CN113189553A - Multi-modal optimization-based micro-maneuvering multi-target rapid detection method - Google Patents
Multi-modal optimization-based micro-maneuvering multi-target rapid detection method Download PDFInfo
- Publication number
- CN113189553A CN113189553A CN202110121747.2A CN202110121747A CN113189553A CN 113189553 A CN113189553 A CN 113189553A CN 202110121747 A CN202110121747 A CN 202110121747A CN 113189553 A CN113189553 A CN 113189553A
- Authority
- CN
- China
- Prior art keywords
- particle
- subgroup
- particles
- target
- individual optimal
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 52
- 238000005457 optimization Methods 0.000 title claims abstract description 32
- 239000002245 particle Substances 0.000 claims abstract description 145
- 238000000034 method Methods 0.000 claims abstract description 53
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 10
- 230000009191 jumping Effects 0.000 claims abstract description 7
- 238000009825 accumulation Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 230000006835 compression Effects 0.000 claims description 4
- 238000007906 compression Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 230000009466 transformation Effects 0.000 description 7
- 230000001427 coherent effect Effects 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 241000238631 Hexapoda Species 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000002592 echocardiography Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000009828 non-uniform distribution Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000005295 random walk Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
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
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/415—Identification of targets based on measurements of movement associated with the target
-
- 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
- G01S13/88—Radar or analogous systems specially adapted for specific applications
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a micro-motor multi-target rapid detection method based on multi-modal optimization, belongs to the technical field of radar target detection and intelligent optimization, and can improve the detection efficiency of a radar for micro-motor multi-targets. Firstly, radar echo signals are obtained, multi-mode particle swarm optimization algorithm parameters are set, and the positions and the speeds of all particles are initialized. The fitness value of each particle is calculated and its individual optimal solution is updated. And dividing the particles into different subgroups by adopting an improved multi-modal subgroup division method, and determining seed particles of the subgroups. And (4) carrying out blind speed side lobe jumping of the individual optimal solution on all the subgroup seeds, and updating the individual optimal solution. The velocity and position of each particle is updated and bound. The invalid subgroup is determined and reset. And executing iteration until the set maximum iteration number is reached. And taking the individual optimal solution of all the subgroup seeds as the detected target, and removing false targets from the detected target.
Description
Technical Field
The invention relates to the technical field of radar target detection and intelligent optimization, in particular to a micro-maneuvering multi-target rapid detection method based on multi-modal optimization.
Background
With the wide application of radar in the civil field, tiny maneuvering targets such as birds, insects and small unmanned aerial vehicles bring new challenges to radar target detection. The modern radar generally adopts a mode of long-time accumulation of echo pulses to gather target energy and improve the detection signal-to-noise ratio of a tiny target, and the main accumulation is divided into non-coherent accumulation and coherent accumulation.
During long-term accumulation, the phenomenon of cross-range unit (ARU) and cross-Doppler unit (ADU) which are easily generated can cause that the traditional Moving Target Detection (MTD) technology can not realize effective accumulation of target energy. The existing Hough transformation and dynamic programming method can realize non-coherent accumulation of target echoes under ARU, but certain accumulation gain can be lost due to the loss of phase information, and a good detection effect can not be realized for a tiny maneuvering target.
In long-time coherent accumulation, the ARU phenomenon is corrected through the common Keystone transformation on the scale change of a slow time axis, but the method is only suitable for detecting the uniform motion of the target, and a large amount of extra operations are introduced by the method aiming at the search and compensation of speed fuzzy numbers. Generalized Radon-Fourier transform (GRFT) converts the problem of long-time accumulation into a parameter model matching problem, can span unit accumulation and simultaneously realize motion phase compensation of any order, and realizes effective detection of weak and small maneuvering targets.
The existing method for shortening the time consumption of GRFT target detection by adopting a particle swarm optimization algorithm has obvious loss in detection performance compared with a GRFT method realized by traversal search, cannot detect a plurality of targets simultaneously, and particularly cannot meet the radar detection requirement of simultaneously appearing a plurality of targets such as insects, birds and the like.
Therefore, in a scene of a plurality of maneuvering tiny targets, how to use the radar to perform efficient detection and detection of the targets is a problem to be solved urgently at present.
Disclosure of Invention
In view of the above, the invention provides a micro-maneuvering multi-target rapid detection method based on multi-modal optimization, which adopts an improved multi-modal particle swarm optimization algorithm to rapidly realize multi-target radar detection based on a GRFT method, and is beneficial to improving the detection efficiency of a radar for micro-maneuvering multi-target.
In order to achieve the purpose, the technical scheme of the invention comprises the following steps:
s1: and acquiring M frames of radar echo signals, wherein each frame of radar echo signal corresponds to one frame of one-dimensional range profile, and each frame of one-dimensional range profile comprises more than one range unit.
