CN109635419A - Laser atmospheric turbulence Transmission characteristics method based on machine learning - Google Patents
Laser atmospheric turbulence Transmission characteristics method based on machine learning Download PDFInfo
- Publication number
- CN109635419A CN109635419A CN201811499782.2A CN201811499782A CN109635419A CN 109635419 A CN109635419 A CN 109635419A CN 201811499782 A CN201811499782 A CN 201811499782A CN 109635419 A CN109635419 A CN 109635419A
- Authority
- CN
- China
- Prior art keywords
- variance
- signal strength
- light signal
- neural network
- step102
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The present invention discloses a kind of laser atmospheric turbulence Transmission characteristics method based on machine learning.This method executes first under different extraneous transmission conditions and receives light signal strength fluctuating normalization variance measurement operation, create the training sample set of multilayer feedforward neural network；Then it with the training sample set training multilayer feedforward neural network being created that, approaches multilayer feedforward neural network and receives the functional relation that light signal strength rises and falls between normalization variance and extraneous transmission conditions；The reception light signal strength fluctuating normalization variance under specific extraneous transmission conditions is finally obtained using multilayer feedforward neural network.Since this method uses dimensionless parameter as the input of multilayer feedforward neural network, the multilayer feedforward neural network that this method is established can be used for predicting that those did not carry out the reception light signal strength fluctuating normalization variance under the extraneous transmission conditions of experiment measurement.
Description
Technical field
The invention belongs to atmospheric channel optical signal transmission technical fields, and it is rapid to be related to a kind of atmospheric laser based on machine learning
Spread defeated characteristic analysis method.
Background technique
Laser will appear optical flare phenomenon when transmitting in atmospheric turbulance, so that the light signal strength hair that photoreceiver receives
Raw random fluctuation.It is to measure Laser Beam Through Onflow Atmosphere Channel journey that the light signal strength received, which rises and falls and normalizes variance,
One important indicator of degree, is defined asP indicates that the light signal strength received, angle brackets indicate
Seek ensemble average.
Scholars did a large amount of solutions to reception light signal strength fluctuating normalization variance of the laser after atmospheric turbulance transmits
Analyse theoretical research.However, current analytic theory research is only compared under weak turbulent-flow conditions or extremely strong turbulent-flow conditions
Reliable result.Other than analytic theory research, experiment measurement is another means of Study of Laser atmospheric turbulance transmission, it
By measuring a large amount of reception light signal strength fluctuating sampled values after atmospheric turbulance transmits, then again with statistical analysis technique meter
It calculates and receives the normalization variance that light signal strength rises and falls.Experimental measurements can reflect the reception light in Real Atmosphere Turbulence Channels
Signal strength fluctuating statistical property, but each measurement result is corresponding with specific extraneous transmission conditions.How basis
Existing experimental measurements, the reception light signal strength fluctuating estimated under the extraneous transmission conditions that other did not carried out measurement are returned
One change variance is a problem.The present invention will provide a kind of method to solve this technical problem.
" the Laser beam propagation through random published according to SPIE publishing house in 2005
media,2ndEdition ", for Propagation of Gaussian Beam, usually described on the plane of departure with two dimensionless groups
Gaussian beam, i.e. Fresnel ratio Λ0(Fresnel Ratio) and curvature parameters Θ0(Curvature Parameter), wherein
Λ0=2L/ (kW0 2), Θ0=1-L/F0, L expression transmission range, k expression light wave wave number, W0Indicate the Gauss light on the plane of departure
Beam radius, F0Indicate the Gaussian beam wave-front curvature radius on the plane of departure.Practical photoreceiver is usually with certain size
Circular aperture collect optical signal, it is assumed that receiving aperture radius is WG, then nondimensional receiving aperture Fresnel can be defined
Number ΩG=2L/ (kWG 2) relative size of receiving aperture described." the modern atmosphere published according to Science Press in 2012
Optics ", finite aperture of the spherical wave after atmospheric turbulance transmits receives arrival angle fluctuation variance and can be write as
In formula,Indicate the Refractive-index-structure parameter at z=ξ L.According to SPIE publishing house in 2010
" the Numerical Simulation of Optical Wave Propagation with Examples in published
MATLAB ", spherical wave atmospheric coherence length can be write as
It is can be found that according to formula (1) and formula (2)
In laser atmospheric turbulence transmission field, commonly using spherical wave atmospheric coherence length as the whole transmission path of description
On atmospheric turbulence intensity parameter.Similar to receiving aperture radius, it is luxuriant and rich with fragrance can equally to define a nondimensional coherence length
Alunite ear numberTo describe atmospheric turbulence intensity.Reception light signal strength of the laser after atmospheric turbulance transmits rises and falls
Normalizing variance has dependence to the path profile of Refractive-index-structure parameter.In practice, atmospheric refraction is accurately measured
The path profile of rate structural constant is a relatively difficult thing.Application No. is 201811382166.9 Chinese invention patents
It is close with arrival angle fluctuation variance measuring system to disclose a kind of light-intensity oscillation variance transmitted using two-way lightwave atmospheric turbulance
Like the method for estimation Refractive-index-structure parameter path profile, this method is with three discrete random phase screens come Approximate Equivalent
Influence of the continuous atmospheric turbulance to optical transport in transmission path, each random phase screen correspond to an effective refractive index structure
Constant is expressed asWithWithOn the basis of, it can defineWithCome
The relative size relationship of the effective refractive index structural constant of these three random phase screens is described.Actually η and η ' describes atmosphere
Substantially variation of the refractive index structure parameter along transmission path.
