CN110795892B - Channel simulation method and device based on generation countermeasure network - Google Patents
Channel simulation method and device based on generation countermeasure network Download PDFInfo
- Publication number
- CN110795892B CN110795892B CN201911009738.3A CN201911009738A CN110795892B CN 110795892 B CN110795892 B CN 110795892B CN 201911009738 A CN201911009738 A CN 201911009738A CN 110795892 B CN110795892 B CN 110795892B
- Authority
- CN
- China
- Prior art keywords
- network
- phase difference
- difference coefficient
- countermeasure
- loss
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- 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
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The embodiment of the invention provides a channel simulation method and device based on a generation countermeasure network, wherein a first phase difference coefficient is obtained by inputting noise data of a specified type into a generation countermeasure network model obtained by pre-training, wherein the generation countermeasure network model is obtained by training based on a plurality of noise data with specified types and a second phase difference coefficient; determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen; and determining an atmospheric turbulence simulation result according to the phase screen. The distribution of the first phase difference coefficient obtained by generating the countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase screen corresponding to the first phase difference coefficient and the actual light beam phase value is reduced, the accuracy of atmospheric turbulence simulation is improved, and the influence of atmospheric turbulence on the light beam transmission quality is reduced.
Description
Technical Field
The invention relates to the technical field of adaptive optics, in particular to a channel simulation method and device based on a generative countermeasure network.
Background
As the beam passes through the atmosphere, changes in refractive index caused by atmospheric turbulence can cause wavefront phase distortion of the beam, which can degrade the transmission quality of the beam. The adaptive optics system can compensate wave front phase distortion caused by atmospheric turbulence, and simulating the atmospheric turbulence is an important part for designing and debugging the adaptive optics system. Therefore, the influence of the atmospheric turbulence on the transmission quality of the light beam can be reduced to some extent by simulating the atmospheric turbulence.
Since the effect of the atmospheric turbulence on the transmission quality of the light beam is mainly reflected in phase, the simulation process of the atmospheric turbulence can be regarded as the generation process of the phase screen. Atmospheric turbulence can now be simulated by the following method: obtaining an atmospheric turbulence refractive index power spectrum according to a physical model, and obtaining the phase screen through the atmospheric turbulence refractive index power spectrum.
However, because the assumption of the physical model is ideal, there is still some gap between the phase value of the actual beam and the phase screen obtained by the method, which results in low accuracy of the phase screen, and thus cannot play a role in reducing the influence of atmospheric turbulence on the transmission quality of the beam.
Disclosure of Invention
The embodiment of the invention aims to provide a channel simulation method and a channel simulation device based on a generation countermeasure network, so as to reduce the influence of atmospheric turbulence on the transmission quality of a light beam. The specific technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a channel simulation method based on a generative countermeasure network, where the method includes:
inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, wherein the generated countermeasure network model is obtained through training based on a plurality of noise data of the specified type and a second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wavefront distortion data detected by a wavefront detector;
determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen;
and determining an atmospheric turbulence simulation result according to the phase screen.
Optionally, the generating the countermeasure network model includes generating a network and a countermeasure network,
the training process for generating the confrontation network model comprises the following steps:
inputting a plurality of noise data with the specified type into the generation network to obtain a third phase difference coefficient, wherein the generation network is used for representing the mapping relation between the noise data with the specified type and the third phase difference coefficient;
inputting the third phase difference coefficient into the countermeasure network to obtain a first output loss, wherein the countermeasure network is used for judging whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient;
inputting the second phase difference coefficient into the impedance network to obtain a second output loss;
calculating a countermeasure loss of the countermeasure network based on the first output loss and the second output loss;
adjusting the network parameters of the countermeasure network according to the countermeasure loss to obtain an updated countermeasure network;
inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network;
and adjusting the network parameters of the generated network according to the generation loss to obtain an updated generated network.
Optionally, the generation network is composed of three fully-connected layers, and the countermeasure network is composed of two fully-connected layers.
Optionally, the adjusting the network parameters of the countermeasure network according to the countermeasure loss includes:
adjusting network parameters of the antagonistic network by using a back propagation algorithm according to the antagonistic loss;
the adjusting the network parameters of the generated network according to the generation loss includes:
and adjusting the network parameters of the generated network by using a back propagation algorithm according to the generation loss.
Optionally, the first phase difference coefficient is a zernike polynomial coefficient.
In a second aspect, an embodiment of the present invention provides a channel simulation apparatus based on a generation countermeasure network, where the apparatus includes:
the device comprises a first phase difference coefficient acquisition module, a second phase difference coefficient acquisition module and a third phase difference coefficient acquisition module, wherein the first phase difference coefficient acquisition module is used for inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, the generated countermeasure network model is obtained through training based on a plurality of noise data of the specified type and the second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wave front distortion data detected by a wave front detector;
the phase screen determining module is used for determining the phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and the preset corresponding relation between the first phase difference coefficient and the phase screen;
and the atmospheric turbulence simulation result determining module is used for determining an atmospheric turbulence simulation result according to the phase screen.
