 1
The y-axis data matrix of y1 --- ----target 1
The z-axis data matrix of z1 --- ----target 1
The speed data matrix of v1 --- ----target 1
The yaw angle data matrix of f1 --- ----target 1
The angle of attack data matrix of s1 --- ----target 1
The initial time of ob1tf --- ----target 1
The termination time of ob1tl --- ----target 1
The x-axis data matrix of x2 --- ----target 2
The y-axis data matrix of y2 --- ----target 2
The z-axis data matrix of z2 --- ----target 2
The speed data matrix of v2 --- ----target 2
The yaw angle data matrix of f2 --- ----target 2
The angle of attack data matrix of s2 --- ----target 2
The initial time of ob2tf --- ----target 2
The termination time of ob2tl --- ----target 2
Step 2：Carrier aircraft flight path configures
The present invention carries out configuration accordingly, it is necessary to which the parameter set includes to the flight path of carrier aircraft：The model of carrier aircraft, carrier aircraft
Flight path is rectilinear flight, and original position is the coordinate (x, y, z) relative to geographic coordinate system, and starting velocity (m/s), starting accelerate
Spend (m/s2).The initial velocity default of carrier aircraft have to be larger than 80m/s, and elemental height is more than 500m.The track data of carrier aircraft
Also single data file is saved as, the track documents of carrier aircraft track documents and tracked target are as target source data.
Step 3：Sensor configuration
Configuration of the present invention to sensor includes radar and both infrared different types, can produce this two classes sensor
Data source, the tracking data result for radar is the three dimensional local information of target, and be target believes relative to the position of carrier aircraft
Breath, and infrared tracking data is the relative distance and angle information of target.Need to set its target to examine for radar sensor
Probability and false-alarm probability are surveyed, for infrared sensor except setting target detection probability and false-alarm probability, there is provision of infrared regard
The observation scope of field, including horizontal form angle and vertical form angle.In order to simulate real tracking environmental, it is also necessary to add
Influence of the ambient noise to tracking process.Different noise models, including uniformly distributed noise model, normal distribution can be selected
Noise model, Poisson distribution noise model.
Step 4：Loading algorithm
The present invention uses different load modes, including basic algorithm loading, local calculation for different evaluated algorithms
Method loading, remote algorithm loading.Basic algorithm loading is the existing classical track algorithm of loading, and classical track algorithm includes five
Link：Track initiation uses the Track initialization algorithm of logic-based, and tracking gate uses rectangle, and data correlation is using JPDA (joints
Probabilistic data association), filtering uses Kalman filtering, and flight path termination is used using Bayes tracking termination algorithms, basic algorithm loading
It is analyzed in other two ways；Local algorithm loading, it is the track algorithm progress developed, verified to user
Loading, by way of file replacement, local file is introduced directly into, existing algorithm file is covered, is tracked algorithm
The assessment of energy, the mode of local algorithm loading do not have confidentiality, it is not necessary to the 5th step network service is carried out, can be straight after loading
Tap into the algorithm evaluation of the step of row the 6th；If evaluated algorithm need for confidentiality processing, can use by the way of remote algorithm loads,
Remote algorithm loading uses server/customer end pattern, and by way of network service, evaluated track algorithm operates in visitor
Family end, sensing data source is handled by track algorithm of the network transmission to long-range client, then again will tracking
As a result local is returned to, remote algorithm loading can isolate evaluated algorithm, play privacy functions.
Step 5：Network service
Evaluated algorithm for needing progress secrecy processing in step 4, using udp protocol, is carried out by network service
Data send and receive, and evaluated algorithm operates in Terminal Server Client, and the assessment to algorithm completes data by network service
Transmission and reception, network service uses Socket web socket programming techniques.