For the transmission of Gaussian beam atmospheric turbulance, Λ can be used0、Θ0、ΩG、Ωr0, η, η ' extraneous transmission described
Condition.Light signal strength fluctuating normalization variance measurement result is received for each, there is corresponding Λ0、Θ0、ΩG、
Ωr0, η, η ' parameter value.In other words, receiving light signal strength fluctuating normalization variance should be Λ0、Θ0、ΩG、Ωr0、η、
The function of η ', i.e. reception light signal strength rise and fall normalization variance by Λ0、Θ0、ΩG、Ωr0, η, η ' determine sextuple space
Middle value.Reception light signal strength fluctuating normalizing in the case where obtaining a large amount of different conditions (different location in corresponding sextuple space)
It, can be using these measurement results and its corresponding extraneous transmission conditions as training sample set, to instruct after changing variance measurement result
Practice multilayer feedforward neural network, approaches multilayer feedforward neural network and receive light signal strength fluctuating normalization variance and Λ0、
Θ0、ΩG、Ωr0, η, the functional relation between η '." the artificial neural network principle that China Machine Press published in 2003
And simulation example " multilayer feedforward neural network was discussed in detail.It is concentrated in aforementioned training sample, each sample is one corresponding
Input vector [Λ0,Θ0,ΩG,Ωr0, η, η '] and desired output reception light signal strength rise and fall normalization variance.Specific sample
The reception light signal strength fluctuating normalization variance of this corresponding desired output is obtained by measurement, corresponding input vector
[Λ0,Θ0,ΩG,Ωr0, η, η '] each component value then corresponding each extraneous transmission conditions when executing measurement operation.?
It, can be taking human as control Λ when executing measurement operation0、Θ0、ΩGValue, Ωr0, η, η ' value then determined by real atmosphere turbulent-flow conditions
Fixed, usual same transmission path is in the corresponding Ω of different time sectionsr0, η, η ' different values can be taken.
When analyzing laser atmospheric turbulence transmission characteristic, need to specify transmission range L, the wave number k of light wave, transmitting flat first
The initial radium W of Gaussian laser beam on face0, Gaussian laser beam on the plane of departure primary wave before radius of curvature F0, receiver hole
The radius W of diameterG, Refractive-index-structure parameter along transmission path profileIf it is not known that the atmosphere in transmission path
Refractive index structure parameter profilePrecise forms horizontal transport path usually assume thatFor constant,
Or rule of thumb assume a profile form.According toIt can calculateWithSpecific method is to solve
Following equation group:
Summary of the invention
The laser atmospheric turbulence Transmission characteristics method based on machine learning that the object of the present invention is to provide a kind of,
It is risen and fallen according to existing reception light signal strength and normalizes variance experimental measurements, obtain other external worlds for not carrying out measurement
Reception light signal strength under transmission conditions, which rises and falls, normalizes variance.
The technical solution of this method is achieved in that a kind of laser atmospheric turbulence transmission characteristic based on machine learning point
Analysis method, it is characterised in that: the Gaussian laser beam half on the plane of departure is arranged in laser atmospheric turbulence transmission path selected first
Gaussian laser beam wave-front curvature radius and receiving aperture radius parameter on diameter, the plane of departure, hold in different time period respectively
Row receives light signal strength fluctuating and normalizes variance experiment measurement operation, and each external world executed when experiment measurement operates of record passes
Corresponding vector [the Λ of defeated condition0,Θ0,ΩG,Ωr0, η, η '] value and the obtained reception light signal strength fluctuating normalizing of measurement
Change varianceThey are added to training sample as a sample to concentrate, wherein Λ0Indicate Fresnel ratio, Θ0Indicate curvature
Parameter, ΩGIndicate receiving aperture Fresnel number, Ωr0Indicate coherence length Fresnel number, WithRespectively indicate the continuous atmospheric turbulance on effective transmission path
The effective refractive index structural constant of three random phase screens of the influence to optical transport；Return executing to receive light signal strength and rise and fall
When one change variance experiment measurement operation, other than changing the period, also needs to change the gauss laser beam radius on the plane of departure, sends out
The Gaussian laser beam wave-front curvature radius and receiving aperture radius parameter in plane are penetrated, to obtain under different extraneous transmission conditions
Reception light signal strength rise and fall normalization variance measurement result, each extraneous transmission conditions and in its lower reception obtained
Light signal strength rises and falls normalization variance measurement result as a sample and is added to training sample concentration.It is concentrated with training sample
All sample training multilayer feedforward neural networks, make multilayer feedforward neural network approach receive light signal strength rise and fall normalization
Variance and Λ0、Θ0、ΩG、Ωr0, η, the functional relation between η '；When analyzing laser atmospheric turbulence transmission characteristic, according to dividing
The Gaussian laser beam wave on gauss laser beam radius, the plane of departure on the plane of departure of the laser atmospheric turbulence Transmission system of analysis
Preceding radius of curvature and receiving aperture radius parameter, and the Refractive-index-structure parameter path profile considered, calculate outgoing vector
[Λ0,Θ0,ΩG,Ωr0, η, η '] each component value, vector [Λ0,Θ0,ΩG,Ωr0, η, η '] as before multilayer to
The reception light signal strength fluctuating normalization side obtained is exactly analyzed in the input of neural network, the output of multilayer feedforward neural network
Difference.
As shown in Figure 1, the light-intensity oscillation variance and arrival angle fluctuation variance measuring system of the transmission of two-way lightwave atmospheric turbulance
Rising and falling with the reception light signal strength of laser atmospheric turbulence transmission, it is placed side by side to normalize variance measuring system.Two-way lightwave atmosphere
The light-intensity oscillation variance and arrival angle fluctuation variance measuring system of turbulence transfer include receiving and transmitting terminals A (101) and receiving and transmitting terminals B
(102).It includes Laser emission terminal that the reception light signal strength of laser atmospheric turbulence transmission, which rises and falls and normalizes variance measuring system,
(201), optics receiving subsystem (202), photodetector (203) and computer (204).Laser emission terminal (201) transmitting
Laser enter after atmospheric turbulence channels the receiving aperture for reaching optics receiving subsystem (202), optics receiving subsystem (202)
The optical signal received is incident on photodetector (203), and the signal of photodetector (203) output is transmitted to computer
(204) on data acquisition card.It is surveyed according to the light-intensity oscillation variance of two-way lightwave atmospheric turbulance transmission and arrival angle fluctuation variance
The measurement data that amount system obtains, can calculate influence of the continuous atmospheric turbulance to optical transport on effective transmission path
Three random phase screens effective refractive index structural constant, and then obtain the value of extraneous transmission conditions parameter η and η '.Laser is big
The reception light signal strength fluctuating normalization variance measuring system of gas turbulence transfer is returned for measuring to receive light signal strength and rise and fall
One changes variance.
