CN106912105A - 3-D positioning method based on PSO_BP neutral nets - Google Patents
3-D positioning method based on PSO_BP neutral nets Download PDFInfo
- Publication number
- CN106912105A CN106912105A CN201710133596.6A CN201710133596A CN106912105A CN 106912105 A CN106912105 A CN 106912105A CN 201710133596 A CN201710133596 A CN 201710133596A CN 106912105 A CN106912105 A CN 106912105A
- Authority
- CN
- China
- Prior art keywords
- terminal
- pso
- error
- base station
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W64/00—Locating users or terminals or network equipment for network management purposes, e.g. mobility management
- H04W64/006—Locating users or terminals or network equipment for network management purposes, e.g. mobility management with additional information processing, e.g. for direction or speed determination
Abstract
The present invention devises a kind of terminal 3-D positioning method of the BP neural network after optimization based on particle cluster algorithm (Particle Swarm Optimizer, PSO), can be widely applied to wireless positioning field.This method process is：Measure the range data of multiple base stations and terminal in certain region.Measured distance is sorted from small to large, four nearest base stations of chosen distance calculate the terminal location influenceed containing non line of sight with this four base stations by least square method.All terminal locations containing non line of sight are calculated again, and calculate each base station to the three-dimensional angle of terminal.Finally using gained terminal location coordinate, base station to terminal distance and three-dimensional angle as the characteristic value input layer of PSO_BP neutral nets, output layer is revised terminal location coordinate.Using PSO algorithm optimizations BP neural network, acquired results eliminate terminal location measurement error that non line of sight factor is brought to the present invention, and the algorithm of proposition has the advantages that stable performance, algorithmic statement be fast, positioning precision is high, is adapted to promote the use of.
Description
Technical field
The present invention devises a kind of terminal 3-D positioning method of the BP neural network after optimization based on particle cluster algorithm, can
It is widely used in wireless positioning field.
Background technology
With flourishing for cordless communication network and mobile Internet, there is provided the service based on geographical location information
(Location Based Service, abbreviation LBS) has become one of business of most market prospects and development potentiality.From biography
The GPS navigation of system, to ecommerce, food and drink platform etc., the consumption information service based on geographical position and social software, realize it
The basis of function seeks to obtain the position of user's handheld terminal (including the equipment such as mobile phone or flat board).
Although commercialization GPS is widely used with the development of smart mobile phone, it is such as indoor,
Under, in many scenes such as built-up urban district, GPS location poor-performing.Alignment system based on wireless network base station is compared
GPS has that wide coverage, signal quality be high, user expects the strong advantage of its mobile terminal wish, meanwhile, based on wireless network
Network base station, the profit model of operator's propulsion positioning service is clear, outside the data, services on basis, can also be by being user
Value-added service is provided and promotes the business development of operator.Therefore, the location technology based on radio communication base station has wide
Application prospect and huge commercial value.
Limited by conditions such as positioning time, positioning precision and complex indoor environments, especially in three-dimensional localization environment, than
More perfect location technology cannot also be utilized well at present.By taking indoor environment as an example, meeting in the communication process of radio signal
By the multiple reflections of metope, the refraction of indoor object and absorption etc..These physical factors can cause communication base station measurement to obtain
The information such as distance, angle there is noise.How these noisy measurements are based on, obtain accurate for positional information
Estimate, be also that communication base station realization needs the problem of solution to terminal positioning.
Mobile terminal location relevant issues has novelty and the strong algorithm of realizability under the conditions of solution analyzing three-dimensional, all
In would be possible to by rapid deployment to modern commerce communication network, huge social and economic benefit is brought.
The content of the invention
The technical problem to be solved in the present invention is directed to the terminal positioning under above-mentioned many base station cases indoor in the prior art
Method precision problem not high and a kind of terminal 3-D positioning method based on PSO_BP neutral nets for proposing.
Can there are multiple base stations in one region, from the location of terminal (mxi,myi,mzi) from the close-by examples to those far off selection before
Four base stations in the same plane, r is not respectively between terminal to base station using the distance of TDOA technologies actual measurement1i,r2i,r3i,
r4i, there is r1i≤r2i≤r3i≤r4i。
By the position of the measured distance computing terminal containing non-market value of four base stations to mobile terminal, using least square
Method is estimated.Then the least-squares estimation position of three-dimensional coordinate is
Using AOA methods, calculate by base station coordinates (bxj, byj, bzj) arrive the three-dimensional terminal position that estimation is participated in by the base station
PutIdentified three-dimensional angle
Site error is produced with azimuth angle error and produced by 2 parts, respectively by non-line-of-sight propagation error
And measurement error2 parts constitute.