Step 6：Algorithm evaluation
Assessment of the present invention to the performance indications of multiple-target system algorithm, has trace debug function, special using covering
Calot's method, single operation Monte Carlo simulation can be used as algorithm debugging to use, and single Monte Carlo simulation does not have statistical value, but
Result of the algorithm to data processing can be shown, realizes that every frame of data is shown, user can be carried out to Simulation result data
Analysis, the difference for analyzing different track algorithms at particular point, such as the crosspoint of multiple target；Multiple Monte Carlo Method can obtain
To the statistical property of evaluated Image Tracking Algorithms Performance, index that multiple Monte Carlo simulation is assessed have track initiation correct probability,
Track initiation average time, the track initiation probability of success, flight path terminate error probability, false track initial probability, data correlation
The probability of success, tracking accuracy, flight path maintain correct probability, divergence, validity, algorithm complex, pass through what these were assessed
Comprehensive analysis assesses the performance of track algorithm.
Track initiation correct probability be real goal correctly originates in M Monte Carlo simulation number and simulation times it
Than；
Track initiation average time is the average time confirmed from foundation hypothesis track to track；
The track initiation probability of success be defined as the number that real goal successfully originates in multiple Monte Carlo emulation and
The ratio of simulation times：
In formula：M is Monte Carlo simulation number, and m represents Monte Carlo simulation each time, and NT is target to be tracked
Number, j represent each target to be tracked, INI_Num(m)(j) number is successfully originated in Monte Carlo simulation each time for target j,
INI_Pro (j) is the target j numbers that success originates in M emulation；
It is the ratio that flight path terminates errors number and simulation times in M Monte Carlo simulation that flight path, which terminates error probability,；
False track initial probability is that M Monte Carlo is emulated, the automatic initial probability of false track of each emulation
Average value：
In formula：：M is Monte Carlo simulation number, and m represents Monte Carlo simulation each time, Clu_Prob(m)：Each time
The automatic initial probability of false track in emulation；
The data correlation probability of success is the number of targets of successful association with being traced the ratio between total number of targets；
Tracking accuracy is mean value errors (ME) of the target i in a Monte Carlo is emulated, and is：
In formula：J is j-th of component of state vector, and N is data length, and K represents data each time,For target following
xijPosition；
Flight path maintains correct probability relevant with probability with losing, and it is that real goal loses in multiple Monte Carlo simulation to lose with probability
With number and simulation times ratio, to M Monte Carlo simulation, the mistake of j-th of target is with rate：
In formula：M is Monte Carlo simulation number, and m represents Monte Carlo simulation each time, Los_Prob(m)(j) it is m
Secondary emulation, target j mistake is with indicating, then flight path maintenance correct probability is：1-Los_Prob(j)；
When tracking gate is rectangle, volume is：V=(2K)M,
When tracking gate is oval, volume is：
Wherein：M is observation dimension, and K is that ripple door constant is tabled look-up and can obtained, and γ is the thresholding of oval tracking gate, and S (k) is filtering
Residual covariance matrix,
Diverging rate is that the position root-mean-square error RMSE for continuously having sampled point since filtering stable moment t0 is more than diverging
Threshold value, then it is assumed that this filtering divergence, be specially：
Xyz represents three directions in formula, and subscript c represents the position at stable moment, and subscript 1 represents the position of sampled point
Put；
Availability is to terminate to have the κ values of sampled point to be more than 1 to filtering since filtering stable moment t0, then it is assumed that this filter
Ripple is invalid, and otherwise effectively, availability formula is：
In formula：NFiltering is effectiveAnd NEmulationFiltering effective degree and total simulation times are represented respectively；
Algorithm complex is program runtime t or CPU expense.
Finally caused effect is the present invention：A kind of multiple-target system performance estimating method is devised, is employed more
The different approach loaded targets flight path of kind, realizes the diversity in targetpath path, passes through self-defined targetpath, maximum limit
Degree meets demand of the user to diversity flight path type configuration；Radar and infrared two kinds of sensing data sources are employed, it is real
The Performance Evaluation to different multiple-target systems is showed；Server/customer end (C/S) pattern is employed, evaluated algorithm can
To operate in client, algorithm secrecy is realized, the mode that single is combined with multiple Monte Carlo method is employed, can obtain
The statistical property of evaluated Image Tracking Algorithms Performance, and can realize the debugging to being evaluated algorithm, find the discrete point of algorithm.