The first part of this method, which executes, to be received light signal strength and rises and falls normalization variance measurement operation, create before multilayer to
The training sample set of neural network, the specific steps are as follows:
Step101: the atmospheric turbulance transmission path PATH1 that a selected length is L, Laser emission terminal (201) are located at
The end A of path P ATH1, optics receiving subsystem (202) are located at the end B of path P ATH1；Laser emission terminal (201) transmitting
The light wave wave number of Gaussian laser beam is k；Along the direction perpendicular to path P ATH1 path P ATH1 translation distance Ds, path is obtained
PATH2, receiving and transmitting terminals A (101) are located at the end A of path P ATH2, and receiving and transmitting terminals B (102) is located at the end B of path P ATH2；Enable instruction
Practicing sample set is sky；
Step102: n is enabled1=0, n2=0, n3=0；It performs the following operations:
Step102-1: x is enabled1=-b1+n1δ1, x2=b2+n2δ2, x3=-b3+n3δ3, b1For a positive real number, b2It is one
Real number less than 1, b3For a positive real number, δ1=2b1/N1, δ2=2 (1-b2)/N2, δ3=2b3/N3；It calculatesΘ0
=x2,Calculate W0=[2L/ (k Λ0)]1/2, F0=L/ (1- Θ0), WG=[2L/ (k ΩG)]1/2；Laser emission end
The radius of the Gaussian laser beam of machine (201) transmitting is adjusted to W0, the wave of the Gaussian laser beam of Laser emission terminal (201) transmitting
Preceding radius of curvature is adjusted to F0, the radius in the circular reception aperture of optics receiving subsystem (202) is adjusted to WG；
Step102-2: in period DUR, at interval of H hours, step Step102-2-1 to step Step102- is executed
The operation of 2-3:
Step102-2-1: the reception light signal strength for transmitting laser atmospheric turbulence, which rises and falls, normalizes variance measuring system
It works normally；The light-intensity oscillation variance and the normal work of arrival angle fluctuation variance measuring system for transmitting two-way lightwave atmospheric turbulance
Make；
Step102-2-2: computer (204) is by data acquisition card with fsThe frequency of hertz is to photodetector (203)
The signal A001 of output is sampled, and the sampled value A002 of NUM signal A001 is obtained, and calculates adopting for this NUM signal A001
The normalization variance of sample value A002At the same time, the light-intensity oscillation variance transmitted using two-way lightwave atmospheric turbulance and arrival
Angle fluctuating variance measuring system combination atmospheric turbulance refractive index structure parameter path profile approximate measure method, measure for etc.
Imitate the effective refractive index structural constant of three random phase screens of influence of the continuous atmospheric turbulance in transmission path to optical transportWithThe slave receiving and transmitting terminals A (101) that record measurement obtains passes to the angle of arrival of the light wave of receiving and transmitting terminals B (102)
Fluctuating varianceIt calculates
Step102-2-3: one new samples A003 of creation, enabling the corresponding input vector of new samples A003 is [Λ0,Θ0,
ΩG,Ωr0, η, η '], enable the reception light signal strength of the corresponding desired output of new samples A003 rise and fall normalization variance as normalizing
Change varianceNew samples A003 is added to training sample to concentrate；
Step102-3: if n1Less than N1, then n is enabled1=n1+ 1 and Step102-1 is gone to step, otherwise enables n1=0 and turn step
Rapid Step102-4；
Step102-4: if n2Less than N2, then n is enabled2=n2+ 1 and Step102-1 is gone to step, otherwise enables n2=0 and turn step
Rapid Step102-5；
Step102-5: if n3Less than N3, then n is enabled3=n3+ 1 and Step102-1 is gone to step, otherwise gone to step
Step103；
Step103: training sample set creation operation terminates；
The second part of this method is trained before multilayer Godwards using the training sample set that the first part of this method creates
Through network, approaches multilayer feedforward neural network and receive light signal strength fluctuating normalization variance and Λ0、Θ0、ΩG、Ωr0、η、
Functional relation between η ', the specific steps are as follows:
Step201: one multilayer feedforward neural network of creation, there are six input knots for the input layer of multilayer feedforward neural network
Point, the output layer of multilayer feedforward neural network have an output node；
Step202: the corresponding input vector [Λ of each sample that training sample is concentrated0,Θ0,ΩG,Ωr0, η, η ']
The value of six components respectively as six input nodes of multilayer feedforward neural network input, the corresponding expectation of each sample
Output receives desired output of the light signal strength fluctuating normalization variance as multilayer feedforward neural network, before realizing to multilayer
Training to neural network approaches multilayer feedforward neural network and receives light signal strength fluctuating normalization variance and Λ0、Θ0、
ΩG、Ωr0, η, the functional relation between η '.
The Part III of this method obtains the specific external world using the multilayer feedforward neural network of the second part of this method
Reception light signal strength under transmission conditions, which rises and falls, normalizes variance, the specific steps are as follows:
Step301: specified transmission range L, the radius W of light wave wave number k, Gaussian laser beam on the plane of departure0, transmitting it is flat
The wave-front curvature radius F of Gaussian laser beam on face0, receiving aperture radius WG, calculate Λ0=2L/ (kW0 2)、Θ0=1-L/
F0、ΩG=2L/ (kWG 2)；Profile of the specified Refractive-index-structure parameter along transmission pathZ is indicated from transmission path
On Laser emission end distance, calculate spherical wave atmospheric coherence length:
It calculatesIt calculates
Wherein
It calculatesWith
Step302: the Λ that step Step301 is obtained0、Θ0、ΩG、Ωr0, η, η ' be combined into input vector [Λ0,Θ0,
ΩG,Ωr0, η, η '] and the obtained input of multilayer feedforward neural network of second part as this method, multilayer BP Neural Network
The output of network is exactly corresponding reception light signal strength fluctuating normalization variance.