S2: and setting parameters of the multi-mode particle swarm optimization algorithm according to the target characteristics in the radar detection scene, and randomly initializing the positions and the speeds of all particles.
S3: and calculating the fitness value of each particle, and updating the individual optimal solution of the particle to be the current particle position if the current fitness value of the particle is superior to the fitness value of the individual optimal solution of the particle.
S4: an improved multi-modal subgroup partitioning method is employed to partition particles into different subgroups while determining seed particles for each subgroup.
S5: and (4) carrying out blind speed side lobe jumping of the individual optimal solution on all the subgroup seeds, and updating the individual optimal solution.
S6: the velocity of each particle is updated and the position of the particle in the next iteration is updated based on the velocity.
S7: and judging to obtain an invalid subgroup according to the position relation among the individual optimal solutions of the seeds of different subgroups, and resetting all the particles in the invalid subgroup.
S8: repeating S3-S7 until the set maximum iteration number kmax。
S9: and collecting the individual optimal solutions of all the subgroup seeds as the detected targets, using the position information of the targets as the motion parameters of the detected targets, and removing false targets from the position information.
Further, according to the target characteristics in the radar detection scene, setting multi-mode particle swarm optimization algorithm parameters, and randomly initializing the positions and the speeds of all particles, specifically comprising
The target motion order D is also the search space dimension of the motion parameters.
Search space Range [ Xmin,Xmax]Wherein X ismaxAnd XminSearching upper and lower boundaries of the space for the parameters respectively; the search space corresponding to each dimension is [ X ]min0,Xmax0]～[XminD-1,XmaxD-1]Wherein X ismax0～XmaxD-1Respectively searching the upper bound of the space for the parameters of the 0 th dimension to the D-1 th dimension; xmin0～XminD-1The lower bound of the parameter search space is respectively the 0 th dimension to the D-1 th dimension.
Number of subgroups NsnSubgroup size NssMaximum number of iterations kmaxA speed constraint factor alpha and a reset factor beta.
Randomly initializing the positions and speeds of all particles in the whole parameter search space, namely the position x of the ith particlei＝(x0,x1,…,xD-1) With the velocity v of the ith particlei＝(v0,v2,…,vD-1) Where i is 1,2, …, N, and the total number of particles N is NsnNss；x0,x1,…,xD-1Respectively the position information of the 0 th to D-1 th dimension in the particle position; v. of0,v2,…,vD-1Respectively, the velocity information of the 0 th to D-1 th dimensions among the particle velocities.
Further, calculating a fitness value of each particle, and if the current fitness value of the particle is superior to the fitness value of the individual optimal solution, updating the individual optimal solution of the particle to be the current particle position, specifically:
taking generalized Radon-Fourier transform, namely GRFT transform as a fitness function f (x)i) Calculating the fitness value of each particle, namely taking the particle position information as an estimated target motion parameter and extracting a corresponding target motion track from M frames of radar echo signals, wherein the target motion track is formed by range units where targets in each frame of one-dimensional range profile are located, performing GRFT (generalized regression transform) on radar echo data corresponding to the range units, and taking an accumulation result obtained after the conversion as the fitness value of the particle, wherein the GRFT specifically comprises the following steps:
wherein,
for radar echo signals obtained after pulse compression, fcIs the radar carrier frequency, tm＝mTrTo indicate the slow time of the radar echo pulse instant, TrM is the serial number of radar echo pulse, M is 0,1,2, …, M-1, M is the total number of accumulated pulses, and time delay is
Comprises the following steps:
c is the speed of light, xi,dThe motion parameter indicating the d-th order of the target corresponding to the i-th particle corresponds to the position information of the d-th dimension in the particle position.
The individual optimal solution is a historical position with an optimal fitness value recorded in an iteration process, and an initial value of the historical position is empty; and if the current fitness value of the particle is superior to the fitness value of the individual optimal solution, updating the individual optimal solution to be the current particle position.
Further, the improved multi-modal subgeneration method is as follows: firstly, sorting all particles according to the individual optimal fitness value, selecting the sorted particles with the optimal fitness as new subgroup seeds from all the undivided particles, then calculating the distance from all the undivided subgroup particles to the subgroup seeds, and selecting the N closest to the subgroup seedssnThe individual particles form a subgroup and the above operation is repeated until all the particles are divided.
Further, performing blind speed side lobe jumping of the individual optimal solution on all the subgroup seeds, and updating the individual optimal solution, specifically:
wherein,
for the jth subgroup of seed particles sjThe individual optimal solution of (2).
q is the number of blind speed ambiguities,
vbthe speed is blind.
