CN110336637A - A kind of unmanned plane interference signal feature selection approach - Google Patents
A kind of unmanned plane interference signal feature selection approach Download PDFInfo
- Publication number
- CN110336637A CN110336637A CN201910635110.8A CN201910635110A CN110336637A CN 110336637 A CN110336637 A CN 110336637A CN 201910635110 A CN201910635110 A CN 201910635110A CN 110336637 A CN110336637 A CN 110336637A
- Authority
- CN
- China
- Prior art keywords
- particle
- interference signal
- unmanned plane
- plane interference
- selection approach
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04K—SECRET COMMUNICATION; JAMMING OF COMMUNICATION
- H04K3/00—Jamming of communication; Counter-measures
- H04K3/60—Jamming involving special techniques
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04K—SECRET COMMUNICATION; JAMMING OF COMMUNICATION
- H04K3/00—Jamming of communication; Counter-measures
- H04K3/80—Jamming or countermeasure characterized by its function
- H04K3/82—Jamming or countermeasure characterized by its function related to preventing surveillance, interception or detection
- H04K3/825—Jamming or countermeasure characterized by its function related to preventing surveillance, interception or detection by jamming
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a kind of unmanned plane interference signal feature selection approach, belong to unmanned plane interference signal cognitive techniques field.The method includes constructing the initial characteristics set of discrete binary version of particle swarm optimization algorithm;Construct fitness function;Using discrete binary version of particle swarm optimization algorithm, particle position and speed are updated;If population optimal location is globally optimal solution, the corresponding character subset of globally optimal solution is exported;Otherwise, inertia weight or setting maximum number of iterations are modified, is updated again.The present invention can be balanced in terms of computing resource and classification accuracy, and excellent effect, reached the statistical correlation for removing feature under same performance index, saved the purpose of calculation resources;The present invention considers influence of the Characteristic Number to algorithm performance, in the case where same category recognition accuracy, selects Characteristic Number fewer, algorithm performance is better, and the number of iterations is less.
Description
Technical field
The invention belongs to unmanned plane interference signal cognitive techniques fields, in particular to one kind to be based on discrete binary population
The unmanned plane interference signal feature selection approach of optimization.
Background technique
Most of civilian unmanned planes carry out the premise that cognition is electronic countermeasure to the signal of communication that its data center receives
And basis, even more realize the intelligentized key of unmanned systems data-link.In addition, unmanned plane application scenarios are constantly expanded in recent years,
Demand to the cognition of its interference signal is more urgent.For the use for meeting the unmanned planes such as scientific research, the disaster relief, mitigation, consider each
To the data-link cognitive question and time resource of unmanned plane interference signal and the consumption of computing resource under scene, in the design people
Computing resource should can be saved when the scheme recognized with unmanned plane signal intelligent, is met in the case where consumption resource is small as far as possible
Feature selecting and itself cognition need, therefore need an efficient unmanned plane interference signal feature selection approach comprehensively.
In civilian UAV Communication system, in order to which signal can be transmitted in various complex environments, signal is not using
Same modulation system.The modulation for carrying out amplitude, phase and angle etc. for signal brings different signal spies to signal
Property.Signal characteristic is many kinds of, and the type of signal can not be all distinguished from single features, to need according to different applied fields
Scape constructs multiple features set to distinguish, and multi dimensional analysis is carried out to signal, to realize final perception target.But if
Choose at random feature construction combination, then can be between existing characteristics the problem of statistical correlation, this problem will lead to classification accuracy
It reduces and calculation resources consumption increases.
Summary of the invention
The present invention interferes to solve the problems, such as unmanned plane interference signal classification accuracy and computing resource in conjunction with unmanned plane
The characteristics of signal, devises a kind of unmanned plane interference signal feature selection approach based on discrete binary particle group optimizing, root
It is realized from the character subset for meeting current application scene is wherein filtered out to unmanned plane according to the characteristic set currently having been built up
The selection of interference signal feature.
The present invention provides a kind of unmanned plane interference signal feature selection approach, the specific steps are as follows:
The first step constructs the initial characteristics set of discrete binary version of particle swarm optimization algorithm;
According to interference signal signature analysis as a result, determining characteristic set, including wavelet character, spectrum correlated characteristic and envelope are special
Sign etc., emphasis are that building can be for characteristic set that discrete binary version of particle swarm optimization algorithm uses.
Second step constructs fitness function according to initial characteristics set;
According to the requirement of discrete binary version of particle swarm optimization algorithm, features all in characteristic set are subjected to feature coding,
Initialization feature vector constructs the fitness letter of discrete binary version of particle swarm optimization algorithm according to the classification accuracy of interference signal
Number.