The good effect of this method is, can relatively easily be risen and fallen normalization side according to existing reception light signal strength
Level difference measurements obtain the reception light signal strength fluctuating normalization side under the extraneous transmission conditions that other do not carry out measurement operation
Difference.Before the multilayer for using dimensionless parameter to establish as the input of multilayer feedforward neural network, this method due to this method
It can be used for predicting that those did not carry out the reception light signal strength under the extraneous transmission conditions of experiment measurement to neural network
Rise and fall normalization variance；For example, the multilayer feedforward neural network that this method is established can be used for predicting that those were not carried out in fact
Reception light signal strength under the conditions of the transmission range of test amount, which rises and falls, normalizes variance.
Detailed description of the invention
Fig. 1 is the light-intensity oscillation variance and arrival angle fluctuation variance measuring system and laser of two-way lightwave atmospheric turbulance transmission
The reception light signal strength of atmospheric turbulance transmission, which rises and falls, normalizes variance measuring system schematic diagram placed side by side.
Specific embodiment
In order to which the feature and advantage of this method are more clearly understood, this method is made into one combined with specific embodiments below
The description of step.In the present embodiment, it is measured by method disclosed in the Chinese invention patent application No. is 201811382166.9
The effective refractive index knot of three random phase screens of influence of the continuous atmospheric turbulance to optical transport on effective transmission path
Structure constantWithApplication No. is 201811382166.9 Chinese invention patent, to provide a kind of two-way lightwave big
The light-intensity oscillation variance and arrival angle fluctuation variance measuring system of gas turbulence transfer, can measure the arrival angle fluctuation side of light wave
Difference.The wavelength X of the Gaussian laser beam of Laser emission terminal (201) transmitting is 800 nanometers, correspondingly wave number k=2 π/λ of light wave.
Photodetector (203) selects PIN photoelectric detector, and distance Ds is equal to 3 meters, b1=3, b2=0, b3=3, N1=7, N2=5, N3
=7, L=2 km, period DUR value are 5 days, and H is equal to 1, fsIt is equal to 10000 equal to 100, NUM, multilayer BP Neural Network
Network includes 2 hidden layers.Path P ATH1 is parallel and not far from one another with path P ATH2, is accordingly used in effective transmission path P ATH1
On influence of the continuous atmospheric turbulance to optical transport three random phase screens effective refractive index structural constantThree of influence with the continuous atmospheric turbulance on effective transmission path P ATH2 to optical transport are random
The effective refractive index structural constant of phase screenIt can regard as approximately uniform.
The technical solution of this method is achieved in that a kind of laser atmospheric turbulence transmission characteristic based on machine learning point
Analysis method, it is characterised in that: the Gaussian laser beam half on the plane of departure is arranged in laser atmospheric turbulence transmission path selected first
Gaussian laser beam wave-front curvature radius and receiving aperture radius parameter on diameter, the plane of departure, hold in different time period respectively
Row receives light signal strength fluctuating and normalizes variance experiment measurement operation, and each external world executed when experiment measurement operates of record passes
Corresponding vector [the Λ of defeated condition0,Θ0,ΩG,Ωr0, η, η '] value and the obtained reception light signal strength fluctuating normalizing of measurement
Change varianceThey are added to training sample as a sample to concentrate, wherein Λ0Indicate Fresnel ratio, Θ0Indicate curvature
Parameter, ΩGIndicate receiving aperture Fresnel number, Ωr0Indicate coherence length Fresnel number, WithRespectively indicate the continuous atmospheric turbulance on effective transmission path
The effective refractive index structural constant of three random phase screens of the influence to optical transport；Return executing to receive light signal strength and rise and fall
When one change variance experiment measurement operation, other than changing the period, also needs to change the gauss laser beam radius on the plane of departure, sends out
The Gaussian laser beam wave-front curvature radius and receiving aperture radius parameter in plane are penetrated, to obtain under different extraneous transmission conditions
Reception light signal strength rise and fall normalization variance measurement result, each extraneous transmission conditions and in its lower reception obtained
Light signal strength rises and falls normalization variance measurement result as a sample and is added to training sample concentration.It is concentrated with training sample
All sample training multilayer feedforward neural networks, make multilayer feedforward neural network approach receive light signal strength rise and fall normalization
Variance and Λ0、Θ0、ΩG、Ωr0, η, the functional relation between η '；When analyzing laser atmospheric turbulence transmission characteristic, according to dividing
The Gaussian laser beam wave on gauss laser beam radius, the plane of departure on the plane of departure of the laser atmospheric turbulence Transmission system of analysis
Preceding radius of curvature and receiving aperture radius parameter, and the Refractive-index-structure parameter path profile considered, calculate outgoing vector
[Λ0,Θ0,ΩG,Ωr0, η, η '] each component value, vector [Λ0,Θ0,ΩG,Ωr0, η, η '] as before multilayer to
The reception light signal strength fluctuating normalization side obtained is exactly analyzed in the input of neural network, the output of multilayer feedforward neural network
Difference.