Optionally, the generating the countermeasure network model includes generating a network and a countermeasure network,
the device further comprises a training module;
the training module is used for inputting a plurality of noise data with the specified type into the generating network to obtain a third phase difference coefficient, wherein the generating network is used for representing the mapping relation between the noise data with the specified type and the third phase difference coefficient; inputting the third phase difference coefficient into the countermeasure network to obtain a first output loss, wherein the countermeasure network is used for judging whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient; inputting the second phase difference coefficient into the impedance network to obtain a second output loss; calculating a countermeasure loss of the countermeasure network based on the first output loss and the second output loss; adjusting the network parameters of the countermeasure network according to the countermeasure loss to obtain an updated countermeasure network; inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network; and adjusting the network parameters of the generated network according to the generation loss to obtain an updated generated network.
Optionally, the generation network is composed of three fully-connected layers, and the countermeasure network is composed of two fully-connected layers.
Optionally, when the training module performs the adjustment of the network parameters of the countermeasure network according to the countermeasure loss, the training module is specifically configured to adjust the network parameters of the countermeasure network by using a back propagation algorithm according to the countermeasure loss;
the training module is specifically configured to adjust the network parameters of the generated network by using a back propagation algorithm according to the generation loss when the training module adjusts the network parameters of the generated network according to the generation loss.
Optionally, the first phase difference coefficient is a zernike polynomial coefficient.
In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor and the communication interface complete communication between the memory and the processor through the communication bus;
a memory for storing a computer program;
a processor, configured to implement the channel simulation method based on the generative countermeasure network according to any one of the first aspect described above when executing the computer program stored in the memory.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when executed by a processor, the computer program implements the channel simulation method based on generation of a countermeasure network according to any one of the first aspect.
The embodiment of the invention provides a channel simulation method and a device based on a generation countermeasure network, wherein the method comprises the following steps: inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, wherein the generated countermeasure network model is obtained through training based on a plurality of noise data of the specified type and a second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wavefront distortion data detected by a wavefront detector; determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen; and determining an atmospheric turbulence simulation result according to the phase screen. Since the generation countermeasure network model is trained based on a plurality of noise data with the specified type and the second phase difference coefficient, and the second phase difference coefficient is obtained by fitting a plurality of wavefront distortion data detected by the wavefront detector, the distribution of the first phase difference coefficient obtained by inputting the noise data with the specified type into the generation countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase value of the actual light beam and the phase value of the phase screen corresponding to the first phase difference coefficient is reduced, the accuracy of atmospheric turbulence simulation is improved, and the influence of atmospheric turbulence on the transmission quality of the light beam is reduced.
Of course, not all of the advantages described above need to be achieved at the same time in the practice of any one product or method of the invention.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic flowchart of a first embodiment of a channel simulation method based on a generation countermeasure network according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of a training method for generating a countermeasure network according to an embodiment of the present invention;
fig. 3 is a schematic flowchart of a second embodiment of a channel simulation method based on a generative countermeasure network according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a channel simulation apparatus based on a generative countermeasure network according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to reduce the influence of the atmospheric turbulence on the transmission quality of the optical beam, embodiments of the present invention provide a channel simulation method and apparatus based on a generation countermeasure Network (GAN), which are described in detail below.
Fig. 1 is a schematic flowchart of a first embodiment of a channel simulation method based on a generation countermeasure network according to an embodiment of the present invention, and as shown in fig. 1, the method according to the embodiment of the present invention may include:
s101, inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient.
The channel simulation method based on the generation countermeasure network provided by the embodiment of the invention is applied to the server.
The generated countermeasure network model is obtained based on a plurality of noise data with specified types and a second phase difference coefficient training, and the second phase difference coefficient is obtained by fitting a plurality of wavefront distortion data detected by a wavefront detector.
The noise data of the above-mentioned specified type may be data conforming to a gaussian distribution, and the noise data may be data conforming to a gaussian distribution directly acquired or data conforming to a gaussian distribution obtained after data processing.
Since the generation countermeasure network model is trained based on the second phase difference coefficient obtained by fitting a plurality of wavefront distortion data detected by the wavefront sensor and the noise data having the designated type, the distribution of the first phase difference coefficient obtained by inputting the designated type of noise data into the generation countermeasure network model is close to the distribution of the second phase difference coefficient.
S102, determining the phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and the preset corresponding relation between the first phase difference coefficient and the phase screen.