BP neural network is built, the input feature vector value of network is terminal location, base station to distance between terminal, three-dimensional AOA
Deflection.Network is multilayer, is determined after hidden layer configuration is to be optimized, and output layer is the end point position of final prediction.
Using PSO algorithm optimization BP neural networks, PSO_BP neural network algorithm models are set up.
Sample data after being processed using the part of collection carries out the error analysis of PSO_BP neutral nets, by nerve net
The error analysis of network determines a preferable network structure.
Using the data sample that characteristic value is converted into after all treatment, BP nerve nets are trained for the first time first with PSO algorithms
Network, then using gradient descent method double optimization, the final weight threshold parameter for determining network.The real-time performance is from feature
Vector M is to end point exact position (x in samplei,yi,zi) a Function Mapping relation, eliminate what environment non line of sight brought
Error, obtains more accurate end point position.
The present invention first position of terminal according to a preliminary estimate in three dimensions according to many base stations, be extracted estimated location coordinate,
Base station to the distance of terminal and base station to the three dimensional orientation angle of terminal as characteristic value input vector, this group of characteristic value to
Amount preferably reflects terminal location and base station can reflect the phase of non line of sight factor and measurement error influence with terminal relativeness
Pass factor.Extract characteristic quantity data set from sample point to be trained in BP neural network, the optimizing of error function is calculated according to PSO
Method is carried out, and when a region is converged to, then is iterated near error function minimum point with small step-length with gradient decline.For
Whether error function carries out error analysis, and the structure of network is determined come comprehensive by learning curve, and increases the behaviour such as data volume
Make.After determining network structure and training network, more accurate end in the case of be eliminated non-market value and measurement error
End position point.The method fast convergence rate, accuracy are high, adapt to various specific environments, are adapted to promote the use of.
Brief description of the drawings
Fig. 1 is the method for the invention implementation process flow chart；
Fig. 2 is the schematic diagram that four base stations determine terminal location；
Fig. 3 be in the present invention base sites to the end point three-dimensional azimuthal schematic diagrames of AOA；；
Fig. 4 is the topological diagram of BP networks in the present invention；
Fig. 5 is using PSO algorithm optimization BP neural network parametric procedure schematic diagrames in the present invention；
Fig. 6 is the flow chart of the terminal 3-D positioning method algorithmic procedure based on PSO_BP neutral nets in the present invention.
Specific embodiment
The detailed of further volume will be made to the three-dimensional method of locating terminal based on PSO_BP neutral nets of the invention below
Description, Fig. 1 is implementation process flow chart of the present invention, implements step as follows：
S1. there is the influence of non line of sight to during the distance of terminal by Radio Signal Propagation calculation base station, it is this
Influence can make to bring deviation during many architecture end errors.Can there are multiple base stations, the position residing for terminal in one region
Put (mxi,myi,mzi) from the close-by examples to those far off selection before four base stations, four base station coordinates are respectively designated as (bx in order1,by1,
bz1), (bx2,by2,bz2),(bx3,by3,bz3),(bx4,by4,bz4).Fig. 2 is the signal that four base stations determine terminal location
Figure, range finding coverage rate of base station one (A), range finding coverage rate of base station two (B), range finding coverage rate of base station three (C) and the range finding of base station four are covered
It is the sphere of the centre of sphere that capping (D) is respectively with four base stations, and the coverage rate of four base stations meets at terminal M, when four base stations do not exist
In approximately the same plane, r is respectively using the distance of TDOA technologies actual measurement between terminal to base station1i,r2i,r3i,r4i, there is r1i≤r2i
≤r3i≤r4i.The present invention discusses the situation of four base stations not planar, current four base sites be in a plane even
It is continuous to choose preceding 5 base stations, by that analogy.