As shown in Figure 1, the light-intensity oscillation variance and arrival angle fluctuation variance measuring system of the transmission of two-way lightwave atmospheric turbulance
Rising and falling with the reception light signal strength of laser atmospheric turbulence transmission, it is placed side by side to normalize variance measuring system.Two-way lightwave atmosphere
The light-intensity oscillation variance and arrival angle fluctuation variance measuring system of turbulence transfer include receiving and transmitting terminals A (101) and receiving and transmitting terminals B
(102).It includes Laser emission terminal that the reception light signal strength of laser atmospheric turbulence transmission, which rises and falls and normalizes variance measuring system,
(201), optics receiving subsystem (202), photodetector (203) and computer (204).Laser emission terminal (201) transmitting
Laser enter after atmospheric turbulence channels the receiving aperture for reaching optics receiving subsystem (202), optics receiving subsystem (202)
The optical signal received is incident on photodetector (203), and the signal of photodetector (203) output is transmitted to computer
(204) on data acquisition card.It is surveyed according to the light-intensity oscillation variance of two-way lightwave atmospheric turbulance transmission and arrival angle fluctuation variance
The measurement data that amount system obtains, can calculate influence of the continuous atmospheric turbulance to optical transport on effective transmission path
Three random phase screens effective refractive index structural constant, and then obtain the value of extraneous transmission conditions parameter η and η '.Laser is big
The reception light signal strength fluctuating normalization variance measuring system of gas turbulence transfer is returned for measuring to receive light signal strength and rise and fall
One changes variance.
The first part of this method, which executes, to be received light signal strength and rises and falls normalization variance measurement operation, create before multilayer to
The training sample set of neural network, the specific steps are as follows:
Step101: the atmospheric turbulance transmission path PATH1 that a selected length is L, Laser emission terminal (201) are located at
The end A of path P ATH1, optics receiving subsystem (202) are located at the end B of path P ATH1；Laser emission terminal (201) transmitting
The light wave wave number of Gaussian laser beam is k；Along the direction perpendicular to path P ATH1 path P ATH1 translation distance Ds, path is obtained
PATH2, receiving and transmitting terminals A (101) are located at the end A of path P ATH2, and receiving and transmitting terminals B (102) is located at the end B of path P ATH2；Enable instruction
Practicing sample set is sky；
Step102: n is enabled1=0, n2=0, n3=0；It performs the following operations:
Step102-1: x is enabled1=-b1+n1δ1, x2=b2+n2δ2, x3=-b3+n3δ3, b1For a positive real number, b2It is one
Real number less than 1, b3For a positive real number, δ1=2b1/N1, δ2=2 (1-b2)/N2, δ3=2b3/N3；It calculatesΘ0
=x2,Calculate W0=[2L/ (k Λ0)]1/2, F0=L/ (1- Θ0), WG=[2L/ (k ΩG)]1/2；Laser emission end
The radius of the Gaussian laser beam of machine (201) transmitting is adjusted to W0, the wave of the Gaussian laser beam of Laser emission terminal (201) transmitting
Preceding radius of curvature is adjusted to F0, the radius in the circular reception aperture of optics receiving subsystem (202) is adjusted to WG；
Step102-2: in period DUR, at interval of H hours, step Step102-2-1 to step Step102- is executed
The operation of 2-3:
Step102-2-1: the reception light signal strength for transmitting laser atmospheric turbulence, which rises and falls, normalizes variance measuring system
It works normally；The light-intensity oscillation variance and the normal work of arrival angle fluctuation variance measuring system for transmitting two-way lightwave atmospheric turbulance
Make；
Step102-2-2: computer (204) is by data acquisition card with fsThe frequency of hertz is to photodetector (203)
The signal A001 of output is sampled, and the sampled value A002 of NUM signal A001 is obtained, and calculates adopting for this NUM signal A001
The normalization variance of sample value A002At the same time, the light-intensity oscillation variance transmitted using two-way lightwave atmospheric turbulance and arrival
Angle fluctuating variance measuring system combination atmospheric turbulance refractive index structure parameter path profile approximate measure method, measure for etc.
Imitate the effective refractive index structural constant of three random phase screens of influence of the continuous atmospheric turbulance in transmission path to optical transportWithThe slave receiving and transmitting terminals A (101) that record measurement obtains passes to the angle of arrival of the light wave of receiving and transmitting terminals B (102)
Fluctuating varianceIt calculates
Step102-2-3: one new samples A003 of creation, enabling the corresponding input vector of new samples A003 is [Λ0,Θ0,
ΩG,Ωr0, η, η '], enable the reception light signal strength of the corresponding desired output of new samples A003 rise and fall normalization variance as normalizing
Change varianceNew samples A003 is added to training sample to concentrate；
Step102-3: if n1Less than N1, then n is enabled1=n1+ 1 and Step102-1 is gone to step, otherwise enables n1=0 and turn step
Rapid Step102-4；
Step102-4: if n2Less than N2, then n is enabled2=n2+ 1 and Step102-1 is gone to step, otherwise enables n2=0 and turn step
Rapid Step102-5；
Step102-5: if n3Less than N3, then n is enabled3=n3+ 1 and Step102-1 is gone to step, otherwise gone to step
Step103；
Step103: training sample set creation operation terminates；
The second part of this method is trained before multilayer Godwards using the training sample set that the first part of this method creates
Through network, approaches multilayer feedforward neural network and receive light signal strength fluctuating normalization variance and Λ0、Θ0、ΩG、Ωr0、η、
Functional relation between η ', the specific steps are as follows:
Step201: one multilayer feedforward neural network of creation, there are six input knots for the input layer of multilayer feedforward neural network
Point, the output layer of multilayer feedforward neural network have an output node；
Step202: the corresponding input vector [Λ of each sample that training sample is concentrated0,Θ0,ΩG,Ωr0, η, η ']
The value of six components respectively as six input nodes of multilayer feedforward neural network input, the corresponding expectation of each sample
Output receives desired output of the light signal strength fluctuating normalization variance as multilayer feedforward neural network, before realizing to multilayer
Training to neural network approaches multilayer feedforward neural network and receives light signal strength fluctuating normalization variance and Λ0、Θ0、
ΩG、Ωr0, η, the functional relation between η '.