Subgroup seeds
The center position of a blind speed side lobe corresponding to the blind speed fuzzy number q is set as
Further, the velocity of each particle is updated, and the position of the particle in the next iteration is updated according to the velocity, specifically:
vi(k+1)＝ωvi(k)+c1r1(pi(k)-xi(k))+c2r2(pni(k)-xi(k))
xi(k+1)＝xi(k)+vi(k+1)
wherein v isi(k) And xi(k) Respectively, the velocity and position of the ith particle at the k-th iteration, k is the current iteration number, k is 1,2, …, kmax；kmaxIs the maximum iteration number; the number i is 1, i is equal to 1,2, …, N, N is the number of all particles; omega is an inertia factor; c. C1,c2Two preset learning factors are set; r is1,r2Are all intervals of [0,1 ]]The random number of the inner part of the random number,
as an individual optimal solution for particle i, niThe seed number of the subgroup where the particle i is located.
At the same time, a maximum speed limit V is setmaxIs composed of
Vmax＝α(Xmax-Xmin)
Wherein α ∈ (0,1) is the speed limiting factor, if the absolute value of the updated speed exceeds the maximum speed limit VmaxResetting the updated speed to VmaxAnd if the updated position exceeds the upper and lower boundaries of the parameter search space of the search space range, resetting the updated particle position as a random value in the search space range.
Further, according to the position relationship between the individual optimal solutions of different subgroup seeds, an invalid subgroup is judged and obtained, and all particles in the invalid subgroup are reset, and the specific method comprises the following steps:
if the target motion trajectories extracted by the seeds of the two subgroups according to the individual optimal solutions of the seeds of the two subgroups overlap by more than beta M distance units, the subgroup with the poor fitness in the seeds of the two subgroups is judged to be an invalid subgroup, the positions and the speeds of all the particles in the invalid subgroup are initialized randomly before the next iteration is started, and the individual optimal solutions of all the particles in the invalid subgroup are empty;
wherein β is a predetermined reset factor.
Further, the false target is removed, specifically:
aiming at all the targets collected in the iteration result, if the fitness value calculated by utilizing GRFT is smaller than the threshold gammadAnd (4) if the target is a false target, removing the false target.
Threshold gammadThe calculation method comprises the following steps:
wherein, PfaFor a set false alarm rate, σ2Is the average noise power after accumulation.
Has the advantages that:
the invention provides a micro maneuvering multi-target rapid detection method based on multi-modal optimization, which is mainly used for detecting micro maneuvering targets such as insects and the like, and has a good detection effect on the micro maneuvering targets with higher movement component orders. The scheme of the invention is as follows: the method comprises the steps of converting a micro maneuvering multi-target detection problem based on a GRFT method into a multi-modal optimization problem, after a radar acquires a plurality of frames of one-dimensional range profiles, reasonably setting algorithm parameters according to radar echo characteristics of a target to be detected, evaluating particles by taking GRFT transformation as a fitness function in an iteration process, dividing subgroups and jumping out blind speed side lobes of the seeds by adopting an improved method, constraining the speed and the position of the particles, resetting invalid subgroups, and collecting the individual optimal solution of all the subgroups of the seeds as the motion parameters of a plurality of detected targets, so that a method for quickly realizing the micro maneuvering multi-target radar detection is provided. Compared with the existing GRFT multi-target detection method realized by traversing search parameters, the method provided by the invention uses a multi-modal optimization algorithm to carry out random heuristic search on target parameters, and realizes more rapid GRFT multi-target radar detection and more accurate target motion parameter estimation. Compared with the existing multi-modal optimization method based on fixed radius subgroup division and the like, the method has the advantages that the subgroup division method is improved, the invalid subgroup is reset, and better radar detection performance is realized for micro-maneuvering multi-targets. In conclusion, the invention achieves higher detection efficiency and relatively better detection performance in the aspect of radar detection of micro maneuvering multiple targets.
Drawings
FIG. 1 is a flow chart of a micro-maneuvering multi-target rapid detection method based on multi-modal optimization according to the present invention;
fig. 2 is a schematic diagram of target detection probability and peak rate after removing false targets for 40 targets in the embodiment provided by the present invention, where (a) in fig. 2 is a schematic diagram of target detection probability and (b) in fig. 2 is a schematic diagram of peak rate.
Detailed Description
The invention is described in detail below by way of example with reference to the accompanying drawings.
As shown in FIG. 1, the implementation of the invention provides a micro-maneuvering multi-target rapid detection method based on multi-modal optimization, which comprises the following specific steps:
s1: according to an actually used radar system, pulse compression is carried out on each received echo pulse to obtain a frame of one-dimensional range profile comprising a plurality of range units, and radar echoes are repeatedly received until M frames of one-dimensional range profiles are obtained to form a radar echo range-time plane. The number of range cells is determined by the radar range resolution and range detection range used.