S2. by the position of the measured distance computing terminal containing non-market value of four base stations to mobile terminal, due to existing
Measurement error, four balls may not give a point, to reduce the influence of error, be estimated using least square method.It is then three-dimensional to sit
Target least-squares estimation position is
In S2, when the measurement of the distance between base station to terminal is error free, a point can be met at, the distance according to point-to-point transmission is public
Formula lists four equations：
Then coordinate (the mx of i pointsi,myi,mzi) matrix be expressed as：
Due to there is measurement error, four balls can not give a point, in order to reduce the influence of error, using least square method pair
Data are processed.The least-squares estimation position of three-dimensional coordinate isThen
Solved with least square method：
Wherein
Due to present invention research is 4 non-coplanar base stations, therefore n=4.Can be according to this formula more than 4 situations of base station
Calculate.
S3. AOA methods are utilized, is calculated by base station coordinates (bxj, byj, bzj) arrive the three-dimensional terminal that estimation is participated in by the base station
PositionIdentified three-dimensional angleIf Fig. 3 is base sites to the end point three-dimensional azimuthal signals of AOA
Figure, represents base station MiTo terminal BiThree-dimensional position angle,It is free from error three-dimensional angle.Then PSO_BP neutral nets
Input feature vector value vector be M.
Due to present invention research is therefore the n=4 to be special case based on 4 non-coplanar base stations.More than 4 feelings of base station
Condition can be represented according to this general type.
Wherein
In S3, three-dimensional position angleCalculating process be：
Then
Have
S4. site error and azimuth angle error are produced and are all made up of 2 parts, wherein the estimated location coordinate of terminalWith error free position coordinates (xi,yi,zi) between error be non-line-of-sight propagation errorMissed with measurement
DifferenceThen have
From base station location coordinate (bxj, byj, bzj) arrive terminal estimated locationThree-dimensional angleAlso include
Due to the error of non line of sight factor influenceWith measurement errorHave WhereinIt is free from error three-dimensional angle.
S5. BP neural network is built, BP neural network is made up of input layer, hidden layer and output layer.Input feature vector value to
It is M to measure.Input feature value is tieed up for 3n+4.N=4 in the present invention.Output layer is 3-dimensional, and output layer is required accurate terminal position
Put coordinate.Hidden layer is multilayer, and the error analysis by neural network learning curve of hidden layer configuration determines.If Fig. 4 is the BP networks
Topological diagram.
In S5, the hidden layer transmission function tansig functions of BP neural network, its expression formula is：
Input value can be arbitrary value, and output valve codomain is [- 1,1].Output layer is made up of 3 neurons, and output layer is using linear biography
Delivery function fo(x)=kx.Output vector is [xi, yi,zi]T, this vector element is to eliminate the coordinate exact value after error.If
The corresponding physical locations of sample point i are
If Pi otuput=[xi, yi, zi]T,
Then error function isW and B are the weights and threshold parameter of network.| | | | it is
Norm sign, m is input sample number.
The process of S6.PSO_BP neural network algorithms is to build BP neural network according to S5, using PSO algorithms in given feature
Search error function global minimum solution in the domain of definition of value, when PSO algorithmic statements are to a zonule and after terminate iteration, adopts
The mode part declined with gradient seeks minimum value solution so that smaller step-length is secondary.If Fig. 5 is PSO algorithm optimizations BP neural network ginseng
The schematic diagram of number process.
In S6, the specific mathematical description of PSO algorithms is：
In D dimensions search space, a colony includes N number of particle, is counted as X=[x1,...,xN]T, the position rate of particle i
It is denoted as vi=[vi1,vi2,...,viD]T, its position is denoted as xi=[xi1,xi2,...,xiD]T, wherein i=1,2 ..., N；Particle i
The optimal location that itself finds to current iteration is denoted as pi=[pi1, pi2,...,piD]T, all particles hair to current
Existing optimal location is denoted as pg=[pg1,pg2,...,pgD]T, after the two values are found, particle is carried out more by following equation
Newly：
vid(t+1)=ω (t) vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t))
xid(t+1)=xid(t)+vid(t+1)
Wherein,
In formula, i=1,2 ..., N, d=1,2 ..., D；c1,c2It is Studying factors, is nonnegative constant, is typically set to 2；r1,r2
It is the random number in [0,1]；ω (t) is inertia weight；T is iterations；TmaxIt is maximum iteration；vid(t),xid
T () is respectively the current speed of particle i and position；pidIt is the personal best particle of particle i discoveries；
It is as shown in Figure 6 the flow chart of the terminal 3-D positioning method algorithmic procedure based on PSO_BP neutral nets, PSO_
BP algorithm is comprised the following steps that：
(1) initiation parameter.These parameters include scale, iterations, Studying factors, speed, the position of population.