The preset corresponding relation between the first phase difference coefficient and the phase screen is that the first phase difference coefficient and the phase screen have a one-to-one corresponding relation, and the phase screen corresponding to the first phase difference coefficient can be obtained according to the one-to-one corresponding relation. The one-to-one correspondence relationship may be preset, or may be a preset algorithm, a formula, or the like, so that the phase screen corresponding to the first phase difference coefficient is obtained by calculation using the preset algorithm, the formula, or the like.
And S103, determining an atmospheric turbulence simulation result according to the phase screen.
Since the effect of the atmospheric turbulence on the transmission quality of the light beam is mainly reflected in phase, the simulation process of the atmospheric turbulence can be regarded as the generation process of the phase screen. The resulting phase screen can therefore be used as a simulation result of atmospheric turbulence.
According to the channel simulation method based on the generation countermeasure network provided by the embodiment of the invention, the noise data of the designated type is input into the generation countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, wherein the generation countermeasure network model is obtained based on a plurality of noise data with the designated type and a second phase difference coefficient obtained through training, and the second phase difference coefficient is obtained by fitting a plurality of wavefront distortion data detected by a wavefront detector; determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen; and determining an atmospheric turbulence simulation result according to the phase screen. Because the generation countermeasure network model is obtained by training based on a plurality of noise data with specified types and the second phase difference coefficient, and the second phase difference coefficient is obtained by fitting a plurality of wave front distortion data detected by the wave front detector, the distribution of the first phase difference coefficient obtained by inputting the noise data with the specified types into the generation countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase value of the actual light beam and the phase value of the phase screen corresponding to the first phase difference coefficient is reduced, the accuracy of atmospheric turbulence simulation is improved, and the influence of atmospheric turbulence on the transmission quality of the light beam is reduced.
Fig. 2 is a schematic flow chart of a training method for generating a countermeasure network according to an embodiment of the present invention, where generating a countermeasure network model includes generating a network and a countermeasure network, where the generation network is composed of three fully-connected layers, the countermeasure network is composed of two fully-connected layers, and a phase difference coefficient is described by taking a Zernike polynomial coefficient as an example. As shown in fig. 2, a method of an embodiment of the invention may include:
s201, a plurality of noise data with specified types are input into a generating network to obtain a third phase difference coefficient.
Wherein the generation network may be used to characterize a mapping relationship between the specified type of noise data and the third phase difference coefficient.
In a specific implementation, 100-dimensional noise data z conforming to a gaussian distribution can be input into a generation network composed of three fully-connected layers. The full-junction layer can be obtained by calculating the following formula (1), wherein the formula (1) is specifically as follows:
out＝W·in+b (1)
where out is the output of the fully-connected layer, in is the input of the fully-connected layer, b is the bias, and W is the parameter matrix. The generation network is composed of three fully-connected layers, and the number of neurons in the hidden layer of each of the three fully-connected layers can be: 2048, 1024, 400, resulting in a vector of 400 dimensions, i.e. resulting in 400 th order Zernike polynomial coefficients (third phase difference coefficients).
S202, inputting the third phase difference coefficient into the countermeasure network to obtain a first output loss.
The countermeasure network may be configured to determine whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient. In specific implementation, if the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient, the countermeasure network outputs '1'; if the distribution of the third phase difference coefficient is not close to the distribution of the second phase difference coefficient, the countermeasure network outputs "0".
Specifically, the countermeasure network is composed of two fully-connected layers, and the number of neurons in the hidden layer of the two fully-connected layers may be: 256,1. In a specific implementation, the 400 th order Zernike polynomial coefficient obtained in step S201 is input into the countermeasure network, so as to obtain a first output loss.
And S203, inputting the second phase difference coefficient into the countermeasure network to obtain a second output loss.
And inputting Zernike polynomial coefficients (second phase difference coefficients) obtained by fitting a plurality of wavefront distortion data detected by the wavefront detector into the impedance network to obtain second output loss.
S204, calculating the countermeasure loss of the countermeasure network according to the first output loss and the second output loss.
Specifically, the countermeasure loss of the countermeasure network can be calculated by the following equation (2), where equation (2) is specifically as follows:
LD＝L1-L2 (2)
where LD is the countermeasure loss of the countermeasure network, L1 is the first output loss obtained in step S202, and L2 is the second output loss obtained in step S203.
S205, according to the countermeasure loss, network parameters of the countermeasure network are adjusted to obtain the updated countermeasure network.
As one implementation, the updated countermeasure network can be obtained by calculating the gradient of the countermeasure network according to the countermeasure loss and adjusting the network parameters of the countermeasure network by using a back propagation algorithm.
And S206, inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network.
And S207, adjusting the network parameters of the generated network according to the generation loss to obtain the updated generated network.
The specific implementation of step S206 may refer to step S202, and the specific implementation of step S207 may refer to step S205, which is not described herein again.