The number of the one-dimensional range image frames M to be accumulated in the embodiment of the invention can be determined according to the requirement on the target detection result, and the detection result is analyzed in a trial value mode to obtain a better M value.
S2: according to the target characteristics in a radar detection scene, setting multi-modal particle swarm optimization algorithm parameters, specifically comprising the following steps:
the target motion order D, i.e. the search space dimension of the motion parameter;
search space Range [ Xmin,Xmax]Wherein X ismaxAnd XminRespectively searching upper and lower boundaries of a space for the parameters; the search space corresponding to each dimension is [ X ]min0,Xmax0]～[XminD-1,XmaxD-1]Wherein X ismax0～XmaxD-1Respectively searching the upper bound of the space for the parameters of the 0 th dimension to the D-1 th dimension; xmin0～XminD-1Respectively searching the lower bound of the space for the parameters of the 0 th dimension to the D-1 th dimension;
number of subgroups NsnSubgroup size NssMaximum number of iterations kmaxA speed constraint factor alpha and a reset factor beta.
At the same time, in the wholeRandomly initializing the positions and speeds of all particles in the parameter search space, and the position x of the ith particlei＝(x0,x1,…,xD-1) With the velocity v of the ith particlei＝(v0,v2,…,vD-1) Wherein
Total number of particles N ═ NsnNss。x0,x1,…,xD-1Respectively the position information of the 0 th to D-1 th dimension in the particle position; v. of0,v2,…,vD-1Respectively, the velocity information of the 0 th to D-1 th dimensions among the particle velocities.
S3: taking generalized Radon-Fourier transform, namely GRFT transform as a fitness function f (x)i) Calculating the fitness value of each particle, namely taking the particle position information x as an estimated target motion parameter and extracting a corresponding target motion track from M frames of radar echo signals, wherein the track is formed by range units where targets in each frame of one-dimensional range image are located, performing GRFT (generalized regression transform) on radar echo data corresponding to the range units, and taking an accumulation result obtained after transformation as the fitness value of the particle, wherein the GRFT specifically comprises the following steps:
wherein,
for radar echo signals obtained after pulse compression, fcIs the radar carrier frequency, tm＝mTr(M-0, 1,2, …, M-1) is slow time, TrFor pulse repetition interval, time delay
Comprises the following steps:
c is the speed of light, xi,dRepresenting the motion parameter of the d-th order of the target corresponding to the ith particle, usually xi,0Denotes the starting distance, xi,1Representing the speed, xi,2Indicating acceleration, and so on, corresponding to xi＝(x0,x1,…,xD-1)。
When the motion parameters of the particle search are closer to the real target, the accumulated peak value obtained after GRFT transformation is higher, so that the position of the particle in the parameter search space can be effectively evaluated whether to be close to the real target by selecting the GRFT transformation as the fitness function.
The individual optimal solution is a historical position with an optimal fitness value recorded in an iteration process, and an initial value of the historical position is empty; and if the current fitness value of the particle is superior to the fitness value of the individual optimal solution, updating the individual optimal solution to be the current particle position.
S4: an improved multi-modal subgroup division method is adopted to divide particles into different subgroups and determine seed particles of each subgroup at the same time, and the specific subgroup division method comprises the following steps: firstly sorting all particles according to the fitness value of the individual optimal solution, selecting the particles with the optimal fitness (the maximum or the minimum fitness value) after sorting from all the particles which are not divided, namely the particles with the best accumulation result after GRFT (generalized regression transformation), as new subgroup seeds, calculating the distance from the particles of other non-divided subgroups to the seeds, and selecting the N closest to the seedsssThe individual particles form a subgroup and the above operation is repeated until all the particles are divided. In the embodiment, the Euclidean distance is adopted to measure the distance between particles, and other distance judgment methods such as Manhattan distance, normalized Euclidean distance and the like can also be adopted.
S5: performing blind speed side lobe jumping of the individual optimal solution on all the subgroup seeds, and updating the individual optimal solution, wherein the specific blind speed side lobe jumping method comprises the following steps:
wherein,
for the jth subgroup of seed particles sjThe individual optimal solution of (a); q is the blind speed fuzzy number
seed psjThe center position of the corresponding blind speed side lobe with fuzzy number q is
Because the accumulated peak value of the position of the main lobe where the real target is located is higher, target missing detection caused by the fact that the seed falls into a blind speed side lobe local optimal solution can be avoided by searching all parameters forming a blind speed side lobe relation with the seed.