(2) BP neural network is set up according to input, hidden layer, output, excitation function.A population X is generated at randomi=[xi1,
xi2...,xiN]T, i=1,2 ..., n, the element that each particle is included are all of weights W and threshold value B in BP neural network,
Element is placed in one-dimensional vector X by certain tandem rule of correspondence in W and B.
In formula, i is the sequence number of particle, and n is the population number of particle.
(3) by the fitness value of the evaluation function of BP network calculations each particles.Initialization BP neural network parameter, (2)
The weights and threshold value that each particle of the population of middle determination is included are brought into BP neural network respectively.Input feature vector value square
Battle array is to obtaining training output vector P in the BP networks of each particlei output, now the fitness value of the particle is the error of network
Function E (W, B)：
In formula, Pi outputIt is network output vector；Pi realTo expect output vector；| | | | it is norm sign, m is input sample
Number.
(4) X is calculatedi(i=1,2 ..., n) corresponding fitness value, fitness value to current particle and previous optimal suitable
Answer angle value to be compared, less value in the two is substituted for the local extremum of current particle.Then all particles are selected to adapt to
Angle value it is minimum as current global extremum.
(5) updated in iterative process each time according to the position described in PSO algorithm mathematics models and the iterative formula of speed
The speed of each particle and position；
Speed more new formula：vid(t+1)=ω (t) vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t))
Location updating formula：xid(t+1)=xid(t)+vid(t+1)
(6) fitness value of new particle is calculated, according to population each particle fitness value now according to (4th) step more new particle
Individual extreme value and colony's extreme value.
(7) meet maximum iteration or meet the error criterion backed off after random PSO algorithms of setting, otherwise return to step (2).
(8) weights and threshold value for being included the optimal particle that particle cluster algorithm is obtained are assigned to BP networks and carry out double optimization, two
It is a process declined by gladient rule that optimal solution is found during suboptimization, is to connect by the weights and threshold value of PSO Algorithm for Training
Nearly global optimum, this is due to may completely not reached most because its flight step-length is long when PSO algorithms terminate
It is excellent, a search for the BP networks of the learning efficiency of very little small step-length of further small range on last position is set herein.
If the result of search so exports this result better than the result before search, otherwise keep original result constant.
S7. the sample data after being processed using the part of collection carries out the error analysis of PSO_BP neutral nets, by god
Determine a preferable network hidden layer configuration through the error analysis of network.
The hidden layer configuration of neutral net is relevant with scene in S7, and scene is more complicated, then network structure should be more complicated.By nerve
Variance in the learning curve of network and aberration curve can be seen that network whether poor fitting or over-fitting.For different feelings
Condition takes different measure adjustable strategies.These conventional measures include singly being not limited to：(1) extraction of new feature amount；(2) train
The data volume of sample；(3) size of the penalty term factor；(4) the network number of plies or neuron number purpose increase and decrease.
When setting up a neural network algorithm and starting, a simple network is built first, train simple network, drafting
Learning curve, what kind of the learning curve large deviations and variance curve for judging simple network are.Determined to take according to learning curve
Which kind of measure or which measure coordinate is improved neutral net so that neutral net has appropriate fitting to show, so as to improve net
The Generalization Capability of network.
S8. using the data sample that characteristic value is converted into after all treatment, training PSO_BP neutral nets determine network
Weight threshold parameter.So as to be believed by error caused by non line of sight factor and measurement error in the specific specific environment of the e-learning
Breath, finally realizes from characteristic vector.
The invention provides a kind of terminal three-dimensional accurate positioning method based on PSO_BP neutral nets, it is applicable and various spies
There is more irregular slalom thing in fixed environment, particularly environment, scattering and multipath phenomenon are obvious, such as megastore, underground tunnel
In hole, the environment such as lofty mountains area, be a kind of practical location technology.The method is sat according to actual measurement terminal location in three dimensions
Mark information comes the accurate position of computing terminal.From actual measurement terminal location in be extracted terminal point coordinates, base station to terminal away from
From and base station to the three-dimensional position angle of terminal as characteristic value, the essence of target is calculated by PSO_BP neural network algorithms
True position.The invention provides the non line of sight that three-dimensional method of locating terminal effectively can be brought by neural network learning to environment
Error influences and the error brought of measurement influences, with feasibility it is high, adapt to specific environment, high precision advantage.Can promote the use of.