The above steps S201 to S207 are repeatedly and iteratively executed until the countermeasure network convergence is generated. The convergence condition for generating the countermeasure network may be that the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient, or that the number of iterations reaches a preset number, where one iteration process includes performing the above steps S201 to S207.
According to the training method for generating the countermeasure network, provided by the embodiment of the invention, the network parameters of the countermeasure network are adjusted according to the countermeasure loss, and the network parameters of the generation network are adjusted according to the generation loss, so that the distribution of the third phase difference coefficient obtained by training the generation countermeasure network is close to the distribution of the second phase difference coefficient obtained by fitting a plurality of wavefront distortion data detected by a wavefront detector, and the trained generation countermeasure network can output the phase difference coefficient similar to the distribution of the phase difference coefficient corresponding to the real wavefront distortion data, and the purpose of simulation is achieved.
A specific implementation manner is taken as an example to describe the channel simulation method based on the generation countermeasure network according to the embodiment of the present invention, and fig. 3 is a schematic flow diagram of a second embodiment of the channel simulation method based on the generation countermeasure network according to the embodiment of the present invention, as shown in fig. 3, the method according to the embodiment of the present invention may include:
s301, wavefront distortion data are collected and fitted to obtain a corresponding first phase difference coefficient.
In an actual atmosphere turbulence environment, 5000 groups of wavefront distortion data influenced by the atmosphere turbulence can be collected by using a wavefront detector. Then, 5000 groups of acquired wavefront distortion data can be fitted by using Zernike polynomials to obtain corresponding Zernike polynomial coefficients. The fitting method of the Zernike polynomials may be the Gram-Schmidt orthogonal algorithm. Here, both the phase detection technology of the wavefront detector and the fitting method of the Zernike polynomial are mature prior arts, and as for the method for converting wavefront distortion data into Zernike polynomial coefficients, the embodiment of the present invention is not specifically limited, as long as it is ensured that the Zernike polynomial coefficients corresponding to the wavefront distortion data in the actual turbulent flow environment can be obtained.
The Zernike polynomial may be in the form of equation (3), where equation (3) is specifically as follows:
wherein j is the order of a Zernike polynomial; theta and r are polar coordinates; n and m are natural numbers related to the order of the Zernike polynomials; zevenj、Zoddj、zjCorresponding Zernike polynomials when different values are taken for m; and two conditions are implicit in equation (3): (n-m) is an even number, and n is greater than or equal to m; s represents a natural number. n and m may occur in combination and follow a particular orderThe method comprises the following steps: firstly, n is determined to be zero, and then m meeting the condition is found in sequence from small to large. Then n is determined to be 1, and then m meeting the condition is found in sequence from small to large, and so on. For example, if n is 0, then m is 0; if n is 1, there is no m satisfying the condition; if n is 2, m may be 0 and 2, and if n is 3, m may be 1 and 3.
For each combination of n and m satisfying the condition, the calculation method of the corresponding Zernike polynomials can be found by using equation (3), and the Zernike polynomials are labeled in the above order. In particular, when n and m are equal, two Zernike polynomials are corresponded, wherein a polynomial using the homodromous component cos (m θ) is labeled as odd order and a polynomial using the orthogonal component sin (m θ) is labeled as even order.
And S302, training to generate a countermeasure network.
The generation of the countermeasure network is divided into two parts: a generation network and a countermeasure network. The generating network is used for generating Zernike polynomial coefficients similar to real data from noise data of a specified type through a multilayer network, wherein the real data are a plurality of wavefront distortion data detected by a wavefront detector; the countermeasure network may be a classification network composed of a multi-layer network structure for discriminating whether the generated Zernike polynomial coefficients are similar to the real data. The generation network and the countermeasure network can be trained alternately, the generation network generates a phase distribution similar to the real data as much as possible, and the countermeasure network distinguishes the generation data and the real data as much as possible. The generation network and the countermeasure network fight each other, and finally the countermeasure network stops because the data generated by the generation network cannot be distinguished from the true and false, so that the generation network can generate Zernike polynomial coefficients almost consistent with the real data distribution.
Because the more the number of network layers, the stronger the description capability of the model, and the better the generation effect, the embodiment of the present invention does not limit the network structure for generating the countermeasure network.
And S303, generating a phase screen by using the trained generation countermeasure network.
Specifically, any one of the phase modes in the unit circle can be approximated by a linear combination of Zernike polynomials. The phase screen can be obtained by calculating according to formula (4), wherein formula (4) is as follows:
wherein the content of the first and second substances,
representing a phase screen; m is the maximum order of a preset Zernike polynomial; a isiCoefficients of a Zernike polynomial of order i; zi(r, θ) represents an ith order Zernike polynomial.