S6: updating the speed of each particle, and updating the position of the particle in the next iteration according to the speed, wherein the specific position and speed updating method comprises the following steps:
vi(k+1)＝ωvi(k)+c1r1(pi(k)-xi(k))+c2r2(pni(k)-xi(k))
xi(k+1)＝xi(k)+vi(k+1)
wherein v isi(k) And xi(k) Respectively the speed and the position of the ith particle at the kth iteration, k being the current iterationNumber, k ═ 1,2, …, kmax；kmaxIs the maximum iteration number; omega is an inertia factor and is generally used for balancing the global exploration capability and the local development capability of particles in the random search process; c. C1,c2The weight of the optimal solution of the individual and subgroup seeds is adjusted respectively for learning factors; r is1,r2Is the interval [0,1]Random numbers in the search space, for ensuring random walk of the particles throughout the search space,
for the individual optimal solution position, n, of the particle iiThe seed number of the subgroup where the particle i is located.
At the same time, a maximum speed limit V is setmaxComprises the following steps:
Vmax＝α(Xmax-Xmin)
wherein α ∈ (0,1) is the speed limiting factor, if the absolute value of the updated speed exceeds the maximum speed limit VmaxResetting the updated speed to VmaxAnd if the updated position exceeds the upper and lower boundaries of the parameter search space of the search space range, resetting the updated particle position as a random value in the search space range.
The constraint aiming at the speed and the position of the particle can effectively avoid the reduction of the searching efficiency caused by the transitional migration of the particle in the searching process.
S7: according to the position relation among the individual optimal solutions of different subgroup seeds, judging to obtain an invalid subgroup and resetting all particles in the invalid subgroup, wherein the specific judging and resetting method comprises the following steps: if more than beta M distance units are overlapped in the target motion trail extracted by the seeds of the two subgroups according to the individual optimal solutions, wherein beta is a reset factor, the subgroup with the poor fitness of the seeds of the two subgroups is judged to be an invalid subgroup, the positions and the speeds of all particles in the invalid subgroup are initialized randomly before the next iteration is started, and the individual optimal solutions of the particles are also emptied.
S8: repeating S3-S7 until the set maximum iteration number kmax；
S9: collection stackThe individual optimal solution of all the subgroup seeds in the generation result is used as a detected target, the position information of the detected target is used as a motion parameter of the detected target, and the GRFT accumulation result (fitness value) is removed from the detected target, wherein the fitness value is smaller than a threshold gammadTarget, threshold γ ofdThe calculation method comprises the following steps:
wherein, PfaFor a set false alarm rate, σ2Is the average noise power after accumulation.
In order to verify the effectiveness of the multi-modal optimization method, the invention adopts the micro-maneuvering multi-target rapid detection method based on the multi-modal optimization based on simulation experiment data to complete the simultaneous detection of a plurality of micro-maneuvering targets. The radar system simulation parameters are shown in table 1:
TABLE 1 Radar System simulation parameters
| Radar carrier frequency | Bandwidth of radar | Pulse width | Sampling frequency | Pulse repetition frequency | Accumulated pulse number |
| 16.2GHz | 800MHz | 1us | 1.28GHz | 500Hz | 300 |
For the convenience of verification, the target motion order D is 3, and is between-250 m, -5m/s and-3 m/s2]And [250m, 5m/s,3m/s2]The 20 simulation targets are randomly generated in the motion parameter range, the targets have the maneuvering phenomenon, and the starting distances of the targets are distributed on the distance axis in a non-uniform distribution mode.
Setting the number and size of subgroups to Nsn＝40，Nss25, the upper and lower bounds of the motion parameter search space are Xmin＝[-400m,-10m/s,-5m/s2]，Xmax＝[400m,10m/s,5m/s2]Randomly generating the positions and the speeds of 1000 particles in the space range, and then carrying out iterative evolution on all the particles by taking GRFT as a fitness function until the maximum iteration number k is reachedmaxThe iteration is terminated 6000. In the updating of the position and the speed, the inertia factor omega is made to linearly decrease along with the iteration number from 0.9 to 0.4, and the learning factor c1＝2.8,c2The speed limiting factor α is equal to 0.25, and the reset factor β is equal to 0.1 in resetting the non-active subgroup. And finally obtaining the target detection probability and the peak value rate of 40 targets after the false targets are removed.
It can be seen that when the signal-to-noise ratio is greater than 0dB, the detection performance of the BMMO in the method is far better than that of a BMMO-radius multi-modal optimization method based on a traditional fixed-radius niche, and a performance curve is similar to that of a GRFT traversal method. In the aspect of estimation precision of motion parameters, the root mean square error of the GRFT traversal method is [0.0275m,0.0038m/s,0.0125m/s ]2]The maximum error of the method of the invention is [0.0188m, 0.0029m/s,0.0094m/s2]And is superior to the traversal implementation method. Under the test in the same hardware and software environment, the time consumption of one detection process of the method is 505s, the time consumption of a multi-modal optimization method based on the fixed-radius niche is 754s, and the time consumption of a GRFT traversal search method is 60703 s. Hair brushThe method only loses certain performance under the condition of extremely low signal-to-noise ratio, meanwhile, the detection efficiency is greatly improved, and the effectiveness of the method is verified.