Particular embodiments described above, has been carried out further in detail to the purpose of the present invention, technical scheme and beneficial effect
Describe in detail bright, should be understood that and the foregoing is only specific embodiment of the invention, be not intended to limit the invention, it is all
Within the spirit and principles in the present invention, any modification, equivalent substitution and improvements done etc., should be included in guarantor of the invention
Within the scope of shield.
Claims (6)
1. a kind of terminal 3-D positioning method based on PSO_BP neutral nets, it is characterised in that：Patent of the present invention can be used for from
The mobile terminal radio signal that many base stations receive orients the accurate position in base station；The important step of this method is to utilize
The good fit ability of BP neural network is eliminated due to the terminal positioning error that non line of sight influence is caused, and utilizes PSO algorithms
Global optimizing ability overcomes the local optimal searching defect of BP neural network, and BP neural network importation using distance, estimate
Meter position, three-dimensional angle improve network learning and training to end point locating effect as characteristic value., realize interior many
Mobile terminal location under base station case, specifically includes：
S1. there is the influence of non line of sight, this influence to during the distance of terminal by Radio Signal Propagation calculation base station
Can make to bring deviation during many architecture end errors；Can there are multiple base stations in one region, from the location of terminal
(mxi,myi,mzi) from the close-by examples to those far off selection before four base stations, four base station coordinates are respectively designated as (bx in order1,by1,bz1),
(bx2,by2,bz2),(bx3,by3,bz3),(bx4,by4,bz4), when four base stations are not in approximately the same plane, terminal to base station
Between using TDOA technologies actual measurement distance be respectively r1i,r2i,r3i,r4i, there is r1i≤r2i≤r3i≤r4i.The present invention discusses four
The situation of individual base station not planar, current four base sites are in a plane just continuous preceding 5 base stations of selection, with such
Push away；
S2. by the position of the measured distance computing terminal containing non-market value of four base stations to mobile terminal, measured due to existing
Error, four balls may not give a point, be the influence for reducing error, be estimated using least square method, then three-dimensional coordinate
Least-squares estimation position is
S3. AOA methods are utilized, is calculated by base station coordinates (bxj,byj,bzj) arrive the three-dimensional terminal location that estimation is participated in by the base stationIdentified three-dimensional angleThen the input feature vector value vector of PSO_BP neutral nets is M, due to this
Invention research is therefore the n=4 to be special case based on 4 non-coplanar base stations；Can be general according to this more than 4 situations of base station
Form represents, wherein
S4. site error and azimuth angle error are produced and are all made up of 2 parts, wherein the estimated location coordinate of terminal
With error free position coordinates (xi,yi,zi) between error be non-line-of-sight propagation errorAnd measurement errorThen have
From base station location coordinate (bxj,byj,bzj) arrive terminal estimated locationThree-dimensional angleAlso include
Due to the error of non line of sight factor influenceWith measurement errorHave WhereinIt is free from error three-dimensional angle；
S5. BP neural network is built, BP neural network is made up of input layer, hidden layer and output layer；Input feature vector value vector is
M, input feature value is tieed up for 3n+4, n=4 in the present invention, and output layer is 3-dimensional, and output layer is that required accurate terminal location is sat
Mark, hidden layer is multilayer, and the error analysis by neural network learning curve of hidden layer configuration determines；
The process of S6.PSO_BP neural network algorithms is to build BP neural network according to S5, using PSO algorithms in given characteristic value
Search error function global minimum solution in domain of definition, when PSO algorithmic statements are to a zonule and after terminate iteration, using ladder
Spend the mode for declining and locally seek minimum value solution so that smaller step-length is secondary；
S7. the sample data after being processed using the part of collection carries out the error analysis of PSO_BP neutral nets, by nerve net
The error analysis of network determines a preferable network structure；
S8. using the data sample that characteristic value is converted into after all treatment, training PSO_BP neutral nets determine the weights of network
Threshold parameter, so that by control information caused by non line of sight factor and measurement error in the specific specific environment of the e-learning, most
After realize from characteristic vector M to sample in exact position (xi,yi,zi) a Function Mapping relation, eliminate error, obtain
Compared with exact position.