The Part III of this method obtains the specific external world using the multilayer feedforward neural network of the second part of this method
Reception light signal strength under transmission conditions, which rises and falls, normalizes variance, the specific steps are as follows:
Step301: specified transmission range L, the radius W of light wave wave number k, Gaussian laser beam on the plane of departure0, transmitting it is flat
The wave-front curvature radius F of Gaussian laser beam on face0, receiving aperture radius WG, calculate Λ0=2L/ (kW0 2)、Θ0=1-L/
F0、ΩG=2L/ (kWG 2)；Profile of the specified Refractive-index-structure parameter along transmission pathZ is indicated from transmission path
On Laser emission end distance, calculate spherical wave atmospheric coherence length:
It calculatesIt calculatesWhereinIt calculatesWith
Step302: the Λ that step Step301 is obtained0、Θ0、ΩG、Ωr0, η, η ' be combined into input vector [Λ0,Θ0,
ΩG,Ωr0, η, η '] and the obtained input of multilayer feedforward neural network of second part as this method, multilayer BP Neural Network
The output of network is exactly corresponding reception light signal strength fluctuating normalization variance.
Claims (1)
1. a kind of laser atmospheric turbulence Transmission characteristics method based on machine learning, it is characterised in that: laser selected first
It is bent that the gauss laser beam radius on the plane of departure, the Gaussian laser beam wavefront on the plane of departure is arranged in atmospheric turbulance transmission path
It is real to execute reception light signal strength fluctuating normalization variance in different time period respectively for rate radius and receiving aperture radius parameter
The operation of test amount records the corresponding vector [Λ of extraneous transmission conditions when executing experiment measurement operation every time0,Θ0,ΩG,Ωr0,
η, η '] value and the obtained reception light signal strength of measurement rise and fall normalization varianceIt is added using them as a sample
It is concentrated to training sample, wherein Λ0Indicate Fresnel ratio, Θ0Indicate curvature parameters, ΩGIndicate receiving aperture Fresnel number, Ωr0
Indicate coherence length Fresnel number, WithIt respectively indicates for equivalent biography
The effective refractive index structural constant of three random phase screens of influence of the continuous atmospheric turbulance to optical transport on defeated path；It is holding
When row receives light signal strength fluctuating normalization variance experiment measurement operation, other than changing the period, also need change transmitting flat
The Gaussian laser beam wave-front curvature radius and receiving aperture radius parameter on gauss laser beam radius, the plane of departure on face, from
And the reception light signal strength fluctuating normalization variance measurement result under different extraneous transmission conditions is obtained, each external world is passed
Defeated condition and the reception light signal strength obtained under it rise and fall normalization variance measurement result as a sample and are added to instruction
Practice in sample set；All sample training multilayer feedforward neural networks concentrated with training sample, force multilayer feedforward neural network
Proximity receives light signal strength fluctuating normalization variance and Λ0、Θ0、ΩG、Ωr0, η, the functional relation between η '；In analysis laser
When atmospheric turbulance transmission characteristic, according to the Gaussian laser beam on the plane of departure of the laser atmospheric turbulence Transmission system to be analyzed half
Gaussian laser beam wave-front curvature radius and receiving aperture radius parameter on diameter, the plane of departure, and the air index considered
Structural constant path profile calculates outgoing vector [Λ0,Θ0,ΩG,Ωr0, η, η '] each component value, vector [Λ0,
Θ0,ΩG,Ωr0, η, η '] and input as multilayer feedforward neural network, the output of multilayer feedforward neural network is exactly to analyze
Reception light signal strength out, which rises and falls, normalizes variance；
The light-intensity oscillation variance and arrival angle fluctuation variance measuring system and laser atmospheric turbulence of two-way lightwave atmospheric turbulance transmission
The reception light signal strength fluctuating normalization variance measuring system of transmission is placed side by side；The light intensity of two-way lightwave atmospheric turbulance transmission
Fluctuating variance and arrival angle fluctuation variance measuring system include receiving and transmitting terminals A and receiving and transmitting terminals B；Laser atmospheric turbulence transmission connects
Receiving light signal strength fluctuating normalization variance measuring system includes Laser emission terminal, optics receiving subsystem, photodetector
And computer；The laser of Laser emission terminal transmitting enters the receiver hole of arrival optics receiving subsystem after atmospheric turbulence channels
Diameter, the optical signal that optics receiving subsystem receives are incident on photodetector, and the signal of photodetector output is transmitted to
On the data acquisition card of computer；It is surveyed according to the light-intensity oscillation variance of two-way lightwave atmospheric turbulance transmission and arrival angle fluctuation variance
The measurement data that amount system obtains, can calculate influence of the continuous atmospheric turbulance to optical transport on effective transmission path
Three random phase screens effective refractive index structural constant, and then obtain the value of extraneous transmission conditions parameter η and η '；Laser is big
The reception light signal strength fluctuating normalization variance measuring system of gas turbulence transfer is returned for measuring to receive light signal strength and rise and fall
One changes variance；
First part's execution reception light signal strength fluctuating normalization variance measurement of this method operates, to nerve before creation multilayer
The training sample set of network, the specific steps are as follows:
Step101: the atmospheric turbulance transmission path PATH1 that a selected length is L, Laser emission terminal are located at path P ATH1
The end A, optics receiving subsystem is located at the end B of path P ATH1；The light wave wave number of the Gaussian laser beam of Laser emission terminal transmitting