Therefore, by applying the embodiment of the invention, because the generation countermeasure network model is obtained by training based on a plurality of noise data with specified types and the second phase difference coefficient, and the second phase difference coefficient is obtained by fitting real data, wherein the real data is a plurality of wavefront distortion data detected by the wavefront detector, the distribution of the first phase difference coefficient obtained by inputting the noise data with the specified types into the generation countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase value of the actual light beam and the phase value of the phase screen corresponding to the first phase difference coefficient is reduced, the accuracy of atmospheric turbulence simulation is improved, and the influence of atmospheric turbulence on the light beam transmission quality is reduced.
Corresponding to the above method embodiment, fig. 4 is a schematic structural diagram of a channel simulation apparatus based on a generation countermeasure network according to an embodiment of the present invention, as shown in fig. 4, the channel simulation apparatus may include:
a first phase difference coefficient obtaining module 410, configured to input noise data of an assigned type into a generated countermeasure network model obtained through pre-training, so as to obtain a first phase difference coefficient, where the generated countermeasure network model is obtained through training based on a plurality of noise data of the assigned type and a second phase difference coefficient, and the second phase difference coefficient is obtained by fitting a plurality of wavefront distortion data detected by a wavefront detector;
a phase screen determining module 420, configured to determine a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relationship between the first phase difference coefficient and the phase screen;
and an atmospheric turbulence simulation result determining module 430, configured to determine an atmospheric turbulence simulation result according to the phase screen.
Optionally, the generating the countermeasure network model includes generating a network and a countermeasure network,
at this time, the device may further include a training module;
the training module is used for inputting a plurality of noise data with specified types into a generating network to obtain a third phase difference coefficient, wherein the generating network is used for representing the mapping relation between the noise data with the specified types and the third phase difference coefficient; inputting the third phase difference coefficient into a countermeasure network to obtain a first output loss, wherein the countermeasure network is used for judging whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient; inputting the second phase difference coefficient into the countermeasure network to obtain a second output loss; calculating the countermeasure loss of the countermeasure network according to the first output loss and the second output loss; adjusting network parameters of the countermeasure network according to the countermeasure loss to obtain an updated countermeasure network; inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network; and adjusting the network parameters of the generated network according to the generation loss to obtain the updated generated network.
Optionally, the generation network is formed by three full-connection layers, and the countermeasure network is formed by two full-connection layers.
Optionally, when the training module performs the adjustment of the network parameters of the countermeasure network according to the countermeasure loss, the training module is specifically configured to adjust the network parameters of the countermeasure network by using a back propagation algorithm according to the countermeasure loss;
the training module is specifically configured to adjust the network parameters of the generated network by using a back propagation algorithm according to the generation loss when the training module adjusts the network parameters of the generated network according to the generation loss.
Optionally, the first phase difference coefficient is a zernike polynomial coefficient.
According to the channel simulation device based on the generation countermeasure network provided by the embodiment of the invention, through the first phase difference coefficient acquisition module, noise data of an appointed type are input into a generation countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, wherein the generation countermeasure network model is obtained through training based on a plurality of noise data with appointed types and a second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wavefront distortion data detected by a wavefront detector; determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen through a phase screen determining module; and determining an atmospheric turbulence simulation result through an atmospheric turbulence simulation result determining module according to the phase screen. Because the generation countermeasure network model is obtained by training based on a plurality of noise data with specified types and the second phase difference coefficient, and the second phase difference coefficient is obtained by fitting a plurality of wave front distortion data detected by the wave front detector, the distribution of the first phase difference coefficient obtained by inputting the noise data with the specified types into the generation countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase value of the actual light beam and the phase value of the phase screen corresponding to the first phase difference coefficient is reduced, the accuracy of atmospheric turbulence simulation is improved, and the influence of atmospheric turbulence on the transmission quality of the light beam is reduced.
An embodiment of the present invention further provides an electronic device, as shown in fig. 5, which includes a processor 501, a communication interface 502, a memory 503 and a communication bus 504, where the processor 501, the communication interface 502 and the memory 503 complete mutual communication through the communication bus 504,
a memory 503 for storing a computer program;
the processor 501, when executing the program stored in the memory 503, implements the following steps:
inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, wherein the generated countermeasure network model is obtained through training based on a plurality of noise data of the specified type and a second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wavefront distortion data detected by a wavefront detector;
determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen;
and determining an atmospheric turbulence simulation result according to the phase screen.
Optionally, generating the countermeasure network model includes generating a network and a countermeasure network, and the processor 501 may further specifically implement training to generate the countermeasure network model:
inputting a plurality of noise data with specified types into a generation network to obtain a third phase difference coefficient, wherein the generation network is used for representing the mapping relation between the noise data with the specified types and the third phase difference coefficient;
inputting the third phase difference coefficient into a countermeasure network to obtain a first output loss, wherein the countermeasure network is used for judging whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient;
inputting the second phase difference coefficient into the countermeasure network to obtain a second output loss;
calculating the countermeasure loss of the countermeasure network according to the first output loss and the second output loss;
adjusting network parameters of the countermeasure network according to the countermeasure loss to obtain an updated countermeasure network;
inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network;
and adjusting the network parameters of the generated network according to the generation loss to obtain the updated generated network.