2. a kind of terminal 3-D positioning method based on PSO_BP neutral nets according to claim 1, it is characterised in that：
In claim 1 step S1 simultaneously in the case where there is multiple base stations to position, self adaptation chooses four nearest base stations to terminal
Positioned, and to sequence that four base stations to the position of terminal carry out from small to large according to the distance that TDOA technologies are surveyed；Cause
It is that distance is nearer, is influenceed to produce error smaller by non line of sight, so that the evaluated error as position is also smaller.
3. a kind of terminal 3-D positioning method based on PSO_BP neutral nets according to claim 1, it is characterised in that：
In claim 1 step S3, using the three-dimensional terminal location coordinate estimated in step S2Calculate three-dimensional anglePosition coordinates, three-dimensional angle and the base station estimated using least square method constitute god to the range data of terminal room
Through the input feature value M of network, this feature vector combination is more complete, and position accurately can be predicted, therefore can be more comprehensive
The information of terminal location after reflection non line of sight influence.
4. a kind of terminal 3-D positioning method based on PSO_BP neutral nets according to claim 1, it is characterised in that：
In claim 1 step S5, the optimizing of error function is carried out to the BP neural network built in S4 using PSO algorithms；BP nerves
Network carries out local optimal searching using gradient descent method in the learning process of backpropagation to error function, relies heavily on
In the selection of initial feasible solution, and PSO algorithms can in the domain of definition of characteristic value global optimizing.
5. a kind of terminal 3-D positioning method based on PSO_BP neutral nets according to claim 1, it is characterised in that：
The detailed process of PSO_BP algorithms is in claim 1 step S6：
S6-1. using PSO algorithms to the error function global optimizing of the BP neural network in S5, PSO algorithms have good complete
The performance of office's optimizing；
S6-2. gradient descent algorithm has good local optimal searching performance；PSO algorithms are with larger particle flight step-length in the overall situation
Scope inner iteration, when iteration reaches end condition, further with the gradient descent algorithm part of less Studying factors
Optimizing, reaches the exact value for causing that error function is minimum, so as to obtain more preferable weights and Network of Threshold parameter.
6. a kind of terminal 3-D positioning method based on PSO_BP neutral nets according to claim 1, it is characterised in that：
In claim 1 step S7, the hidden layer configuration of neutral net is relevant with scene, and scene is more complicated, then network structure should be more complicated；
By the variance in the learning curve of neutral net and aberration curve can be seen that network whether poor fitting or over-fitting；For
Different situations takes different measure adjustable strategies.These measures include：(1) dimension of characteristic quantity；(2) number of characteristic value
According to amount；(3) penalty term size；(4) the network number of plies or neuron number purpose increase and decrease.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710133596.6A CN106912105B (en) | 2017-03-08 | 2017-03-08 | Three-dimensional positioning method based on PSO _ BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710133596.6A CN106912105B (en) | 2017-03-08 | 2017-03-08 | Three-dimensional positioning method based on PSO _ BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106912105A true CN106912105A (en) | 2017-06-30 |
| CN106912105B CN106912105B (en) | 2020-06-09 |
Family
ID=59187298
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710133596.6A Expired - Fee Related CN106912105B (en) | 2017-03-08 | 2017-03-08 | Three-dimensional positioning method based on PSO _ BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106912105B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108132458A (en) * | 2017-12-22 | 2018-06-08 | 北京锐安科技有限公司 | Indoor distance measuring method, device, equipment and storage medium |