Third step updates particle position and speed using discrete binary version of particle swarm optimization algorithm;
According to fitness function, discrete binary version of particle swarm optimization algorithm is iterated, obtain particle optimal location and
Population optimal location.
4th step exports the corresponding character subset of globally optimal solution if population optimal location is globally optimal solution;
Otherwise, inertia weight or setting maximum number of iterations are modified, third step is returned.
Main advantages of the present invention are:
(1) interference signal various dimensions characteristic set is constructed according to the result of signature analysis, signal description comprehensive comprising information
Ability is strong;
(2) it can be balanced, and excellent effect, reach in same performance in terms of computing resource and classification accuracy
The statistical correlation that feature is removed under index, saves the purpose of calculation resources;
(3) in terms of fitness function, the fitness function newly proposed is on the basis of Classification and Identification accuracy rate, it is contemplated that
Influence of the number of parameters to algorithm performance.In the case where same category recognition accuracy, select Characteristic Number fewer, algorithm
Can be better, and the number of iterations is less.
Detailed description of the invention
Fig. 1 is a kind of overall step flow chart of unmanned plane interference signal feature selection approach of the invention;
Fig. 2 is to update iterative process figure using discrete binary particle swarm algorithm in the present invention;
Fig. 3 is the optimal feature subset accuracy rate curve that the present invention obtains.
Specific embodiment
With reference to the accompanying drawing, specific implementation method of the invention is described in detail.
The present invention is a kind of unmanned plane interference signal feature selection approach, and overall flow figure is as shown in Figure 1, and the prior art
Difference be embodied in fitness function foundation and discrete binary particle swarm algorithm iteration update, embody of the invention
Superiority of the selection method in terms of feature selecting.The unmanned plane interference signal feature selection approach specifically includes following step
It is rapid:
The first step determines signal characteristic, constructs the initial characteristics set of discrete binary particle swarm algorithm.
In signal characteristic abstraction technology, the case where all modulation systems can be separated there is no any one feature, because
This must be combined single feature according to application scenarios, improve the separating capacity of feature.But the feature after combination is often
Face the excessive problem of feature redundancy, consumption resource.Present invention selection uses discrete binary particle swarm algorithm, and discrete two into
Granulation swarm optimization operation needs to construct an initial characteristics set, therefore the present invention is constructed first by wavelet character, instantaneous spy
The initial characteristics set of the signal characteristics compositions such as sign, envelope characteristic and spectrum correlated characteristic, is calculated for subsequent discrete binary population
Method uses, as shown in table 1.
12 initial characteristics in 1 initial characteristics set of table
In table 1, CWT indicates multiple dimensioned continuous wavelet transform, at the beginning of including 12 altogether in the characteristic set that the present invention provides
Beginning feature.
Second step carries out feature coding and initialization to the feature in initial characteristics set, establishes and calculate fitness letter
Number.
In discrete binary particle swarm algorithm, each particle position is a candidate solution, all particle positions
Candidate solution forms common solution space, and target search space is D dimension.Optimization aim is the feature vector of particle position.Each
Particle position is all a character subset, constitutes the solution space of optimization aim, represents the feature finally chosen.
Each particle position is encoded using binary form, and particle position is a multi-C vector, initialization
Each of characteristic set feature can be indicated with the coding of particle position, and particle position is all corresponded to per one-dimensional
One feature indicates this feature in the character subset of selection if the i-th bit (dimension) of particle position is 1, if it is 0, table
Show this feature not in the character subset of selection.
Such as, if aiIndicate some feature of interference signal, then { a1,a2,...,anConstitute the feature set of interference signal
It closes, if particle position is [101...1], the character subset selected is { a1,a3,...,an}。
By the feature vector, X of particle positioni=(xi1,xi2,...,xiD) be updated in fitness function, calculate fitness
Value, the quality of the feature vector of particle position, the i.e. quality of this candidate solution are evaluated with fitness.Fitness function is as follows:
Wherein, F (i) is the fitness of i-th of particle position, and p (i) is the classification accuracy of i-th of particle position, n (i)
Indicate the feature vector, X of i-th of particle positioniThe Characteristic Number of middle selection, t are the duration of classifier operation, and λ is characterized number
Parameter, generally take 0.01.
In addition to this, there are one flying speeds for each particle, are denoted as Vi=(vi1,vi2,...,viD), flying speed is determined
Determine Particles Moving direction and distance, particle position is updated using this flying speed, also determines particle iteration speed
Speed.If flying speed is very big, discrete binary particle swarm algorithm can restrain quickly, but easily fall into local optimum, such as
Fruit flying speed very little, then discrete binary particle swarm algorithm convergence is too slow, consumes excessive calculation resources.