Optionally, the generation network is formed by three full connection layers, and the countermeasure network is formed by two full connection layers.
Optionally, when the processor 501 executes the step of adjusting the network parameter of the countermeasure network according to the countermeasure loss, the following steps may be specifically implemented: adjusting network parameters of the countermeasure network by using a back propagation algorithm according to the countermeasure loss;
when the processor 501 executes the step of adjusting the network parameter of the generated network according to the generation loss, the following steps may be specifically implemented: and adjusting the network parameters of the generated network by using a back propagation algorithm according to the generation loss.
Optionally, the first phase difference coefficient is a zernike polynomial coefficient.
It can be seen that, with the embodiment of the present invention, since the generation countermeasure network model is trained based on a plurality of noise data with a specified type and the second phase difference coefficient, and the second phase difference coefficient is obtained by fitting a plurality of wavefront distortion data detected by the wavefront detector, the distribution of the first phase difference coefficient obtained by inputting the noise data with the specified type into the generation countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase value of the actual light beam and the phase value of the phase screen corresponding to the first phase difference coefficient becomes small, thereby improving the accuracy of atmospheric turbulence simulation, which helps to reduce the influence of atmospheric turbulence on the light beam transmission quality.
The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.
The communication interface is used for communication between the electronic equipment and other equipment.
The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.
The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but also Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components.
In yet another embodiment of the present invention, a computer-readable storage medium is further provided, in which a computer program is stored, and the computer program is executed by a processor to implement any of the above steps of the channel simulation method based on generation of a countermeasure network.
It can be seen that, with the embodiment of the present invention, since the generation countermeasure network model is trained based on a plurality of noise data with a specified type and the second phase difference coefficient, and the second phase difference coefficient is obtained by fitting a plurality of wavefront distortion data detected by the wavefront detector, the distribution of the first phase difference coefficient obtained by inputting the noise data with the specified type into the generation countermeasure network model is close to the distribution of the second phase difference coefficient, so that the difference between the phase value of the actual light beam and the phase value of the phase screen corresponding to the first phase difference coefficient becomes small, thereby improving the accuracy of atmospheric turbulence simulation, which helps to reduce the influence of atmospheric turbulence on the light beam transmission quality.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the apparatus embodiment, the electronic device embodiment and the computer-readable storage medium embodiment, since they are substantially similar to the method embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (6)
1. A channel simulation method based on a generative countermeasure network, the method comprising:
inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, wherein the generated countermeasure network model is obtained through training based on a plurality of noise data of the specified type and a second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wavefront distortion data detected by a wavefront detector;
determining a phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and a preset corresponding relation between the first phase difference coefficient and the phase screen;
determining an atmospheric turbulence simulation result according to the phase screen;
the generating a countering network model includes generating a network and a countering network,
the training process for generating the confrontation network model comprises the following steps:
inputting a plurality of noise data with the specified type into the generation network to obtain a third phase difference coefficient, wherein the generation network is used for representing the mapping relation between the noise data with the specified type and the third phase difference coefficient;
inputting the third phase difference coefficient into the countermeasure network to obtain a first output loss, wherein the countermeasure network is used for judging whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient;
inputting the second phase difference coefficient into the impedance network to obtain a second output loss;
calculating a countermeasure loss of the countermeasure network based on the first output loss and the second output loss;
adjusting the network parameters of the countermeasure network according to the countermeasure loss to obtain an updated countermeasure network;
inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network;
adjusting the network parameters of the generated network according to the generation loss to obtain an updated generated network;
the first phase difference coefficient is a Zernike polynomial coefficient.
2. The method of claim 1, wherein the generation network is comprised of three fully-connected layers and the countermeasure network is comprised of two fully-connected layers.
3. The method of claim 1,
the adjusting network parameters of the countermeasure network according to the countermeasure loss includes:
adjusting network parameters of the antagonistic network by using a back propagation algorithm according to the antagonistic loss;
the adjusting the network parameters of the generated network according to the generation loss includes:
and adjusting the network parameters of the generated network by using a back propagation algorithm according to the generation loss.