For k；Along the direction perpendicular to path P ATH1 path P ATH1 translation distance Ds, path P ATH2 is obtained, receiving and transmitting terminals A is located at
The end A of path P ATH2, receiving and transmitting terminals B are located at the end B of path P ATH2；Enable training sample set for sky；
Step102: n is enabled1=0, n2=0, n3=0；It performs the following operations:
Step102-1: x is enabled1=-b1+n1δ1, x2=b2+n2δ2, x3=-b3+n3δ3, b1For a positive real number, b2For one less than 1
Real number, b3For a positive real number, δ1=2b1/N1, δ2=2 (1-b2)/N2, δ3=2b3/N3；It calculatesΘ0=x2,Calculate W0=[2L/ (k Λ0)]1/2, F0=L/ (1- Θ0), WG=[2L/ (k ΩG)]1/2；Laser emission terminal is sent out
The radius for the Gaussian laser beam penetrated is adjusted to W0, the wave-front curvature radius of the Gaussian laser beam of Laser emission terminal transmitting is adjusted
For F0, the radius in the circular reception aperture of optics receiving subsystem is adjusted to WG；
Step102-2: in period DUR, at interval of H hours, step Step102-2-1 to step Step102-2-3 is executed
Operation:
Step102-2-1: the reception light signal strength fluctuating normalization variance measuring system for transmitting laser atmospheric turbulence is normal
Work；The light-intensity oscillation variance and arrival angle fluctuation variance measuring system for transmitting two-way lightwave atmospheric turbulance work normally；
Step102-2-2: computer is by data acquisition card with fsHertz frequency signal A001 that photodetector is exported into
Row sampling, obtains the sampled value A002 of NUM signal A001, calculates the normalization of the sampled value A002 of this NUM signal A001
VarianceAt the same time, the light-intensity oscillation variance and arrival angle fluctuation variance transmitted using two-way lightwave atmospheric turbulance, which is measured, is
System combines atmospheric turbulance refractive index structure parameter path profile approximate measure method, measures the company on effective transmission path
The effective refractive index structural constant of three random phase screens of continuous influence of the atmospheric turbulance to optical transportWithNote
The slave receiving and transmitting terminals A that record measurement obtains passes to the arrival angle fluctuation variance of the light wave of receiving and transmitting terminals BIt calculates
Step102-2-3: one new samples A003 of creation, enabling the corresponding input vector of new samples A003 is [Λ0,Θ0,ΩG,
Ωr0, η, η '], enable the reception light signal strength of the corresponding desired output of new samples A003 rise and fall normalization variance as normalization side
DifferenceNew samples A003 is added to training sample to concentrate；
Step102-3: if n1Less than N1, then n is enabled1=n1+ 1 and Step102-1 is gone to step, otherwise enables n1It=0 and goes to step
Step102-4；
Step102-4: if n2Less than N2, then n is enabled2=n2+ 1 and Step102-1 is gone to step, otherwise enables n2It=0 and goes to step
Step102-5；
Step102-5: if n3Less than N3, then n is enabled3=n3+ 1 and Step102-1 is gone to step, otherwise goes to step Step103；
Step103: training sample set creation operation terminates；
The second part of this method trains multilayer BP Neural Network using the training sample set that the first part of this method creates
Network approaches multilayer feedforward neural network and receives light signal strength fluctuating normalization variance and Λ0、Θ0、ΩG、Ωr0, η, η ' it
Between functional relation, the specific steps are as follows:
Step201: one multilayer feedforward neural network of creation, input node that there are six the input layers of multilayer feedforward neural network,
The output layer of multilayer feedforward neural network has an output node；
Step202: the corresponding input vector [Λ of each sample that training sample is concentrated0,Θ0,ΩG,Ωr0, η, η '] six
The value of component respectively as six input nodes of multilayer feedforward neural network input, the corresponding desired output of each sample
The light signal strength that receives rise and fall desired output of the normalization variance as multilayer feedforward neural network, realization to before multilayer Godwards
Training through network approaches multilayer feedforward neural network and receives light signal strength fluctuating normalization variance and Λ0、Θ0、ΩG、
Ωr0, η, the functional relation between η '；
The Part III of this method obtains specific extraneous transmission using the multilayer feedforward neural network of the second part of this method
Under the conditions of reception light signal strength rise and fall normalization variance, the specific steps are as follows:
Step301: specified transmission range L, the radius W of light wave wave number k, Gaussian laser beam on the plane of departure0, on the plane of departure
Gaussian laser beam wave-front curvature radius F0, receiving aperture radius WG, calculate Λ0=2L/ (kW0 2)、Θ0=1-L/F0、ΩG
=2L/ (kWG 2)；Profile C of the specified Refractive-index-structure parameter along transmission pathn 2(z), z is indicated from swashing in transmission path
The distance of light emitting end calculates spherical wave atmospheric coherence length:
It calculatesIt calculates
WhereinIt calculatesWith
Step302: the Λ that step Step301 is obtained0、Θ0、ΩG、Ωr0, η, η ' be combined into input vector [Λ0,Θ0,ΩG,
Ωr0, η, η '] and the obtained input of multilayer feedforward neural network of second part as this method, multilayer feedforward neural network
Output is exactly corresponding reception light signal strength fluctuating normalization variance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811499782.2A CN109635419B (en) | 2018-12-10 | 2018-12-10 | Laser atmospheric turbulence transmission characteristic analysis method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811499782.2A CN109635419B (en) | 2018-12-10 | 2018-12-10 | Laser atmospheric turbulence transmission characteristic analysis method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109635419A true CN109635419A (en) | 2019-04-16 |
| CN109635419B CN109635419B (en) | 2021-02-23 |
Family
ID=66072234
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811499782.2A Active CN109635419B (en) | 2018-12-10 | 2018-12-10 | Laser atmospheric turbulence transmission characteristic analysis method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109635419B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110083977A (en) * | 2019-05-14 | 2019-08-02 | 南京大学 | Atmospheric turbulence monitoring method based on deep learning |