Third step is updated particle position and speed using discrete binary particle swarm algorithm.
It updates iterative process as shown in Fig. 2, initialization feature set first is as discrete binary particle swarm algorithm
Population, the formula for updating particle position and speed are as follows:
Wherein, initializing total number of particles in population is m, i=1,2 ..., m;D=1,2 ..., D indicate particle position
Dimension;For the speed of i-th of particle d dimension in kth time iteration;r1And r2It is the random number between 0 to 1, c1And c2For
The Studying factors of discrete binary particle swarm algorithm, can according to need sets itself, and value range is between 0 to 4;ω is
Inertia weight, the value between 0.1 and 0.9;For the position vector of i-th of particle d dimension in kth time iteration;It is
The optimal location of i particle d dimension in kth time iteration;For the optimal position of population d dimension in kth time iteration
It sets;Function representation particle rapidity mapped into [0,1] section indicate particle binary digit variable take 1 it is general
Rate;ρ is the random number generated from section [0,1], is avoidedToo close 1 or 0.
To prevent particle beyond existing search space, updated speed is generally defined in speed minimum to particle each time
Value vminWith speed maximum value vmaxBetween, and judge that whether fly out existing search of particle is empty according to space boundary every time
Between, if flying out existing search space, boundary value is set by particle position.
The fitness value of particle position, k=1,2 ..., k are calculated after each iterationmax, kmaxFor maximum number of iterations.By
The optimal location searched until i particle to convergence is denoted as Pi=(pi1,pi2,...,piD), entire population is until restraining
The optimal location searched is denoted as Pg=(pg1,pg2,...,pgD)。
4th step, if the optimal location P found in search spacegFor globally optimal solution, then entire search process knot
Beam;Otherwise, it if as a result falling into locally optimal solution, needs to be appropriately modified discrete binary version of particle swarm optimization algorithm,
It is set to jump out locally optimal solution, discrete binary particle swarm algorithm stopping criterion for iteration is typically chosen as maximum number of iterations at this time
kmax.Ability of searching optimum is adjusted using adaptive weighting method, to meet the needs that population dynamic updates, inertia weight ω's
It is as follows to change expression formula:
Wherein, ωminWith ωmaxThe minimum value and maximum value of inertia weight are respectively represented, f indicates this particle in present bit
The fitness value set, fmaxWith fargRespectively indicate population in this time iteration maximum adaptation angle value and average fitness value.
The performance for the optimal feature subset that the method filters out is verified, the character subset of different characteristic number is used
The curve that the accuracy rate Pcc to classify changes with Signal to Noise Ratio (SNR) is fig. 3, it is shown that optimal feature subset divides
Class performance is slightly better than characteristic set, and there was only 4 features in optimal feature subset, remote in the upper consumed resource of training classification
Much smaller than initial characteristics set.
Claims (5)
1. a kind of unmanned plane interference signal feature selection approach, it is characterised in that: specific step is as follows for the method,
The first step constructs the initial characteristics set of discrete binary version of particle swarm optimization algorithm;
Second step constructs fitness function according to initial characteristics set;
According to the requirement of discrete binary version of particle swarm optimization algorithm, features all in characteristic set are subjected to feature coding, initially
Change feature vector, the fitness function of discrete binary version of particle swarm optimization algorithm is constructed according to the classification accuracy of interference signal;
Third step updates particle position and speed using discrete binary version of particle swarm optimization algorithm;
According to fitness function, discrete binary version of particle swarm optimization algorithm is iterated, obtains particle optimal location and particle
Group's optimal location.
4th step exports the corresponding character subset of globally optimal solution if population optimal location is globally optimal solution;It is no
Then, inertia weight or setting maximum number of iterations are modified, third step is returned.
2. a kind of unmanned plane interference signal feature selection approach according to claim 1, it is characterised in that: the initial spy
Collection includes wavelet character, spectrum correlated characteristic and envelope characteristic in closing.
3. a kind of unmanned plane interference signal feature selection approach according to claim 1, it is characterised in that: institute in second step
The fitness function stated is as follows:
Wherein, F (i) is the fitness of i-th of particle position, and p (i) is the classification accuracy of i-th of particle position, and n (i) is indicated
The feature vector, X of i-th of particle positioniThe Characteristic Number of middle selection, t are the duration of classifier operation, and λ is characterized the ginseng of number
Number.