4. A channel simulation apparatus based on a generative countermeasure network, the apparatus comprising:
the device comprises a first phase difference coefficient acquisition module, a second phase difference coefficient acquisition module and a third phase difference coefficient acquisition module, wherein the first phase difference coefficient acquisition module is used for inputting noise data of a specified type into a generated countermeasure network model obtained through pre-training to obtain a first phase difference coefficient, the generated countermeasure network model is obtained through training based on a plurality of noise data of the specified type and the second phase difference coefficient, and the second phase difference coefficient is obtained through fitting a plurality of wave front distortion data detected by a wave front detector;
the phase screen determining module is used for determining the phase screen corresponding to the first phase difference coefficient according to the first phase difference coefficient and the preset corresponding relation between the first phase difference coefficient and the phase screen;
the atmospheric turbulence simulation result determining module is used for determining an atmospheric turbulence simulation result according to the phase screen;
the generating a countering network model includes generating a network and a countering network,
the device further comprises a training module;
the training module is used for inputting a plurality of noise data with the specified type into the generating network to obtain a third phase difference coefficient, wherein the generating network is used for representing the mapping relation between the noise data with the specified type and the third phase difference coefficient; inputting the third phase difference coefficient into the countermeasure network to obtain a first output loss, wherein the countermeasure network is used for judging whether the distribution of the third phase difference coefficient is close to the distribution of the second phase difference coefficient; inputting the second phase difference coefficient into the impedance network to obtain a second output loss; calculating a countermeasure loss of the countermeasure network based on the first output loss and the second output loss; adjusting the network parameters of the countermeasure network according to the countermeasure loss to obtain an updated countermeasure network; inputting the third phase difference coefficient into the updated countermeasure network to obtain the generation loss of the generation network; adjusting the network parameters of the generated network according to the generation loss to obtain an updated generated network;
the first phase difference coefficient is a Zernike polynomial coefficient.
5. The apparatus of claim 4, wherein the generation network is comprised of three fully-connected layers and the countermeasure network is comprised of two fully-connected layers.
6. The apparatus of claim 4,
the training module is specifically configured to adjust the network parameters of the countermeasure network by using a back propagation algorithm according to the countermeasure loss when the training module performs the adjustment of the network parameters of the countermeasure network according to the countermeasure loss;
the training module is specifically configured to adjust the network parameters of the generated network by using a back propagation algorithm according to the generation loss when the training module adjusts the network parameters of the generated network according to the generation loss.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911009738.3A CN110795892B (en) | 2019-10-23 | 2019-10-23 | Channel simulation method and device based on generation countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911009738.3A CN110795892B (en) | 2019-10-23 | 2019-10-23 | Channel simulation method and device based on generation countermeasure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110795892A CN110795892A (en) | 2020-02-14 |
| CN110795892B true CN110795892B (en) | 2021-10-01 |
Family
ID=69441144
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911009738.3A Active CN110795892B (en) | 2019-10-23 | 2019-10-23 | Channel simulation method and device based on generation countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110795892B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1105449A (en) * | 1994-09-16 | 1995-07-19 | 北京理工大学 | Wavefront detecting and reconstituting method based on Zernike multinomial |
| CN109119093A (en) * | 2018-10-30 | 2019-01-01 | Oppo广东移动通信有限公司 | Voice de-noising method, device, storage medium and mobile terminal |
| CN109543740A (en) * | 2018-11-14 | 2019-03-29 | 哈尔滨工程大学 | A kind of object detection method based on generation confrontation network |
| CN109582562A (en) * | 2018-10-19 | 2019-04-05 | 北京航空航天大学 | Based on the intelligent software test and cloud platform construction method for generating confrontation network |
| CN110021052A (en) * | 2019-04-11 | 2019-07-16 | 北京百度网讯科技有限公司 | The method and apparatus for generating model for generating eye fundus image |
| CN110164469A (en) * | 2018-08-09 | 2019-08-23 | 腾讯科技（深圳）有限公司 | A kind of separation method and device of multi-person speech |
| CN110207835A (en) * | 2019-05-23 | 2019-09-06 | 中国科学院光电技术研究所 | A kind of wave front correction method based on out-of-focus image training |
| CN110263848A (en) * | 2019-06-18 | 2019-09-20 | 北京邮电大学 | A kind of multimode fibre imaging method and device based on production confrontation network |
| CN110289927A (en) * | 2019-07-01 | 2019-09-27 | 上海大学 | The channel simulation implementation method of confrontation network is generated based on condition |