Brief description of the drawings
Fig. 1 is the system main assembly block diagram of the present invention；
Fig. 2 is the overall procedure block diagram of the present invention；
Fig. 3 is the targetpath configuration link target source generating structure block diagram of the present invention；
Fig. 4 is the targetpath configuration link targetpath generation composition figure of the present invention；
Fig. 5 is the targetpath configuration self-defined targetpath configuration flow figure of link of the present invention；
Fig. 6 is the flight path combination block diagram of the targetpath configuration self-defined targetpath of link of the present invention；
Carrier aircraft flight path product process figure when Fig. 7 is the carrier aircraft flight path configuration link mould of the present invention；
Fig. 8 is sensor of the invention configuration link radar target source product process figure；
Fig. 9 is sensor of the invention configuration link infrared target source product process figure；
Figure 10 is the loading algorithm link algorithm Path selection flow chart of the present invention；
Figure 11 is the flow chart of track algorithm of the present invention；
Figure 12 is the network service link data interaction figure of the present invention；
Figure 13 is the algorithm evaluation link Monte Carlo simulation procedure chart of the present invention.
Embodiment
Embodiments of the invention are elaborated below in conjunction with the accompanying drawings：The present embodiment using technical solution of the present invention before
Put and implemented, give embodiment in detail and specific operating process, but protection scope of the present invention is not limited to down
The embodiment stated.
The algorithm performance that the present embodiment is applied to Multiple target tracking system is assessed.
As shown in Figure 1, 2, the present embodiment includes following steps：Targetpath configuration, the configuration of carrier aircraft flight path, sensor
Configuration, loading algorithm, network service, algorithm evaluation, wherein：
Target source refers to the input data source of multiple target tracking algorithm, including Multi-target Data, carrier aircraft data, sensor number
According to.
As shown in figure 3, target source is divided into radar target source, infrared target source.By being exported after sensor configuration as detection
Target information afterwards, so the three-dimensional information [Rx, Ry, Rz] (carrier aircraft coordinate system) of radar target source output target, infrared target
Source output is two-dimensional signal [Ax, Ay] (infrared platform coordinate system), and system produces two kinds of target generation data simultaneously, can basis
Configuration information selection uses a kind of data.
As shown in figure 4, targetpath configuration is substantially carried out the configuration of targetpath source, targetpath parameter configuration, target
Flight path generation, targetpath preview.Stored after targetpath generation in the form of data file.
Targetpath source is the path of targetpath, including system path, external path, self-defined path.Target
The diversity in flight path path can facilitate user to enrich targetpath source by different approaches loaded targets flight path, increase
The flexibility of configuration.
Outside flight path is the track documents that outside preserves, and can be called after being loaded by system.Outside flight path path is
For the ease of user's loading in the flight path type of elsewhere generation.The data that outside track documents include need to meet to define
Track documents data format, outside track documents will be called as internal flight path after loading.
System flight path refers to the targetpath included inside emulation platform.Internal system flight path is some more classical mesh
Mark flight path type, have multiple target horizontal linear, flat bank, spiral, it is snakelike it is motor-driven, dive, rise to.Flight path class built in system
Type is the targetpath data file generated, as long as loading can is called, reduces user configuration targetpath
Time.
As shown in figure 5, self-defined targetpath refers to the target that user requires by configuring the satisfaction that target component generates
Flight path.User can be with the initial position of self-defined target, speed, target model, initial heading angular dimensions configuration generation inhomogeneity
The targetpath of type.Self-defined flight path path farthest adds the diversity in targetpath source, can generate a variety of
Meet the multi-target traces formation type of actual requirement.The generation of flight path is by kinetics equation presented hereinbefore, overload side
Journey, kinematical equation resolve what is got.