In summary, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. A micro maneuvering multi-target rapid detection method based on multi-modal optimization is characterized by comprising the following steps:
s1: acquiring M frames of radar echo signals, wherein each frame of radar echo signal corresponds to one frame of one-dimensional range profile, and each frame of one-dimensional range profile comprises more than one range unit;
s2: according to the target characteristics in a radar detection scene, setting multi-mode particle swarm optimization algorithm parameters, and randomly initializing the positions and the speeds of all particles;
s3: calculating the fitness value of each particle, and if the current fitness value of the particle is superior to the fitness value of the individual optimal solution, updating the individual optimal solution of the particle to be the current particle position;
s4: dividing the particles into different subgroups by adopting an improved multi-modal subgroup division method, and simultaneously determining seed particles of each subgroup;
s5: performing blind speed side lobe jumping of the individual optimal solution on all the subgroup seeds, and updating the individual optimal solution;
s6: updating the speed of each particle, and updating the position of the particle in the next iteration according to the speed;
s7: judging to obtain an invalid subgroup and resetting all particles in the invalid subgroup according to the position relation among the individual optimal solutions of the seeds of different subgroups;
s8: repeating S3-S7 until the set maximum iteration number kmax；
S9: and collecting the individual optimal solutions of all the subgroup seeds as the detected targets, using the position information of the targets as the motion parameters of the detected targets, and removing false targets from the position information.
2. The method as claimed in claim 1, wherein the method for rapidly detecting the micro-maneuvering multi-target based on the multi-modal optimization sets parameters of the multi-modal particle swarm optimization algorithm according to target characteristics in a radar detection scene, and randomly initializes the positions and the speeds of all particles, specifically comprising
The target movement order D is also the search space dimension of the movement parameters;
search space Range [ Xmin,Xmax]Wherein X ismaxAnd XminRespectively searching upper and lower boundaries of a space for the parameters; the parameter search space for each dimension is [ X ]min0,Xmax0]～[XminD-1,XmaxD-1]Wherein X ismax0～XmaxD-1Respectively searching the upper bound of the space for the parameters of the 0 th dimension to the D-1 th dimension; xmin0～XminD-1Respectively searching the lower bound of the space for the parameters of the 0 th dimension to the D-1 th dimension;
number of subgroups NsnSubgroup size NssMaximum number of iterations kmaxA speed constraint factor α and a reset factor β;
randomly initializing the position x of the ith particle of the positions of all the particles in the whole search parameter spacei＝(x0,x1,…,xD-1) With the velocity v of the ith particlei＝(v0,v2,…,vD-1) Where i is 1,2, …, N, and the total number of particles N is NsnNss；x0,x1,…,xD-1Respectively the position information of the 0 th to D-1 th dimension in the particle position; v. of0,v2,…,vD-1Respectively, the velocity information of the 0 th to D-1 th dimensions among the particle velocities.
3. The method as claimed in claim 1 or 2, wherein the fitness value of each particle is calculated, and if the current fitness value of a particle is better than the fitness value of the individual optimal solution, the individual optimal solution is updated to the current particle position, specifically:
taking generalized Radon-Fourier transform, namely GRFT transform as a fitness function f (x)i) Calculating the fitness value of each particle, namely taking the particle position information as an estimated target motion parameter and extracting a corresponding target motion track from M frames of radar echo signals, wherein the target motion track is formed by range units where targets in each frame of one-dimensional range profile are located, performing GRFT (generalized regression transform) on radar echo data corresponding to the range units, and taking an accumulation result obtained after the conversion as the fitness value of the particle, wherein the GRFT specifically comprises the following steps:
wherein,
for radar echo signals obtained after pulse compression, fcIs the radar carrier frequency, tm＝mTrSlow time, T, of radar echo pulse timerM is the serial number of radar echo pulse, M is 0,1,2, …, M-1, M is the total number of accumulated pulses, and time delay is
Comprises the following steps:
c is the speed of light, xi,dA motion parameter indicating a d-th order of the target corresponding to the i-th particle corresponds to position information of a d-th dimension in the position of the particle;
the individual optimal solution is a historical position with the optimal fitness value recorded in the iteration process, and the initial value of the historical position is empty; and if the current fitness value of the particle is superior to the fitness value of the individual optimal solution, updating the individual optimal solution to be the current particle position.