| CN110349095A (en) * | 2019-06-14 | 2019-10-18 | 浙江大学 | Learn the adaptive optics wavefront compensation method of prediction wavefront zernike coefficient based on depth migration |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100507936C (en) * | 2007-06-19 | 2009-07-01 | 中山大学 | Zernike moment based robust hashing image authentification method |
| US20180284758A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for industrial internet of things data collection for equipment analysis in an upstream oil and gas environment |
| CN106972921B (en) * | 2017-01-03 | 2020-07-28 | 北京理工大学 | Asymmetric optical information security system based on wave-front sensing and double optical keys |
-
2019
- 2019-10-23 CN CN201911009738.3A patent/CN110795892B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1105449A (en) * | 1994-09-16 | 1995-07-19 | 北京理工大学 | Wavefront detecting and reconstituting method based on Zernike multinomial |
| CN110164469A (en) * | 2018-08-09 | 2019-08-23 | 腾讯科技（深圳）有限公司 | A kind of separation method and device of multi-person speech |
| CN109582562A (en) * | 2018-10-19 | 2019-04-05 | 北京航空航天大学 | Based on the intelligent software test and cloud platform construction method for generating confrontation network |
| CN109119093A (en) * | 2018-10-30 | 2019-01-01 | Oppo广东移动通信有限公司 | Voice de-noising method, device, storage medium and mobile terminal |
| CN109543740A (en) * | 2018-11-14 | 2019-03-29 | 哈尔滨工程大学 | A kind of object detection method based on generation confrontation network |
| CN110021052A (en) * | 2019-04-11 | 2019-07-16 | 北京百度网讯科技有限公司 | The method and apparatus for generating model for generating eye fundus image |
| CN110207835A (en) * | 2019-05-23 | 2019-09-06 | 中国科学院光电技术研究所 | A kind of wave front correction method based on out-of-focus image training |
| CN110349095A (en) * | 2019-06-14 | 2019-10-18 | 浙江大学 | Learn the adaptive optics wavefront compensation method of prediction wavefront zernike coefficient based on depth migration |
| CN110263848A (en) * | 2019-06-18 | 2019-09-20 | 北京邮电大学 | A kind of multimode fibre imaging method and device based on production confrontation network |
| CN110289927A (en) * | 2019-07-01 | 2019-09-27 | 上海大学 | The channel simulation implementation method of confrontation network is generated based on condition |
Non-Patent Citations (3)
| Title |
|---|
| An improvement on the CNN-based OAM Demodulator via Conditional Generative Adversarial Networks;zhe li等;《2019 18th International Conference on Optical Communications and Networks》;20190831;第1-3页 * |
| 可见光序列图像盲复原及质量评价方法研究;吴静;《中国博士学位论文全文数据库》;20190615(第6期);第I138-25页 * |
| 基于生成对抗网络的人脸灰度图上色;李炬;《电脑知识与技术》;20180430;第14卷(第11期);第179-181页，第191页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110795892A (en) | 2020-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110098875B (en) | Adaptive equalization method, apparatus, electronic device and medium in optical fiber communication system | |
| Seo et al. | Analysis of normalized point source sensitivity as a performance metric for large telescopes | |
| Yura et al. | Digital simulation of an arbitrary stationary stochastic process by spectral representation | |
| Barrios et al. | Exponentiated Weibull fading model for free-space optical links with partially coherent beams under aperture averaging | |
| CN115438571B (en) | Ground wave radar wave field calculation method and device based on machine learning | |
| CN114239698A (en) | Data processing method, device and equipment | |
| CN116229036A (en) | Data acquisition system, method, computer device and storage medium | |
| Croaker et al. | Strongly singular and hypersingular integrals for aeroacoustic incident fields | |
| Gao et al. | Dixon resultant theory for stability analysis of distributed delay systems and enhancement of delay robustness | |
| CN110795892B (en) | Channel simulation method and device based on generation countermeasure network | |
| Tan et al. | Solving ordinary differential equations using neural networks | |
| CN110111284A (en) | A kind of method, apparatus, electronic equipment and the storage medium of vignetting image rectification | |
| CN117273109A (en) | Quantum neuron-based hybrid neural network construction method and device | |
| US20230188321A1 (en) | Method for training model based on homomorphic encryption, device, and storage medium | |
| Yang | Simulink‐based teaching design for multipath‐effect elimination utilizing convolution analysis and pade approximation | |
| Floc'h et al. | Explicit SABR calibration through simple expansions | |
| Ma et al. | Sparse normalized least mean absolute deviation algorithm based on unbiasedness criterion for system identification with noisy input | |
| Zhang et al. | Enclosing ellipsoid‐based reachable set estimation for discrete‐time singular systems | |
| CN115859048A (en) | Noise processing method and device for partial discharge signal | |
| Fiori | Blind deconvolution by a newton method on the non‐unitary hypersphere | |
| Bourlier | Low-grazing angle propagation and scattering by an object above a highly conducting rough sea surface in a ducting environment from an accelerated MoM | |
| Wiegand et al. | A series representation for multidimensional Rayleigh distributions | |
| CN110852451A (en) | Recursive kernel self-adaptive filtering method based on kernel function | |
| Song et al. | Algorithm based on the linear canonical transform for QFM signal parameters estimation | |
| Badjate et al. | Multi step ahead prediction of north and south hemisphere sun spots chaotic time series using focused time lagged recurrent neural network model |
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 |