As shown in fig. 6, in self-defined flight path, parameter, which custom-configures, needs user all to carry out the parameter of each target
Set, user is to the configuration process of target component：
(a) selection target number, most 8 targets.By traveling through relevant parameter of the target number to the specified target
Configured, i.e. initiation parameter；
(b) this emulation total time is inputted, unit is the second；
(c) parameter configuration of each target：Initial position (x, y, z), initial velocity v, the acceleration of each section of flight path, each section
The simulation time of flight path, selected target type, target model, flight path type；
(d) due to the design principle using segmentation flight path, so the time of each section of flight path is configured, if
The time of this section of flight path is that 0 this section of flight path of expression does not have data；
(e) data can be just preserved after the completion of correctly to targetpath parameter configuration, generate flight path.
Complete and postpone according to above step matching somebody with somebody for targetpath parameter, can carry out generating the equation solution of flight path
Calculate.Targetpath generates the method using piecewise combination, if the simulation time of targetpath section is set to non-zero
The parameter of this section of flight path can effectively be configured.One complete object flight path divide into three sections by system, so finally
Also the flight path of each effectively section is synthesized.
As shown in fig. 7, the configuration of carrier aircraft flight path is the parameter configuration to carrier aircraft, set location, speed, acceleration, course angle,
Flight path type, the model of carrier aircraft of carrier aircraft are there is provision of simultaneously.Complete the parameter configuration of carrier aircraft and generate track data file
Afterwards, preview can be carried out to multiple target tracking effect under carrier aircraft platform, simulates carrier aircraft and target location is observed by sensor
Live effect, by Coordinate Conversion by under earth axes observing target location become under carrier aircraft coordinate system (body sit
Mark system).
As shown in Figure 8,9, sensor configuration is divided into radar, infrared two kinds of data source types, i.e. radar position data, infrared
Angle-data, by way of initial parameter is set, simulate the model of respective sensor.The parameter for needing to set has：Detection
Probability, false-alarm probability, interference noise, infrared field range.The generation of radar target source is by carrier aircraft model and carrier aircraft flight path meter
The track of carrier aircraft is calculated, the track of target is calculated by object module and targetpath, according to radar sensor model, according to detection
Probability and false-alarm probability, the positional information of the multiple target of detection is calculated.The generation of infrared target source is according to object module, is imitated
True simulation multi-target track, for verifying the validity of multiple target tracking algorithm.Infrared target source is generated according to sensor parameters
Configuration, by carrier aircraft model and the track of carrier aircraft track Calculation carrier aircraft, pass through object module and the rail of targetpath calculating target
Mark, interference noise can be used as by introducing background image, cloud layer, according to infrared sensor model, foundation detection probability and false-alarm probability,
Target gray model, position and the half-tone information of multiple target is calculated.
As shown in Figure 10, in loading algorithm, there are three classes in algorithm path：Basic algorithm, local algorithm, remote algorithm.Base
The loading of plinth algorithm is the existing classical track algorithm of loading, for being analyzed with other two ways；Local algorithm adds
Carry, be user is being developed, track algorithm verify loads, by way of file substitutes, local file is direct
Import, cover existing algorithm file, be tracked the assessment of algorithm performance, the mode of local algorithm loading is without secrecy
Property, it is not necessary to the 5th step network service is carried out, algorithm evaluation can be directly carried out after loading；If evaluated algorithm need for confidentiality
Processing, can be by the way of remote algorithm loading, and remote algorithm loading uses server/customer end pattern, is led to by network
The mode of letter, evaluated track algorithm operate in client, give sensing data source to long-range client by network transmission
The track algorithm at end is handled, and tracking result then is returned into local again, and remote algorithm loading can isolate evaluated calculation
Method, play privacy functions.
As shown in figure 11, the classical tracking process of this example specifically divides following five steps：
A) the track initiation stage：Here using the Track initialization algorithm of logic-based：The initial related ripple door of setting, is obtained
The measurement number of present frame is obtained, reads the transient state flight path of former frame, and then calculates measurement and the position of each transient state flight path of present frame
Error, if the frame number of certain transient state flight path is more than 1, the angle of each point mark and this transient state flight path is calculated, judged, record falls into phase
Bo Mennei measurement is closed, new transient state flight path is established in the measurement for not falling within related Bo Mennei as flight path head.For with transient state
The point mark of track association judges whether every flight path only has a relating dot mark, if an only relating dot mark, then directly protects
In the presence of in association flight path；Otherwise, the most short point mark of selected distance, which is given, associates.For every transient state flight path after association, judge
Whether have in the flight path at 3 points, if meeting at 3 points, using this transient state flight path as successfully originating flight path, and empty the transient state and navigate
Mark.
B) tracking gate：The decision rule of rectangular door is：V=(2K)M, K is ripple door constant, and tabling look-up to obtain.Here use
It is the decision rule of oval tracking gate, the tracking gate area of k-th of flight path is：M is observation dimension,
Here, M=3, C=4 π/3, if the residual vector norm of certain measuring point mark and this flight path is not more than this ripple door, this measuring point mark can
As candidate's echo.
C) data correlation：There are probability data interconnection (PDAF) algorithm and joint probability mutually according to interconnection (JPDA) algorithm, PDAF
Algorithm only decomposes to newest measurement, mainly solves the problems, such as single radar monotrack under clutter environment.JPDA algorithms
Put forward on the basis of PDAF algorithms, be a kind of good calculation for carrying out data interconnection under clutter environment to multiple target
Method.
Here the JPDA algorithms used, the measurement first to present frame produce observation and confirm matrix：Define one complete zero
Initial observation confirms matrix Q1, and it is 0 to set a flag bit, and Q1 first row is first entered as 1, judges certain using nearest neighbor method
Whether the residual vector norm of point mark and each targetpath is less than given threshold value, if meeting to require, flag bit is changed into
1, and the positional value for corresponding to this mark and targetpath in Q1 is assigned to 1, the mark is effective observation station, records this mark,
Add up effective the points of measurement.Add 1 row to be assigned to Q2 Q1 effective observation row and targetpath number, confirm square as final observation
Battle array.
Secondly, interconnection matrix is produced.To confirming that matrix is split, the interconnection matrix after fractionation meets：Each is measured
There is unique target source, i.e., any measurement does not come from a certain target, then must come from clutter；For some target, be up to one
Measure using it as target source.
Then association probability is calculated.The false number that measures of note is effective the points of measurement, and in each feasible event, calculating can
Every a line of row interconnection matrix and, this value is 1 or 0, if 1, is shown in this feasible event, and this is measured and a real goal
Association, falseness is now measured number and subtracts 1.In some feasible event, associated if measured with targetpath, calculate this amount
The residual vector norm with targetpath is surveyed, its normal distribution number is tried to achieve, feasible event is tried to achieve using joint probability calculation formula
Probability, and then try to achieve and measure and the association probability of target.
D) Kalman filtering algorithm：Prediction covariance, the gain of state are tried to achieve using Kalman filtering algorithm, in conjunction with amount
Survey and try to achieve Target state estimator and state covariance estimation with the association probability of target.
E) flight path terminates：2 extrapolation algorithms, if targetpath Bo Mennei does not put mark and fallen into, flight path is carried out straight
Line is extrapolated, if still falling into Bo Mennei without point mark for the targetpath extrapolated once, flight path is continued to extrapolate,
If fall into Bo Men still without a mark, then it is assumed that this flight path terminates.
This example carries out data exchange process such as Figure 12 of network service when evaluated track algorithm need for confidentiality is handled
It is shown.Evaluated track algorithm is run on as client software, and evaluated track algorithm and the communication assessed between software are adopted
It is udp protocol to complete, completes to communicate using WinSock packets socket.Client software passes through network interface
Obtain and assess the data source that software is sent, and complete processing of the algorithm to data source, be i.e. the tracking process of algorithm.Client will
Result data, which is returned to, assesses software progress algorithm performance assessment.
Described WinSock is the abbreviation of Window Sockets (socket).Window Sockets can be divided into three
Type, SOCK_STREAM, SOCK_DGRAM, SOCK_RAM, i.e. stream socket, packet socket and original socket
Word.Stream socket defines a kind of reliably connection-oriented service, realizes zero defect without the sequential data transfers repeated；
Packet socket defines a kind of connectionless service, and data are transmitted by separate message, are unordered, and
And do not ensure the reliability of data transfer, zero defect；Raw socket is rarely employed, and it allows to lower layer protocol such as IP, ICMP
Directly access, be mainly used in examining new protocol realization or access the new equipment configured in existing service.