4. a kind of unmanned plane interference signal feature selection approach according to claim 1, it is characterised in that: grain in third step
The formula that sub- position and speed updates is as follows:
Wherein, initializing total number of particles in population is m, i=1,2 ..., m;D=1,2 ..., D indicate the dimension of particle position
Degree;For the speed of i-th of particle d dimension in kth time iteration;r1And r2It is the random number between 0 to 1, c1And c2It is discrete
The Studying factors of binary particle swarm algorithm;ω is inertia weight;For the position of i-th of particle d dimension in kth time iteration
Set vector;For the optimal location of i-th of particle d dimension in kth time iteration;It is the population in kth time iteration
The optimal location of d dimension;ρ is the random number generated from section [0,1].
5. a kind of unmanned plane interference signal feature selection approach according to claim 1, it is characterised in that: institute in the 4th step
The variation expression formula of the inertia weight ω stated is as follows:
Wherein, ωminWith ωmaxThe minimum value and maximum value of inertia weight are respectively represented, f indicates this particle in current location
Fitness value, fmaxWith fargRespectively indicate population in this time iteration maximum adaptation angle value and average fitness value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910635110.8A CN110336637A (en) | 2019-07-15 | 2019-07-15 | A kind of unmanned plane interference signal feature selection approach |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910635110.8A CN110336637A (en) | 2019-07-15 | 2019-07-15 | A kind of unmanned plane interference signal feature selection approach |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110336637A true CN110336637A (en) | 2019-10-15 |
Family
ID=68145095
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910635110.8A Pending CN110336637A (en) | 2019-07-15 | 2019-07-15 | A kind of unmanned plane interference signal feature selection approach |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110336637A (en) |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103278326A (en) * | 2013-06-14 | 2013-09-04 | 上海电机学院 | Method for diagnosing faults of wind generating set gear case |
CN104200264A (en) * | 2014-09-25 | 2014-12-10 | 国家电网公司 | Two-stage particle swarm optimization algorithm including independent global search |
CN104900235A (en) * | 2015-05-25 | 2015-09-09 | 重庆大学 | Voiceprint recognition method based on pitch period mixed characteristic parameters |
CN105718943A (en) * | 2016-01-19 | 2016-06-29 | 南京邮电大学 | Character selection method based on particle swarm optimization algorithm |
CN106897733A (en) * | 2017-01-16 | 2017-06-27 | 南京邮电大学 | Video stream characteristics selection and sorting technique based on particle swarm optimization algorithm |
CN107590436A (en) * | 2017-08-10 | 2018-01-16 | 云南财经大学 | Radar emitter signal feature selection approach based on peplomer subgroup multi-objective Algorithm |
CN107682109A (en) * | 2017-10-11 | 2018-02-09 | 北京航空航天大学 | A kind of interference signal classifying identification method suitable for UAV Communication system |
US20180089566A1 (en) * | 2015-05-22 | 2018-03-29 | Huawei Technologies Co., Ltd. | Method and apparatus for positioning of artificial neural network |
CN108307444A (en) * | 2018-01-19 | 2018-07-20 | 扬州大学 | Wireless sense network UAV system communication means based on optimization particle cluster algorithm |
CN108664562A (en) * | 2018-04-10 | 2018-10-16 | 华东师范大学 | The text feature selection method of particle group optimizing |
CN108876086A (en) * | 2018-04-12 | 2018-11-23 | 南安市创培电子科技有限公司 | A kind of multiple no-manned plane method for allocating tasks based on particle swarm algorithm |
CN109145960A (en) * | 2018-07-27 | 2019-01-04 | 山东大学 | Based on the data characteristics selection method and system for improving particle swarm algorithm |
-
2019
- 2019-07-15 CN CN201910635110.8A patent/CN110336637A/en active Pending
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103278326A (en) * | 2013-06-14 | 2013-09-04 | 上海电机学院 | Method for diagnosing faults of wind generating set gear case |
CN104200264A (en) * | 2014-09-25 | 2014-12-10 | 国家电网公司 | Two-stage particle swarm optimization algorithm including independent global search |
US20180089566A1 (en) * | 2015-05-22 | 2018-03-29 | Huawei Technologies Co., Ltd. | Method and apparatus for positioning of artificial neural network |
CN104900235A (en) * | 2015-05-25 | 2015-09-09 | 重庆大学 | Voiceprint recognition method based on pitch period mixed characteristic parameters |