4. As claimed in claim 1 or 2The micro maneuvering multi-target rapid detection method based on multi-modal optimization is characterized in that the improved multi-modal subgroup division method comprises the following steps: firstly, sorting all particles according to the individual optimal fitness value, selecting the sorted particles with the optimal fitness as new subgroup seeds from all the undivided particles, then calculating the distance from all the undivided subgroup particles to the subgroup seeds, and selecting the N closest to the subgroup seedssnThe individual particles form a subgroup and the above operation is repeated until all the particles are divided.
5. The method for rapidly detecting the small maneuvering multi-target based on the multi-modal optimization as claimed in claim 2, characterized in that the blind speed side lobe leaping of the individual optimal solution is performed on all the subgroup seeds, and the individual optimal solution is updated, specifically:
wherein,
for the jth subgroup of seed particles sjThe individual optimal solution of (a);
q is the number of blind speed ambiguities,
vbthe speed is the blind speed;
subgroup seeds
The center position of a blind speed side lobe corresponding to the blind speed fuzzy number q is set as
D is the dimension of the search space and corresponds to the movement order of the target;
6. The method for micro-maneuvering multi-target fast detection based on multi-modal optimization as recited in claim 1, characterized by updating the velocity of each particle and updating the position of the particle in the next iteration according to the velocity, specifically:
xi(k+1)＝xi(k)+vi(k+1)
wherein v isi(k) And xi(k) Respectively, the velocity and position of the ith particle at the k-th iteration, k is the current iteration number, k is 1,2, …, kmax；kmaxIs the maximum iteration number; i is 1,2, …, N, N is the number of all particles; omega is an inertia factor; c. C1,c2Two preset learning factors are set; r is1,r2Are all intervals of [0,1 ]]The random number of the inner part of the random number,
as an individual optimal solution for particle i, niThe seed serial number of the subgroup where the particle i is located;
at the same time, a maximum speed limit V is setmaxIs composed of
Vmax＝α(Xmax-Xmin)
Wherein α ∈ (0,1) is the speed limiting factor, if the absolute value of the updated speed exceeds the maximum speed limit VmaxResetting the updated speed to VmaxIf the updated position exceeds the parameter search of the search space rangeAnd searching the upper and lower boundaries of the space, and resetting the updated particle position as a random value in the search space range.
7. The method for rapidly detecting the small maneuvering multiple targets based on the multi-modal optimization as claimed in claim 1, wherein the method for judging and obtaining the invalid subgroup and resetting all the particles in the invalid subgroup according to the position relationship among the individual optimal solutions of different subgroup seeds comprises the following specific steps:
if the target motion trajectories extracted by the two subgroups of seeds according to the individual optimal solutions of the seeds are overlapped by more than beta M distance units, the subgroup with poor fitness in every two subgroups of seeds is judged to be an invalid subgroup, the positions and the speeds of all particles in the invalid subgroup are initialized randomly before the next iteration is started, and the individual optimal solutions of all particles in the invalid subgroup are empty;
wherein β is a predetermined reset factor.
8. The method for rapidly detecting the small maneuvering multiple targets based on the multi-modal optimization as claimed in claim 1, wherein the removing the false targets specifically comprises:
aiming at all the targets collected in the iteration result, if the fitness value calculated by utilizing GRFT is smaller than the threshold gammadIf the target is a false target, removing the false target;
threshold gammadThe calculation method comprises the following steps:
wherein, PfaFor a set false alarm rate, σ2Is the average noise power after accumulation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110121747.2A CN113189553A (en) | 2021-01-28 | 2021-01-28 | Multi-modal optimization-based micro-maneuvering multi-target rapid detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110121747.2A CN113189553A (en) | 2021-01-28 | 2021-01-28 | Multi-modal optimization-based micro-maneuvering multi-target rapid detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113189553A true CN113189553A (en) | 2021-07-30 |
Family
ID=76972744
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110121747.2A Pending CN113189553A (en) | 2021-01-28 | 2021-01-28 | Multi-modal optimization-based micro-maneuvering multi-target rapid detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113189553A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115359337A (en) * | 2022-08-23 | 2022-11-18 | 四川大学 | Searching method, system and application of pulse neural network for image recognition |
| CN116000069A (en) * | 2023-02-06 | 2023-04-25 | 一夫科技股份有限公司 | Method and system for processing waste resources |