This example algorithm uses Monte Carlo method when assessing, and is as shown in figure 13, the step of Monte Carlo method：(1) construct
The probabilistic model of practical problem；(2) according to the characteristics of probabilistic model, design and use reduce all kinds of methods of variance, accelerate examination
The convergence tested；(3) methods of sampling of various different distributions stochastic variables in probabilistic model is provided；(4) statistical results, provide
The solution and Accuracy extimate of problem.
It should be noted that above-described embodiment is only exemplary, rather than limitation of the present invention.It is any without departing substantially from this hair
The technical scheme of bright spirit all should be fallen under the scope of the present invention, and this including the use of occurring not in different embodiments
Same technical characteristic, various features and embodiment can be combined, to obtain beneficial effect.In addition, should not be by claim
In any reference be considered as the involved claim of limitation；The word of " comprising " one is not excluded for other claims or specification
In unlisted device or step.
Claims (5)
1. a kind of multiple-target system Performance Evaluation emulation mode, it is characterised in that comprise the steps：
Step 1：Targetpath configures
Configuration to targetpath using the loading of typical flight path, imports outside flight path using radar, infrared two kinds of input data sources
Three kinds of modes of data file or self-defined targetpath carry out targetpath configuration；Self-defined targetpath is using piecewise combination
Mode；
Step 2：Carrier aircraft flight path configures
Flight path is configured according to the loading condition that carrier aircraft is actual, it is necessary to which the parameter set includes：The model of carrier aircraft, carrier aircraft
Flight path is rectilinear flight, and original position is the coordinate (x, y, z) relative to geographic coordinate system, and starting velocity (m/s), starting accelerate
Spend (m/s2)；
Step 3：Sensor configuration
Configuration to sensor includes radar and infrared both types；Need to set its target detection general for radar sensor
Rate and false-alarm probability, need to set target detection probability, false-alarm probability, the horizontal form angle of infrared visual field for infrared sensor
With vertical form angle；Tracking environmental needs to add ambient noise, and the noise model that can be selected includes uniformly distributed noise mould
Type, normal distribution noise model, Poisson distribution noise model；
Step 4：Loading algorithm
Different load modes, including basic algorithm loading, local algorithm is used to load, be long-range for different evaluated algorithms
Algorithm loads；Basic algorithm loading is the existing classical track algorithm of loading；The loading of local algorithm, be user is being developed,
The track algorithm of checking is loaded；Need for confidentiality handles assessment algorithm by the way of remote algorithm loading, and remote algorithm adds
Load uses server/customer end pattern, and by way of network service, evaluated track algorithm operates in client, will pass
Sensor data source is handled by track algorithm of the network transmission to long-range client, then again returns to tracking result
It is local；
Step 5：Network service
Evaluated algorithm for needing progress secrecy processing in step 4, using udp protocol, data are carried out by network service
Send and receive, evaluated algorithm operates in Terminal Server Client, to algorithm assessment by network service completion data biography
Send and receive, network service uses Socket web socket programming techniques；
Step 6：Algorithm evaluation
Assessment to the performance indications of multiple-target system algorithm, using single and multiple Monte Carlo method, Simulation Evaluation
Index has track initiation correct probability, track initiation average time, the track initiation probability of success, flight path to terminate error probability, void
False track initiation probability, the data correlation probability of success, tracking accuracy, flight path maintain correct probability, divergence, validity, algorithm
Complexity, the overall performance appraisal procedure such as following formula that track algorithm uses：
A=Pqszq×Wqszq+Tqspj×Wqspj+Pqscg×Wqscg+Pzzcw×Wzzcw
+Pxjqs×Wxjqs+Pglcg×Wglcg+Pgz×Wgz+Pwczq×Wxjqs+D×WD+E×WE+C×WC
Wherein：A is track algorithm overall performance index；
PqszqFor track initiation correct probability, WqszqFor the weight of track initiation correct probability；
TqspjFor track initiation average time, WqspjFor the weight of track initiation average time；
PqscgFor the track initiation probability of success, WqscgFor the weight of the track initiation probability of success；
PzzcwError probability, W are terminated for flight pathzzcwFlight path terminates the weight of error probability；
PxjqsFor false track initial probability, WxjqsFor the weight of false track initial probability；
PglcgFor the data correlation probability of success, WglcgFor the weight of the data correlation probability of success；
PgzFor tracking accuracy, WgzFor the weight of tracking accuracy；
PwczqCorrect probability, W are maintained for flight pathxjqsThe weight of correct probability is maintained for flight path；
D is divergence, WDFor the weight of divergence；
E is validity, WEFor the weight of validity；
C is algorithm complex, WCFor the weight of algorithm complex.
2. Performance Evaluation emulation mode according to claim 1, it is characterised in that the self-defined targetpath, which uses, to be divided
Duan Zuhe mode, it is that a complete object flight path is divided into three sections, each section of flight path is synthesized；At the beginning of first paragraph flight path
Beginning information includes the initial geographical coordinate of multiple target, initial yaw angle, and the first paragraph flight path is transferred to after the time t1 by setting
Second segment flight path；The initial geographical position of target of the second segment flight path and initial yaw angle are the last of the first paragraph flight path
The target geographic position of one frame and course drift angle, the second segment flight path are transferred to the 3rd section of flight path after the time t2 by setting；
The initial geographical position of target of the 3rd section of flight path and initial yaw angle are the targets of the last frame of the second segment flight path
Geographical position and course drift angle.
3. Performance Evaluation emulation mode according to claim 2, it is characterised in that the self-defined targetpath configuration bag
Include：Selection target number, the target component is configured by traveling through the target；Input this emulation total time；It is right
The parameter of each target is configured.
4. Performance Evaluation emulation mode according to claim 3, it is characterised in that the target number is not more than 8.
5. Performance Evaluation emulation mode according to claim 3, it is characterised in that the parameter of each target includes：
Target initial position, initial velocity, the acceleration of each section of flight path, the simulation time of each section of flight path, target type, target model,
Flight path type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410134856.8A CN104977022B (en) | 2014-04-04 | 2014-04-04 | Multiple-target system Performance Evaluation emulation mode |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410134856.8A CN104977022B (en) | 2014-04-04 | 2014-04-04 | Multiple-target system Performance Evaluation emulation mode |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104977022A CN104977022A (en) | 2015-10-14 |
| CN104977022B true CN104977022B (en) | 2018-02-27 |
Family
ID=54273759
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410134856.8A Active CN104977022B (en) | 2014-04-04 | 2014-04-04 | Multiple-target system Performance Evaluation emulation mode |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104977022B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106199583A (en) * | 2016-06-30 | 2016-12-07 | 大连楼兰科技股份有限公司 | Multi-target Data coupling and the method and system followed the tracks of |
| CN108279428B (en) * | 2017-01-05 | 2020-10-16 | 武汉四维图新科技有限公司 | Map data evaluating device and system, data acquisition system, acquisition vehicle and acquisition base station |
| CN106707255B (en) * | 2017-03-31 | 2019-12-10 | 成都紫瑞青云航空宇航技术有限公司 | phased array radar simulation system and method |
| CN111292359A (en) * | 2018-12-06 | 2020-06-16 | 西安光启未来技术研究院 | Method and device for measuring performance of algorithm |
| CN111563960B (en) * | 2020-05-08 | 2023-05-16 | 北京理工大学重庆创新中心 | Space group target detection method and device based on Bayesian recurrence and storage medium |
| CN111753411B (en) * | 2020-06-12 | 2022-12-20 | 清华大学 | Cloud application reliability evaluation method considering edge cloud cooperation |
| CN112989685B (en) * | 2021-04-27 | 2021-08-24 | 北京智芯仿真科技有限公司 | Integrated circuit cloud platform electromagnetic simulation parallel method and device with secret function |
| CN113156418B (en) * | 2021-05-08 | 2023-02-24 | 中国科学院空天信息创新研究院 | Monte Carlo simulation-based radar target tracking precision prediction method |