| CN108551653A (en) * | 2018-04-19 | 2018-09-18 | 北京邮电大学 | A kind of indoor orientation method, device, electronic equipment and storage medium |
| CN108924925A (en) * | 2018-07-09 | 2018-11-30 | 上海公路桥梁（集团）有限公司 | A kind of construction personnel's localization method and its positioning system |
| CN109257816A (en) * | 2017-07-12 | 2019-01-22 | 华为技术有限公司 | A kind of localization method and relevant device |
| CN109766969A (en) * | 2019-01-16 | 2019-05-17 | 广西大学 | RFID indoor positioning algorithms based on asynchronous advantage action evaluation |
| CN109857351A (en) * | 2019-02-22 | 2019-06-07 | 北京航天泰坦科技股份有限公司 | The Method of printing of traceable invoice |
| CN109887202A (en) * | 2019-02-22 | 2019-06-14 | 北京航天泰坦科技股份有限公司 | The tax control tray of built-in enhancing module |
| CN109934597A (en) * | 2019-02-22 | 2019-06-25 | 北京航天泰坦科技股份有限公司 | External tax control tray attachment device |
| CN109993877A (en) * | 2019-03-07 | 2019-07-09 | 北京航天泰坦科技股份有限公司 | Anti-fake bill recognition methods based on location information |
| CN110418278A (en) * | 2019-07-25 | 2019-11-05 | 李印 | A kind of 3 D positioning system based on Evolutionary Neural Network |
| CN110611878A (en) * | 2019-08-23 | 2019-12-24 | 华清科盛(北京)信息技术有限公司 | High-precision positioning system and method based on one-dimensional deployment of base station |
| CN110784823A (en) * | 2019-11-05 | 2020-02-11 | 西安电子科技大学 | Outdoor target positioning method based on BP neural network and TDOA |
| CN112040394A (en) * | 2020-08-05 | 2020-12-04 | 深圳市微能信息科技有限公司 | Bluetooth positioning method and system based on AI deep learning algorithm |
| WO2021027948A1 (en) * | 2019-08-15 | 2021-02-18 | 华为技术有限公司 | Positioning method and device thereof |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101466145A (en) * | 2009-01-04 | 2009-06-24 | 上海大学 | Dual-base-station accurate orientation method based on neural network |
| CN101695190A (en) * | 2009-10-20 | 2010-04-14 | 北京航空航天大学 | Three-dimensional wireless sensor network node self-locating method based on neural network |
| CN105307264A (en) * | 2015-07-27 | 2016-02-03 | 河南科技大学 | Mobile node positioning method for wireless sensor network |
-
2017
- 2017-03-08 CN CN201710133596.6A patent/CN106912105B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101466145A (en) * | 2009-01-04 | 2009-06-24 | 上海大学 | Dual-base-station accurate orientation method based on neural network |
| CN101695190A (en) * | 2009-10-20 | 2010-04-14 | 北京航空航天大学 | Three-dimensional wireless sensor network node self-locating method based on neural network |
| CN105307264A (en) * | 2015-07-27 | 2016-02-03 | 河南科技大学 | Mobile node positioning method for wireless sensor network |
Non-Patent Citations (3)
| Title |
|---|
| LILING YU 等: "Eye-Gaze Tracking System Based on Particle Swarm Optimization and BP Neural network", 《2016 12TH WORLD CONGRESS ON INTELLIGENT CONTROL AND AUTOMATION》 * |
| 卞国龙: "基于PSO-BP 算法的无线传感器网络定位优化", 《电讯技术》 * |
| 闫驰: "基于PSO-BP神经网络的无线传感器网络定位算法", 《电子科技》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109257816A (en) * | 2017-07-12 | 2019-01-22 | 华为技术有限公司 | A kind of localization method and relevant device |
| CN109257816B (en) * | 2017-07-12 | 2023-12-08 | 华为技术有限公司 | Positioning method and related equipment |
| CN108132458A (en) * | 2017-12-22 | 2018-06-08 | 北京锐安科技有限公司 | Indoor distance measuring method, device, equipment and storage medium |
| CN108132458B (en) * | 2017-12-22 | 2020-07-17 | 北京锐安科技有限公司 | Indoor distance measuring method, device, equipment and storage medium |
| CN108551653A (en) * | 2018-04-19 | 2018-09-18 | 北京邮电大学 | A kind of indoor orientation method, device, electronic equipment and storage medium |
| CN108551653B (en) * | 2018-04-19 | 2020-01-17 | 北京邮电大学 | Indoor positioning method and device, electronic equipment and storage medium |
| CN108924925A (en) * | 2018-07-09 | 2018-11-30 | 上海公路桥梁（集团）有限公司 | A kind of construction personnel's localization method and its positioning system |
| CN109766969A (en) * | 2019-01-16 | 2019-05-17 | 广西大学 | RFID indoor positioning algorithms based on asynchronous advantage action evaluation |
| CN109766969B (en) * | 2019-01-16 | 2022-01-07 | 广西大学 | RFID indoor positioning algorithm based on asynchronous dominant motion evaluation |
| CN109887202A (en) * | 2019-02-22 | 2019-06-14 | 北京航天泰坦科技股份有限公司 | The tax control tray of built-in enhancing module |