| CN116400318A (en) * | 2023-06-08 | 2023-07-07 | 中国人民解放军国防科技大学 | Multi-observation target position estimation method and device based on online particle swarm optimization |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103336855A (en) * | 2013-05-24 | 2013-10-02 | 浙江工业大学 | Two-dimensional irregular layout method based on multi-subpopulation particle swarm optimization |
-
2021
- 2021-01-28 CN CN202110121747.2A patent/CN113189553A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103336855A (en) * | 2013-05-24 | 2013-10-02 | 浙江工业大学 | Two-dimensional irregular layout method based on multi-subpopulation particle swarm optimization |
Non-Patent Citations (11)
| Title |
|---|
| JACO F. SCHUTTE 等: "A Study of Global Optimization Using Particle Swarms", 《JOURNAL OF GLOBAL OPTIMIZATION》 * |
| L.-C. QIAN 等: "Fast implementat on of generalised Radon-Fourier transform for manoeuvring radar target detection", 《ELECTRONICS LETTERS》 * |
| LICHANG QIAN 等: "Efficient approach of generalized RFT based on PSO", 《IEEE COMPUTER SOCIETY》 * |
| WEIJIE XIA 等: "A Fast Algorithm of Generalized Radon-Fourier Transform for Weak Maneuvering Target Detection", 《INTERNATIONAL JOURNAL OF ANTENNAS AND PROPAGATION》 * |
| 孙娟: "用基于族群的方法求解动态优化问题" * |
| 梁静 等: "求解大规模问题协同进化动态粒子群优化算法", 《软件学报》 * |
| 潘希姣: "多子群粒子群集成神经网络", 《安徽建筑工业学院学报(自然科学版)》 * |
| 王玺: "基于自适应小生境文化基因算法的数据聚类" * |
| 谢红侠 等: "基于多种群的改进粒子群算法多模态优化", 《计算机应用》 * |
| 邵池: "基于适应值地形信息的差分演化算法研究" * |
| 陈小玉: "一种快速多种群的粒子群多模优化算法", 《计算机应用研究》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115359337A (en) * | 2022-08-23 | 2022-11-18 | 四川大学 | Searching method, system and application of pulse neural network for image recognition |
| CN115359337B (en) * | 2022-08-23 | 2023-04-18 | 四川大学 | Searching method, system and application of pulse neural network for image recognition |
| CN116000069A (en) * | 2023-02-06 | 2023-04-25 | 一夫科技股份有限公司 | Method and system for processing waste resources |
| CN116000069B (en) * | 2023-02-06 | 2023-11-17 | 一夫科技股份有限公司 | Method and system for processing waste resources |
| CN116400318A (en) * | 2023-06-08 | 2023-07-07 | 中国人民解放军国防科技大学 | Multi-observation target position estimation method and device based on online particle swarm optimization |
| CN116400318B (en) * | 2023-06-08 | 2023-08-04 | 中国人民解放军国防科技大学 | Multi-observation target position estimation method and device based on online particle swarm optimization |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113189553A (en) | Multi-modal optimization-based micro-maneuvering multi-target rapid detection method | |
| CN107861107B (en) | Double-threshold CFAR (computational fluid dynamics) and trace point agglomeration method suitable for continuous wave radar | |
| CN109100718B (en) | Sparse aperture ISAR self-focusing and transverse calibration method based on Bayesian learning | |
| US20200225321A1 (en) | Machine learning based highway radar vehicle classification across multiple lanes and speeds | |
| CN106226751B (en) | Maneu-vering target detection and tracking based on DP-TBD | |
| US8405540B2 (en) | Method for detecting small targets in radar images using needle based hypotheses verification | |
| CN106842165B (en) | Radar centralized asynchronous fusion method based on different distance angular resolutions | |
| CN102608590B (en) | Coherent integration method based on dynamic programming and back image projection algorithm | |
| CN103439697B (en) | Target detection method based on dynamic programming | |
| CN109633646B (en) | Bistatic ISAR imaging method based on weighted L1 norm constraint | |
| CN105093198B (en) | A kind of flight path fusion method of distributed external illuminators-based radar networking detection | |
| CN103760543A (en) | MM-CFAR target detection method | |
| CN111624567B (en) | Constant false alarm detection method and device | |
| CN114114192B (en) | Cluster target detection method | |
| CN110187335A (en) | Tracking before being detected for the particle filter with discontinuous characteristic target | |
| CN109324315A (en) | Space-time adaptive based on double level block sparsity handles radar clutter suppression method | |
| Cordone et al. | Improved multi-resolution method for mle-based localization of radiation sources | |
| CN112051552B (en) | Main lobe anti-interference method and device based on multiple stations | |
| CN111368653B (en) | Low-altitude small target detection method based on R-D graph and deep neural network | |
| CN107229040B (en) | high-frequency radar target detection method based on sparse recovery space-time spectrum estimation | |
| CN111812670B (en) | Single photon laser radar space transformation noise judgment and filtering method and device | |
| CN113608193A (en) | Radar multi-target distance and speed estimation method based on UNet | |
| CN105652256B (en) | A kind of high-frequency ground wave radar TBD methods based on polarization information | |
| CN112327286A (en) | Low-complexity daily activity classification method, device, equipment and storage medium | |
| CN108508413B (en) | Target detection method based on probability statistics under low signal-to-noise ratio condition |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210730 |