| CN114611266B (en) * | 2021-12-02 | 2024-03-15 | 河北雄安京德高速公路有限公司 | Traffic radar tracking performance evaluation method under truth-free system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7460951B2 (en) * | 2005-09-26 | 2008-12-02 | Gm Global Technology Operations, Inc. | System and method of target tracking using sensor fusion |
| CN102147468B (en) * | 2011-01-07 | 2013-02-27 | 西安电子科技大学 | Bayesian theory-based multi-sensor detecting and tracking combined processing method |
| CN102426357B (en) * | 2011-08-11 | 2013-06-26 | 西北工业大学 | Multi-target tracking method with function of image confirmation |
| CN103218332A (en) * | 2013-03-19 | 2013-07-24 | 中国科学院声学研究所 | Array signal processing device for supporting VME (virtual machine environment) bus and HPI (host port interface) bus loading |
| CN103365266A (en) * | 2013-06-20 | 2013-10-23 | 中国科学院光电技术研究所 | Distributed intelligent photoetching device based on Agent |
-
2014
- 2014-04-04 CN CN201410134856.8A patent/CN104977022B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104977022A (en) | 2015-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104977022B (en) | Multiple-target system Performance Evaluation emulation mode | |
| Vermaak et al. | Monte Carlo filtering for multi target tracking and data association | |
| Hue et al. | Sequential Monte Carlo methods for multiple target tracking and data fusion | |
| CN105354356B (en) | A kind of air situation fusion performance evaluation system and method based on radar information simulation | |
| CN112925223B (en) | Unmanned aerial vehicle three-dimensional tracking virtual test simulation system based on visual sensing network | |
| Wang et al. | An adaptive UKF based SLAM method for unmanned underwater vehicle | |
| JP2016519288A (en) | Scalable real-time position detection based on overlapping neural networks | |
| CN106932771A (en) | A kind of radar simulation targetpath tracking and system | |
| CN109283562A (en) | Three-dimensional vehicle localization method and device in a kind of car networking | |
| CN110779518A (en) | Underwater vehicle single beacon positioning method with global convergence | |
| CN111982124B (en) | Deep learning-based three-dimensional laser radar navigation method and device in glass scene | |
| Frery et al. | Analysis of minute features in speckled imagery with maximum likelihood estimation | |
| CN106054171A (en) | Information entropy-based multi-radar node adaptive selection and tracking method | |
| CN110779519A (en) | Underwater vehicle single beacon positioning method with global convergence | |
| Yang et al. | Linking persistent scatterers to the built environment using ray tracing on urban models | |
| CN111273266A (en) | Multi-station radar positioning system | |
| CN109903367A (en) | Construct the method, apparatus and computer readable storage medium of map | |
| Zhou et al. | A Robust Detection and Optimization Approach for Delayed Measurements in UWB Particle-Filter-Based Indoor Positioning | |
| CN105891793A (en) | Radar seeker space-time correlation coherent K distribution clutter modeling method | |
| CN108304869A (en) | A kind of fusion method and system of the comprehensive magnitude information based on multiple sensors | |
| Soysal et al. | Data fusion in a multistatic radar network using covariance intersection and particle filtering | |
| US10304351B2 (en) | Buried asset detection simulator using retrofitted electromagnetic locate device | |
| Hollinger et al. | Coordinated search in cluttered environments using range from multiple robots | |
| CN113093092B (en) | Underwater robust self-adaptive single beacon positioning method | |
| CN104977581A (en) | Multi-moving target situation awareness method, device and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211229 Address after: 710077 room 11314, unit 1, building 1, Gaoxin Fengshang community, No. 7, Gaoxin 1st Road, Zhangba street, high tech Zone, Xi'an, Shaanxi Patentee after: Xi'an saipute Information Technology Co.,Ltd. Address before: 710072 No. 127 Youyi West Road, Shaanxi, Xi'an Patentee before: Northwestern Polytechnical University |
|
| TR01 | Transfer of patent right |