| CN111666691A (en) * | 2020-06-11 | 2020-09-15 | 中国人民解放军32027部队 | Statistical method for atmospheric optical turbulence parameters |
| CN112948352A (en) * | 2021-02-04 | 2021-06-11 | 中国科学院合肥物质科学研究院 | Method for constructing atmospheric optical turbulence space-time characteristics and probabilistic database |
| CN113686817A (en) * | 2021-08-24 | 2021-11-23 | 桂林电子科技大学 | Non-uniform path atmospheric phase screen distribution method in marine aerial environment |
| CN113985566A (en) * | 2021-09-10 | 2022-01-28 | 西南科技大学 | Scattered light focusing method based on spatial light modulation and neural network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5796105A (en) * | 1995-09-08 | 1998-08-18 | Wang; Ting-I | Extended range optical scintillometer with interruption protection for measuring atmospheric refractive turbulence |
| CN108052956A (en) * | 2017-11-07 | 2018-05-18 | 西安理工大学 | Wireless light communication subcarrier modulation constellation recognition methods under a kind of atmospheric turbulance |
| CN108900245A (en) * | 2018-06-26 | 2018-11-27 | 中国地质大学（武汉） | The emulation mode and system of Bessel-Gaussian beam transmission characteristic in turbulent atmosphere |
-
2018
- 2018-12-10 CN CN201811499782.2A patent/CN109635419B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5796105A (en) * | 1995-09-08 | 1998-08-18 | Wang; Ting-I | Extended range optical scintillometer with interruption protection for measuring atmospheric refractive turbulence |
| CN108052956A (en) * | 2017-11-07 | 2018-05-18 | 西安理工大学 | Wireless light communication subcarrier modulation constellation recognition methods under a kind of atmospheric turbulance |
| CN108900245A (en) * | 2018-06-26 | 2018-11-27 | 中国地质大学（武汉） | The emulation mode and system of Bessel-Gaussian beam transmission characteristic in turbulent atmosphere |
Non-Patent Citations (3)
| Title |
|---|
| 周舟: "基于改进的神经网络的自适应波前重构算法研究", 《中国优秀硕士学位论文全文数据库-信息科技辑》 * |
| 娄岩，姜会林，陈纯毅，佟首峰: "激光大气湍流传输光强起伏及光斑面积实验分析", 《红外与激光工程》 * |
| 娄岩，陈纯毅，赵义武，陶宗慧: "高斯涡旋光束在大气湍流传输中的特性研究", 《中国光学》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110083977A (en) * | 2019-05-14 | 2019-08-02 | 南京大学 | Atmospheric turbulence monitoring method based on deep learning |
| CN110083977B (en) * | 2019-05-14 | 2023-06-06 | 南京大学 | Atmospheric turbulence monitoring method based on deep learning |
| CN111666691A (en) * | 2020-06-11 | 2020-09-15 | 中国人民解放军32027部队 | Statistical method for atmospheric optical turbulence parameters |
| CN111666691B (en) * | 2020-06-11 | 2022-10-21 | 中国人民解放军32027部队 | Statistical method for atmospheric optical turbulence parameters |
| CN112948352A (en) * | 2021-02-04 | 2021-06-11 | 中国科学院合肥物质科学研究院 | Method for constructing atmospheric optical turbulence space-time characteristics and probabilistic database |
| CN113686817A (en) * | 2021-08-24 | 2021-11-23 | 桂林电子科技大学 | Non-uniform path atmospheric phase screen distribution method in marine aerial environment |
| CN113686817B (en) * | 2021-08-24 | 2022-10-28 | 桂林电子科技大学 | Non-uniform path atmospheric phase screen distribution method in marine aerial environment |
| CN113985566A (en) * | 2021-09-10 | 2022-01-28 | 西南科技大学 | Scattered light focusing method based on spatial light modulation and neural network |
| CN113985566B (en) * | 2021-09-10 | 2023-09-12 | 西南科技大学 | Scattered light focusing method based on spatial light modulation and neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109635419B (en) | 2021-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109635419A (en) | Laser atmospheric turbulence Transmission characteristics method based on machine learning | |
| CN102986150B (en) | multiple-acquisition OTDR method and device | |
| CN105606499B (en) | Suspended particulate matter mass concentration real-time detection device, and measuring method | |
| AU2020102396A4 (en) | Machine learning-based method for analyzing characteristics of laser beam propagation through turbulent atmosphere | |
| CN101980056B (en) | Method and device for determining atmospheric turbulence parameter based on M<2> factor and light scintillation index | |
| CN102636270B (en) | Optical measurement instrument and method for atmospheric coherent length | |
| CN113568055A (en) | Aviation transient electromagnetic data retrieval method based on LSTM network | |
| CN105424653A (en) | Fruit pulp tissue optical property detection system and method adopting integrated fiber-optics probe | |
| CN105651373B (en) | A method of based on measurement two o'clock once per revolution vibration in polarized light time domain reflection technology | |
| CN105928850A (en) | Lasso regression method for inverting particle size distribution by virtue of light scattering method | |
| CN113189014A (en) | Ozone concentration estimation method fusing satellite remote sensing and ground monitoring data | |
| CN102546008A (en) | Experimental device for estimating bit error rate of laser space communication system based on atmospheric parameters | |
| CN102788678B (en) | Oceanic turbulence parameter non-contact measuring method and system | |
| CN103148872B (en) | A kind of optical fiber distributed perturbation sensor localization method | |
| David et al. | Cross-sectional measuring of optical beam | |
| Eichinger et al. | The application of a scanning, water Raman-lidar as a probe of the atmospheric boundary layer | |
| Wang et al. | A novel indoor ranging method using weighted altofrequent RSSI measurements | |
| CN103063869B (en) | Measuring device and measuring method of light propagation path transverse average wind speed and wind direction | |
| CN103472033B (en) | A kind of method utilizing CW Laser commercial measurement spherical particle spectrum complex refractive index | |
| CN105092523A (en) | Scheme based on near infrared technology and used for detecting phosphorous in water | |
| CN202734971U (en) | Optical measuring instrument for measuring atmospheric coherent length | |
| TWI484353B (en) | Methods and Systems for Calculating Random Errors | |
| Lin et al. | PM2. 5 measuring method using RBF neural network combined with PSO algorithm | |
| CN117516617A (en) | Measurement uncertainty assessment method and system based on digital simulation | |
| Lin et al. | Enhanced Raman Distributed Temperature Sensor Based on Self-Constructed Fully Connected Neural Network |
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 |