CN105718943A (en) * | 2016-01-19 | 2016-06-29 | 南京邮电大学 | Character selection method based on particle swarm optimization algorithm |
CN106897733A (en) * | 2017-01-16 | 2017-06-27 | 南京邮电大学 | Video stream characteristics selection and sorting technique based on particle swarm optimization algorithm |
CN107590436A (en) * | 2017-08-10 | 2018-01-16 | 云南财经大学 | Radar emitter signal feature selection approach based on peplomer subgroup multi-objective Algorithm |
CN107682109A (en) * | 2017-10-11 | 2018-02-09 | 北京航空航天大学 | A kind of interference signal classifying identification method suitable for UAV Communication system |
CN108307444A (en) * | 2018-01-19 | 2018-07-20 | 扬州大学 | Wireless sense network UAV system communication means based on optimization particle cluster algorithm |
CN108664562A (en) * | 2018-04-10 | 2018-10-16 | 华东师范大学 | The text feature selection method of particle group optimizing |
CN108876086A (en) * | 2018-04-12 | 2018-11-23 | 南安市创培电子科技有限公司 | A kind of multiple no-manned plane method for allocating tasks based on particle swarm algorithm |
CN109145960A (en) * | 2018-07-27 | 2019-01-04 | 山东大学 | Based on the data characteristics selection method and system for improving particle swarm algorithm |
Non-Patent Citations (10)
Title |
---|
JIE GUO: "An improved particle swarm optimization and its application in maneuvering control laws design of the unmanned aerial vehicle", 《2012 8TH INTERNATIONAL CONFERENCE ON NATURAL COMPUTATION》 * |
刘伟丽: "基于粒子群算法和支持向量机的中文文本分类研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
夏朋: "针对民用小型无人机的干扰与反制技术研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
尹良: "基于机器学习的无线电信号识别", 《无线网络规划与建设》 * |
李策: "基于自适应粒子群算法的特征选择", 《计算机技术与发展》 * |
李策: "基于自适应粒子群算法的特征选择研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
杨俊: "离散二进制粒子群算法在M|MO***检测中的应用", 《海南大学学报自然科学版》 * |
沈林成: "离散粒子群优化算法研究现状综述", 《***工程与电子技术》 * |
郭文忠: "《离散粒子群优化算法及其应用》", 31 May 2012, 清华大学出版社 * |
郭红敬: "无人机测控链路干扰感知与识别技术研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2022007376A1 (en) | Multi-objective multimodal particle swarm optimization method based on bayesian adaptive resonance | |
Hassanzadeh et al. | A new hybrid approach for data clustering using firefly algorithm and K-means | |
Bae et al. | Feature selection with intelligent dynamic swarm and rough set | |
CN107392919B (en) | Adaptive genetic algorithm-based gray threshold acquisition method and image segmentation method | |
CN106817248B (en) | APT attack detection method | |
CN112329934A (en) | RBF neural network optimization algorithm based on improved sparrow search algorithm | |
CN107798379B (en) | Method for improving quantum particle swarm optimization algorithm and application based on improved algorithm | |
CN106971091B (en) | Tumor identification method based on deterministic particle swarm optimization and support vector machine | |
CN109871872A (en) | A kind of flow real-time grading method based on shell vector mode SVM incremental learning model | |
Premalatha et al. | Hybrid PSO and GA models for document clustering | |
CN109697451A (en) | Similar image clustering method and device, storage medium, electronic equipment | |
CN105405136A (en) | Self-adaptive spinal CT image segmentation method based on particle swarm optimization | |
Wang et al. | A PSO-based classification rule mining algorithm | |
CN110287985B (en) | Depth neural network image identification method based on variable topology structure with variation particle swarm optimization | |
CN111275132A (en) | Target clustering method based on SA-PFCM + + algorithm | |
CN110751121A (en) | Unsupervised radar signal sorting method based on clustering and SOFM | |
CN109711373A (en) | A kind of big data feature selection approach based on improvement bat algorithm | |
CN109871934A (en) | Feature selection approach based on the distributed parallel binary of Spark a flying moth darts into the fire algorithm | |
Ehsan et al. | Vi-Net: a deep violent flow network for violence detection in video sequences | |
CN114189899B (en) | User equipment selection method based on random aggregation beam forming | |
CN117290721A (en) | Digital twin modeling method, device, equipment and medium | |
Mishra et al. | Kohonen self organizing map with modified k-means clustering for high dimensional data set | |
Shi et al. | PSO-based community detection in complex networks | |
CN110336637A (en) | A kind of unmanned plane interference signal feature selection approach | |
CN117376066A (en) | Asynchronous sampling-oriented digital signal modulation mode identification method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191015 |
|
WD01 | Invention patent application deemed withdrawn after publication |