| CN109934597A (en) * | 2019-02-22 | 2019-06-25 | 北京航天泰坦科技股份有限公司 | External tax control tray attachment device |
| CN109857351A (en) * | 2019-02-22 | 2019-06-07 | 北京航天泰坦科技股份有限公司 | The Method of printing of traceable invoice |
| CN109993877A (en) * | 2019-03-07 | 2019-07-09 | 北京航天泰坦科技股份有限公司 | Anti-fake bill recognition methods based on location information |
| CN110418278A (en) * | 2019-07-25 | 2019-11-05 | 李印 | A kind of 3 D positioning system based on Evolutionary Neural Network |
| WO2021027948A1 (en) * | 2019-08-15 | 2021-02-18 | 华为技术有限公司 | Positioning method and device thereof |
| CN110611878A (en) * | 2019-08-23 | 2019-12-24 | 华清科盛(北京)信息技术有限公司 | High-precision positioning system and method based on one-dimensional deployment of base station |
| CN110611878B (en) * | 2019-08-23 | 2021-10-19 | 华清科盛(北京)信息技术有限公司 | High-precision positioning system and method based on one-dimensional deployment of base station |
| CN110784823B (en) * | 2019-11-05 | 2021-01-05 | 西安电子科技大学 | Outdoor target positioning method based on BP neural network and TDOA |
| CN110784823A (en) * | 2019-11-05 | 2020-02-11 | 西安电子科技大学 | Outdoor target positioning method based on BP neural network and TDOA |
| CN112040394A (en) * | 2020-08-05 | 2020-12-04 | 深圳市微能信息科技有限公司 | Bluetooth positioning method and system based on AI deep learning algorithm |
| CN112040394B (en) * | 2020-08-05 | 2022-06-10 | 深圳市微能信息科技有限公司 | Bluetooth positioning method and system based on AI deep learning algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106912105B (en) | 2020-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106912105A (en) | 3-D positioning method based on PSO_BP neutral nets | |
| CN108512621B (en) | Wireless channel modeling method based on neural network | |
| CN107318084B (en) | Fingerprint positioning method and device based on optimal similarity | |
| CN106851573A (en) | Joint weighting k nearest neighbor indoor orientation method based on log path loss model | |
| CN105704652B (en) | Fingerprint base acquisition and optimization method in a kind of positioning of WLAN/ bluetooth | |
| KR101213363B1 (en) | Wireless localization method using 4 or more anchor nodes based on RSSI at indoor environment and a recording medium in which a program for the method is recorded | |
| CN108141837A (en) | For the device and method of tracking | |
| CN108828643B (en) | Indoor and outdoor seamless positioning system and method based on grey prediction model | |
| CN103885030A (en) | Locating method of mobile node in wireless sensor network | |
| CN109640262B (en) | Positioning method, system, equipment and storage medium based on mixed fingerprints | |
| CN112748397A (en) | UWB positioning method based on self-adaptive BP neural network under non-line-of-sight condition | |
| CN106093849A (en) | A kind of Underwater Navigation method based on range finding with neural network algorithm | |
| CN103888979B (en) | Indoor positioning method based on wireless local area network | |
| Xie et al. | BPNN based indoor fingerprinting localization algorithm against environmental fluctuations | |
| CN106792516B (en) | 3-D positioning method based on radio communication base station | |
| CN108650629A (en) | A kind of indoor three-dimensional location based on radio communication base station | |
| Huan et al. | Indoor location fingerprinting algorithm based on path loss parameter estimation and bayesian inference | |
| CN103096465A (en) | Environment self-adaption multi-target direct locating method | |
| CN113259884B (en) | Indoor positioning base station layout optimization method based on multi-parameter fusion | |
| Chen et al. | TDOA/AOA Hybrid Localization Based on Improved Dandelion Optimization Algorithm for Mobile Location Estimation Under NLOS Simulation Environment | |
| CN111263295B (en) | WLAN indoor positioning method and device | |
| CN113852433B (en) | Wireless channel shadow fading model prediction method based on computer vision | |
| CN116359838A (en) | Positioning resolving model for indoor low-power consumption Bluetooth | |
| Cao | A localization algorithm based on particle swarm optimization and quasi-newton algorithm for wireless sensor networks | |
| CN115226027A (en) | WiFi indoor fingerprint positioning method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200609 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |