CN111610514B - Inversion method and device for propagation characteristics of evaporation waveguide - Google Patents
Inversion method and device for propagation characteristics of evaporation waveguide Download PDFInfo
- Publication number
- CN111610514B CN111610514B CN202010429593.9A CN202010429593A CN111610514B CN 111610514 B CN111610514 B CN 111610514B CN 202010429593 A CN202010429593 A CN 202010429593A CN 111610514 B CN111610514 B CN 111610514B
- Authority
- CN
- China
- Prior art keywords
- evaporation waveguide
- evaporation
- wave propagation
- waveguide
- height
- 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
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
- G01S13/06—Systems determining position data of a target
- G01S13/08—Systems for measuring distance only
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/882—Radar or analogous systems specially adapted for specific applications for altimeters
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/414—Discriminating targets with respect to background clutter
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Electromagnetism (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The application discloses an inversion method and device for propagation characteristics of an evaporation waveguide, wherein the height of the evaporation waveguide acquired by radar is acquired, and an electric wave propagation factor is calculated according to radar parameters of the radar and the height of the evaporation waveguide. And inputting the electric wave propagation factor into a preset space distribution prediction network of the evaporation waveguide to obtain the refractive index profile of the evaporation waveguide. Compared with the prior art, in the embodiment, the first mapping relation between the atmospheric environmental factor and the evaporation waveguide propagation characteristic is represented by the electric wave propagation factor, and the spatial distribution prediction network of the evaporation waveguide is used for representing the second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, so that the interference of the atmospheric environmental factor on the inversion process of the evaporation waveguide propagation characteristic can be avoided. In addition, the spatial distribution prediction network can convert the nonlinear irreparable of the second mapping relation in the low latitude space into the linear divisible of the high-dimensional space, so that the inversion process of the propagation characteristic of the evaporation waveguide can be efficient and accurate.
Description
Technical Field
The present application relates to the field of wireless communication technologies, and in particular, to an inversion method and apparatus for propagation characteristics of an evaporation waveguide.
Background
Tropospheric waveguides have a large impact on wireless communications as well as on radar systems, for example, reducing the communication performance of the radio system, causing radar detection failures, and the like. The evaporation waveguide is used as a troposphere waveguide with high occurrence probability on the sea surface, and great influence is brought to the offshore microwave communication. Therefore, the real-time detection of the offshore evaporation waveguide is of great significance.
Currently, the inversion of evaporative waveguides (RFC) using radar sea clutter is the current research focus. However, the troposphere atmospheric environment belongs to a coherent channel, and the random characteristic of the troposphere atmospheric environment is extremely complex. The existing method for acquiring the atmospheric environmental factors is difficult to satisfy the requirement of quickly acquiring the specific numerical values of all the atmospheric environmental factors on the propagation path of the evaporation waveguide in real time, so that the prediction effect of the RFC inversion method is poor, namely the accuracy of the refractive index profile (used for expressing the propagation characteristics of the evaporation waveguide, including the vertical distribution, the horizontal uniform distribution or the horizontal non-uniform distribution) of the evaporation waveguide output by the RFC inversion method is not high.
Disclosure of Invention
The application provides an inversion method and device for propagation characteristics of an evaporation waveguide, and aims to provide a method capable of accurately inverting the propagation characteristics of the evaporation waveguide.
In order to achieve the above object, the present application provides the following technical solutions:
a method of inversion of propagation characteristics of an evaporation waveguide, comprising:
the method comprises the steps of obtaining the height of an evaporation waveguide acquired by a radar, and calculating a radio wave propagation factor according to radar parameters of the radar and the height of the evaporation waveguide, wherein the radio wave propagation factor is used for representing a first mapping relation between atmospheric environment factors and evaporation waveguide propagation characteristics;
and inputting the electric wave propagation factor into a pre-constructed spatial distribution prediction network of the evaporation waveguide to obtain an evaporation waveguide refractive index profile of the evaporation waveguide, wherein the spatial distribution prediction network of the evaporation waveguide is used for representing a second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, and converting the nonlinear indifference of the second mapping relation in a low-latitude space into linear indifference in a high-dimensional space, and the evaporation waveguide refractive index profile is used for representing the propagation characteristic of the evaporation waveguide.
Optionally, the process for constructing the spatial distribution prediction network of the evaporation waveguide includes:
inputting a radio wave propagation factor sample in a preset experiment database as a training sample of a preset radial basis function neural network to obtain an output result;
training the radial basis function neural network based on the radio wave propagation factor sample, the evaporation waveguide refractive index profile sample corresponding to the radio wave propagation factor sample and the output result until the radial basis function neural network outputs the evaporation waveguide refractive index profile corresponding to the radio wave propagation factor sample;
and determining the trained radial basis function neural network as a spatial distribution prediction network of the evaporation waveguide.
Optionally, the preset process of the experimental database includes:
respectively modeling the vertical distribution of the heights of the evaporation waveguides and the horizontal distribution of the heights of the evaporation waveguides by taking the height samples of the evaporation waveguides obtained by collection as parameters, and taking the combination of the vertical distribution model and the horizontal distribution model of the heights of the evaporation waveguides as an evaporation waveguide refractive index model;
outputting an evaporation waveguide refractive index profile sample by using the evaporation waveguide refractive index model;
inputting the height indicated by the evaporation waveguide refractive index profile sample and a preset radar parameter sample into a preset radio wave propagation factor model to obtain a radio wave propagation factor sample;
and recording the corresponding relation between the evaporation waveguide refractive index profile sample and the electric wave propagation factor sample.
Optionally, the process of modeling the horizontal distribution of the height of the evaporation waveguide includes:
generating a Markov chain matrix by utilizing a Gauss-Markov process;
extracting characteristic values and characteristic vectors of the Markov chain matrix by using a principal component analysis method of K-L transformation to obtain the horizontal distribution model of the height of the evaporation waveguide;
wherein the Markov chain matrix is used to characterize a horizontally distributed sample of the height of the evaporation waveguide.
Optionally, the vertical distribution model of the height of the evaporation waveguide includes:
P-J model.
Optionally, the process of constructing the wave propagation factor model includes:
and establishing a calculation formula of the electric wave propagation factor by taking a parabolic equation of electric wave propagation and electric field distribution of the electric wave in the evaporation waveguide environment as parameters.
Optionally, the training process of the radial basis function neural network includes:
estimating the center and standard deviation of the radial basis function of the hidden layer in the radial basis function neural network;
estimating the weight and the threshold of an output layer in the radial basis function neural network;
estimating respective numbers of nodes of an input layer, the hidden layer, and the output layer in the radial basis function neural network.
An apparatus for inverting propagation characteristics of an evaporation waveguide, comprising:
the calculation unit is used for acquiring the height of the evaporation waveguide acquired by the radar, and calculating a radio wave propagation factor according to the radar parameter of the radar and the height of the evaporation waveguide, wherein the radio wave propagation factor is used for representing a first mapping relation between an atmospheric environment factor and the propagation characteristic of the evaporation waveguide;
the prediction unit is used for inputting the electric wave propagation factor into a pre-constructed spatial distribution prediction network of the evaporation waveguide to obtain an evaporation waveguide refractive index profile of the evaporation waveguide, the spatial distribution prediction network of the evaporation waveguide is used for representing a second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, and converting the nonlinear indifference of the second mapping relation in a low-latitude space into linear divisible of a high-dimensional space, and the evaporation waveguide refractive index profile is used for representing the propagation characteristic of the evaporation waveguide.
A computer storage medium comprising a stored program, wherein the program performs the method of inverting the propagation characteristics of an evaporation waveguide.
An electronic device, comprising: a processor, a memory, and a bus; the processor and the memory are connected through the bus;
the memory is configured to store a program and the processor is configured to execute the program, wherein the program is configured to execute the method for inverting the propagation characteristic of the evaporation waveguide when the program is executed.
According to the technical scheme, the height of the evaporation waveguide acquired by radar collection is acquired, and the radio wave propagation factor is calculated according to the radar parameters of the radar and the height of the evaporation waveguide. And inputting the electric wave propagation factor into a pre-constructed space distribution prediction network of the evaporation waveguide to obtain the evaporation waveguide refractive index profile of the evaporation waveguide. Compared with the prior art, in the embodiment, the first mapping relation between the atmospheric environmental factor and the evaporation waveguide propagation characteristic is represented by the electric wave propagation factor, so that the atmospheric environmental factor is not used as a reference basis in the process of inverting the evaporation waveguide propagation characteristic, and the spatial distribution prediction network of the evaporation waveguide is used for representing the second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, that is, the prediction result of the spatial distribution prediction network of the evaporation waveguide is not influenced by the atmospheric environmental factor, so that the interference of the atmospheric environmental factor on the evaporation waveguide propagation characteristic inversion process can be avoided. In addition, the spatial distribution prediction network of the evaporation waveguide can convert the nonlinear irreparable of the second mapping relation in the low-latitude space into the linear separable of the high-dimensional space, so that the inversion process of the propagation characteristic of the evaporation waveguide can be efficient and accurate.
Drawings
In order to more clearly illustrate the embodiments of the present application 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 application, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic diagram of an inversion method of propagation characteristics of an evaporation waveguide according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of a process for constructing a spatial distribution prediction network of an evaporation waveguide according to an embodiment of the present application;
fig. 3 is a schematic diagram of a specific implementation process for building an experiment database according to an embodiment of the present application;
FIG. 4 is a schematic diagram of an implementation of a process for modeling the horizontal distribution of the height of an evaporation waveguide according to an embodiment of the present disclosure;
FIG. 5 is a diagram illustrating a specific implementation process of a K-means clustering algorithm according to an embodiment of the present application;
fig. 6 is a schematic diagram of a specific implementation process of a steepest descent algorithm based on an error method according to an embodiment of the present application;
fig. 7 is a schematic diagram of a specific process for estimating the number of nodes of an input layer, an output layer, and a hidden layer according to an embodiment of the present disclosure;
FIG. 8 is a schematic diagram of a topology of a spatial distribution prediction network of evaporation waveguides according to an embodiment of the present application;
FIG. 9 is a diagram illustrating a radial basis function unit structure according to an embodiment of the present disclosure;
fig. 10 is a schematic diagram of a unit structure of an output layer according to an embodiment of the present disclosure;
FIG. 11 is a schematic diagram of a horizontal distribution of heights of an evaporation waveguide provided in an embodiment of the present application;
fig. 12a is a schematic diagram of an electric wave propagation factor according to an embodiment of the present application;
fig. 12b is a schematic diagram of another wave propagation factor provided in the embodiment of the present application;
fig. 13a is a schematic diagram of another wave propagation factor provided in the embodiment of the present application;
fig. 13b is a schematic diagram of another wave propagation factor provided in the embodiment of the present application;
FIG. 14 is a schematic diagram of a horizontal distribution sample of the heights of evaporation waveguides provided in an embodiment of the present application;
fig. 15 is a variation curve of the number of nodes in a hidden layer according to an embodiment of the present disclosure;
fig. 16 is a schematic structural diagram of a spatial distribution prediction network of an evaporation waveguide according to an embodiment of the present application;
FIG. 17a is a schematic diagram of a sample horizontal distribution of heights of evaporation waveguides according to an embodiment of the present application;
fig. 17b is a schematic diagram of a wave propagation factor sample according to an embodiment of the present application;
FIG. 18a is a schematic view of the height of an evaporation waveguide provided in an embodiment of the present application;
FIG. 18b is a diagram illustrating an error result according to an embodiment of the present application;
fig. 19a is a schematic diagram of another wave propagation factor provided in the embodiment of the present application;
FIG. 19b is a diagram illustrating absolute error results provided by an embodiment of the present application;
FIG. 20a is a schematic view of the height of another evaporation waveguide provided in embodiments of the present application;
FIG. 20b is a schematic diagram of another error result provided by an embodiment of the present application;
fig. 21a is a schematic diagram of another wave propagation factor provided in the embodiment of the present application;
FIG. 21b is a diagram illustrating another absolute error result provided by an embodiment of the present application;
fig. 22 is a schematic structural diagram of an inversion apparatus for evaporation waveguide propagation characteristics according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, and not all 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 application.
The process provided by the embodiment of the application can be applied to an inversion system of the propagation characteristics of the evaporation waveguide, and can be specifically executed by a computing device such as a server, so that the computed electric wave propagation factor is input into a pre-constructed spatial distribution prediction network of the evaporation waveguide to obtain a refractive index profile of the evaporation waveguide, and the propagation characteristics of the evaporation waveguide can be represented based on the refractive index profile of the evaporation waveguide. In the embodiment of the application, the prediction effect of the evaporation waveguide refractive index profile is superior to that of the evaporation waveguide refractive index profile obtained by RFC inversion model inversion.
As shown in fig. 1, a schematic diagram of an inversion method of propagation characteristics of an evaporation waveguide provided in an embodiment of the present application includes the following steps:
s101: and acquiring the height of the evaporation waveguide acquired by the radar, and calculating the electric wave propagation factor according to the radar parameters of the radar and the height of the evaporation waveguide.
Among these radar parameters are, but not limited to: the height of the transmitting antenna, the frequency of the electric wave, the main beam pointing angle of the electric wave, the beam width of the electric wave, and the polarization mode of the transmitting antenna. The process of collecting the height of the evaporation waveguide is well known to those skilled in the art and will not be described in detail here. The electric wave propagation factor is used for representing a first mapping relation between the atmospheric environment factor and the propagation characteristic of the evaporation waveguide.
It should be noted that, according to the radar parameter of the radar and the height of the evaporation waveguide, the specific implementation process of calculating the radio wave propagation factor is as follows: and inputting the radar parameters and the height of the evaporation waveguide into a pre-constructed electric wave propagation factor model to obtain the electric wave propagation factor. The procedure for constructing the wave propagation factor model can be explained with reference to S302 and S302 shown in fig. 3 below.
S102: and inputting the electric wave propagation factor into a pre-constructed space distribution prediction network of the evaporation waveguide to obtain the evaporation waveguide refractive index profile of the evaporation waveguide.
The space distribution prediction network of the evaporation waveguide is used for representing a second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, and converting the second mapping relation in the nonlinear indifference of the low-latitude space into the linear indifference of the high-dimensional space. In this embodiment, the propagation characteristics of the evaporation waveguide can be characterized based on the evaporation waveguide refractive index profile.
It should be noted that, for the construction process of the spatial distribution prediction network of the evaporation waveguide, reference may be made to the following steps shown in fig. 2 and the explanation of the steps.
In conclusion, the height of the evaporation waveguide is obtained through radar acquisition, and the electric wave propagation factor is calculated according to the radar parameters of the radar and the height of the evaporation waveguide. And inputting the electric wave propagation factor into a pre-constructed space distribution prediction network of the evaporation waveguide to obtain the evaporation waveguide refractive index profile of the evaporation waveguide. Compared with the prior art, in the embodiment, the first mapping relation between the atmospheric environmental factor and the evaporation waveguide propagation characteristic is represented by the electric wave propagation factor, so that the atmospheric environmental factor is not used as a reference basis in the process of inverting the evaporation waveguide propagation characteristic, and the spatial distribution prediction network of the evaporation waveguide is used for representing the second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, that is, the prediction result of the spatial distribution prediction network of the evaporation waveguide is not influenced by the atmospheric environmental factor, so that the interference of the atmospheric environmental factor on the evaporation waveguide propagation characteristic inversion process can be avoided. In addition, the spatial distribution prediction network of the evaporation waveguide can convert the nonlinear irreparable of the second mapping relation in the low-latitude space into the linear separable of the high-dimensional space, so that the inversion process of the propagation characteristic of the evaporation waveguide can be efficient and accurate.
As shown in fig. 2, the process of constructing the spatial distribution prediction network of the evaporation waveguide includes the following steps:
s201: and (5) constructing an experiment database.
The experimental database comprises an electric wave propagation factor sample and an evaporation waveguide refractive index profile sample corresponding to the electric wave propagation factor sample. The evaporation waveguide refractive index profile sample can be obtained by calculation according to an evaporation waveguide refractive index model. The radio wave propagation factor sample can be calculated from the radio wave propagation factor model.
As is known in the art, the radio wave propagation factor is determined by a propagation field (e.g., an electric field) in the air space, and therefore, the radio wave propagation factor is influenced by the atmospheric environmental factors such as sea surface reflection, atmospheric refraction, and atmospheric scattering. Therefore, the mapping relation between the atmospheric environment factor and the evaporation waveguide propagation characteristic can be converted into the mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, and the evaporation waveguide propagation characteristic can be predicted based on the electric wave propagation factor.
It should be noted that the evaporation waveguide refractive index profile is a concrete representation of the propagation characteristics of the evaporation waveguide, and the evaporation waveguide refractive index profile includes a vertical distribution of the height of the evaporation waveguide (i.e., a height variation of the evaporation waveguide) and a horizontal distribution of the height of the evaporation waveguide (the horizontal distribution includes a horizontal uniform distribution and a horizontal non-uniform distribution).
It should be noted that, for a specific implementation process for constructing the experiment database, reference may be made to the following steps shown in fig. 3 and an explanation of the steps.
S202: the method comprises the steps of inputting a radio wave propagation factor sample in an experimental database as a training sample of a preset Radial Basis Function (RBF) neural network to obtain an output result, training the RBF neural network based on the radio wave propagation factor sample, an evaporation waveguide refractive index profile sample corresponding to the radio wave propagation factor sample and the output result until the RBF neural network outputs the evaporation waveguide refractive index profile corresponding to the radio wave propagation factor sample, and determining the RBF neural network obtained by current training as a spatial distribution prediction network of an evaporation waveguide.
The topology of the spatial distribution prediction network of the evaporation waveguide is shown in fig. 8, and in fig. 8, an input layer F (r) is shown N ) In particular to the wave propagation factor model, the hidden layer Φ (c, σ) in particular to the radial basis function (where c denotes the center of the radial basis function and σ denotes the standard deviation of the radial basis function, all of which are well known to those skilled in the art), the output layer y (x) K ) In particular to the evaporation waveguide refractive indexAnd (4) modeling.
It should be noted that the neurons of the hidden layer are formed by radial basis functions, and each neuron corresponds to one radial basis function, and based on the characteristics of the radial basis functions (that is, each radial basis function is independent and orthogonal to each other), the relationship between the evaporation waveguide refractive index model and the electric wave propagation factor model can be converted from the nonlinear inseparability in the low-dimensional space to the linear separability in the high-dimensional space. And in view of the fact that RBF neural network learning can be applied to the aspects of fuzzy recognition, nonlinear problem approximation and the like. Therefore, the input electric wave propagation factor sample is input into the preset RBF neural network, the preset RBF neural network is trained until the output result of the preset RBF neural network is consistent with the evaporation waveguide refractive index profile sample, and then the space distribution prediction network of the evaporation waveguide can be determined.
In this embodiment, the radial basis function of the hidden layer may specifically be a gaussian function, and the unit structure of the radial basis function is shown in fig. 9. Wherein r = { r = 1 ,r 2 ,r 3 ,…r N Denotes different sampling points on the propagation path of the evaporation waveguide, N denotes the number of sampling points, c j Indicating the center, σ, of the input layer jth neuron radial basis function j The standard deviation of the radial basis function of the j-th neuron is indicated. Therefore, based on the radial basis function unit structure shown in fig. 9, the radial basis function Φ corresponding to the jth neuron can be obtained j Is shown in formula (1).
As can be seen from equation (1), taking the euclidean distance between the input vector and the center of the radial basis function as the input of the radial basis function, when the input vector deviates from the center of the radial basis function, the radial basis function decays exponentially and rapidly approaches zero. Therefore, the preset RBF neural network has the advantages of high learning speed, strong convergence and the like.
In addition, the mapping from the hidden layer to the output layer is specifically a linear relationship, and the unit structure of the output layer is shown in fig. 10. It is provided withIn, w j,k Represents the weight from the jth (1 ≦ j ≦ S) unit of the hidden layer to the kth (1 ≦ K ≦ K) unit of the output layer, b k Representing the threshold from the jth (1 ≦ j ≦ S) cell of the hidden layer to the kth (1 ≦ K ≦ K) cell of the output layer.
The radial basis function Φ shown in the formula (1) is combined with the radial basis function cell structure shown in fig. 9 and the output layer cell structure shown in fig. 10 j Regarding the weight value as the output layer, for the input sample F (r) in the preset RBF neural network, the corresponding output result y (x) k ) As shown in equation (2).
In this embodiment, the specific implementation process of training the preset RBF neural network includes, as well known in the art, the following steps: to the center c of the radial basis function j And standard deviation σ j Estimating the weight w of the output layer j,k And a threshold value b k The evaluation is performed and the respective node numbers of the input layer, the output layer and the hidden layer are evaluated.
Note that the standard deviation σ of the radial basis function j Is obtained by performing mean square error operation on the radial basis function, so that only the center c of the radial basis function needs to be obtained j The result of the estimation is that the standard deviation sigma of the radial basis function can be obtained j The estimation result of (2).
In the present embodiment, the center c of the radial basis function is measured j The specific implementation of the estimation is well known to those skilled in the art, and the center c of the radial basis function is calculated by using a K-means clustering algorithm j The estimation result of (2). It should be noted that, for a specific implementation process of the K-means clustering algorithm, reference may be made to the following steps shown in fig. 5 and an explanation of the steps.
In addition, the weight w to the output layer j,k And a threshold value b k The specific implementation of the estimation is well known to those skilled in the art, e.g. using basesAnd calculating the weight value and the threshold value of the output layer by using the steepest descent algorithm of the error method. It should be noted that, for a specific implementation process of the steepest descent algorithm based on the error method, reference may be made to the following steps shown in fig. 6 and an explanation of the steps.
The specific implementation process for estimating the number of nodes in each of the input layer, the output layer and the hidden layer is well known to those skilled in the art, and specific reference may be made to the following steps shown in fig. 7 and the explanation of the steps.
As shown in fig. 3, the specific implementation process of constructing the experiment database includes the following steps:
s301: a height sample of the evaporation waveguide was taken.
S302: and taking the height sample of the evaporation waveguide as a parameter, respectively modeling the vertical distribution of the height of the evaporation waveguide and the horizontal distribution of the height of the evaporation waveguide, and taking the combination of the vertical distribution model and the horizontal distribution model of the height of the evaporation waveguide as a refractive index model of the evaporation waveguide.
The specific implementation process for modeling the vertical distribution of the height of the evaporation waveguide is well known to those skilled in the art, and for example, a P-J model is used as a vertical distribution model of the height of the evaporation waveguide.
It is emphasized that the horizontal uniform distribution of the height of the evaporation waveguide can be regarded as a concrete expression of the horizontal non-uniform distribution of the height of the evaporation waveguide, and therefore, the present embodiment models only the horizontal non-uniform distribution of the height of the evaporation waveguide.
It should be noted that, a specific process of modeling the horizontal distribution of the heights of the evaporation waveguides to obtain a model of the horizontal distribution of the heights of the evaporation waveguides may refer to the following steps shown in fig. 4 and an explanation of the steps.
S303: and constructing a radio wave propagation factor model.
The electric wave propagation factor model is a common knowledge familiar to those skilled in the art, that is, an electric field distribution u (x, z) of an electric wave in an evaporation waveguide environment is calculated by using a parabolic equation of electric wave propagation, wherein x represents a horizontal distance of an evaporation waveguide propagation path (in the present embodiment, the value range of x is 10-40 km), and z represents a height of the evaporation waveguide, and an electric wave propagation factor F (r) at different heights at different horizontal distances is calculated according to the electric field distribution, so the electric wave propagation factor model is as shown in formula (3).
It should be noted that the calculation parameters in the parabolic equation u (x, z) of the wave propagation include, but are not limited to: radar parameters (e.g. transmitting antenna height, wave frequency, main beam pointing angle of the wave, beam width of the wave, polarization of the transmitting antenna) and evaporation waveguide refractive index profile.
Specifically, the radio wave propagation factors shown in fig. 12a, 12b, 13a, and 13b are obtained by calculating the radio wave propagation factors using the formula (3) using the radar parameters shown in table 1 below and the horizontal distribution of the heights of the evaporation waveguides shown in fig. 11 as an example.
TABLE 1
The radio wave propagation factor shown in fig. 12a is a radio wave frequency f =8GHZ and a transmitting antenna height h t And the number of the particles is 5 m.
The radio wave propagation factor shown in fig. 12b is a radio wave frequency f =8GHZ and a transmitting antenna height h t And =20m is calculated according to the formula.
The radio wave propagation factor shown in fig. 13a is a radio wave frequency f =10GHZ and a transmitting antenna height h t And the number of the particles is 5 m.
The radio wave propagation factor shown in fig. 13b is a radio wave frequency f =10GHZ and a transmitting antenna height h t And =20m is calculated according to the formula.
By comparing fig. 12a and 13a, and fig. 12b and 13b, the following conclusions are drawn:
(1) When the height of the transmitting antenna is larger than that of the evaporation waveguide, the higher the frequency of the electric wave is, the easier the electric wave is trapped, the larger the energy of the electric wave is, and the larger the electric wave propagation factor is.
(2) When the height of the transmitting antenna is the same as that of the evaporation waveguide, the possibility that the electric wave is trapped at a position close to the transmitting antenna is low (trapping may also not occur), and after the electric wave is away from the transmitting antenna by a preset distance, the electric wave propagation factor increases with the increase of x (i.e. the horizontal distance of the evaporation waveguide propagation path), and obviously, the change process of the electric wave propagation factor is as follows: first decreasing and then increasing.
(3) The electric wave is more likely to be trapped at a position far from the transmitting antenna due to the horizontal non-uniform distribution of the evaporation waveguide, and the electric wave propagation factor increases as x increases.
From the above, it can be seen that the radar parameters and the refractive index profile of the evaporation waveguide can actually affect the variation of the wave propagation factor. Therefore, the radio wave propagation factor samples in the experimental database need to identify radar parameters in addition to the corresponding evaporation waveguide refractive index profile.
S304: and outputting a sample of the refractive index profile of the evaporation waveguide by using the refractive index model of the evaporation waveguide.
S305: and inputting the height of the evaporation waveguide indicated by the evaporation waveguide refractive index profile sample and preset radar parameters into the electric wave propagation factor model to obtain an electric wave propagation factor sample.
S306: and recording the corresponding relation between the evaporation waveguide refractive index profile sample and the electric wave propagation factor sample.
Optionally, the corresponding relationship between the evaporation waveguide refractive index profile sample, the electric wave propagation factor sample and the radar parameter may also be recorded.
As shown in fig. 4, the specific implementation process for modeling the horizontal distribution of the height of the evaporation waveguide includes the following steps:
s401: and generating a Markov chain matrix by using a Gaussian-Markov process.
From the well-known general knowledgeIt is recognized that the horizontal non-uniform distribution of the height of the evaporation waveguide is caused by the characteristics of the atmospheric environment (e.g., temperature, humidity, and pressure) within a predetermined range of the location of the evaporation waveguide. Thus, the height h of the evaporation waveguide can be simulated using a Markov chain d (x k ) Random variation along the horizontal distance.
Wherein, the height h d (x k ) The random variation characteristic along the horizontal distance specifically refers to: the height of the evaporation waveguide at the current sampling position is only related to the atmospheric environment characteristic within the preset range of the current sampling position, and is not related to the heights of the evaporation waveguides at all the previous sampling positions.
It is known from the central limit theorem that random variables formed by the comprehensive influence of a plurality of mutually independent atmospheric environmental characteristics (i.e., temperature, humidity and pressure) are often approximately in accordance with gaussian distribution (i.e., normal distribution). Therefore, the height between two adjacent sampling positions is as shown in equation (4).
h d (x k+1 )＝h d (x k )+δ k (4)
In the formula (4), d and k are positive integers, δ k Satisfies the mean value of 0 and the variance of σ k Normal distribution of (a), x k Indicating the horizontal distance of the propagation path of the evaporation waveguide.
Further deducing a plurality of Markov chains by formula (4) to obtain a Markov chain matrix, as shown in formula (5)
In equation (5), each row represents a Markov chain, and each Markov chain represents a horizontally distributed sample. In particular, assume an initial value of height h d,1 (x 1 )＝20m，x n (n is an integer) and x n+1 With a spacing distance of 1km, variance σ k At 1m, 100 horizontally distributed samples (i.e., markov chains) generated using equation (4) are shown in fig. 14.
S402: and extracting the eigenvalue and the eigenvector of the Markov chain matrix by using a principal component analysis method of K-L transformation to obtain a horizontal distribution model of the height of the evaporation waveguide.
The implementation process of the principal component analysis method of the K-L transformation comprises the following steps:
calculating H shown in equation (5) d Obtaining a covariance matrix Cov with the dimensionality of C multiplied by C by the covariance of any two columns of elements in the system, and diagonalizing the covariance matrix Cov to obtain a characteristic vector matrix. And calculating to obtain an eigenvalue matrix according to a relational expression between the covariance matrix Cov and the eigenvector matrix. As can be seen from the above equation (4), the horizontal non-uniform distribution of the height of the evaporation waveguide is generated according to the normal distribution, and therefore, the feature value in the feature value matrix is subjected to principal component analysis (a conventional statistical method) to obtain the coe p (specifically, it is expressed as a distribution of the components in a uniform manner
In which λ is a random value p The smallest eigenvalue in the matrix of eigenvalues). Will be composed of a feature vector matrix and coe p The normal distribution is formed as the parameter delta of the above formula (4) k And obtaining a horizontal distribution model of the height of the evaporation waveguide, as shown in formula (6).
In formula (6), P (P) represents a feature vector matrix.
Note that the relation between the covariance matrix Cov and the eigenvector matrix is as shown in equation (7).
Cov＝PΛP T (7)
In formula (7), Λ represents an eigenvalue matrix, and the eigenvalue matrix Λ includes a plurality of eigenvalues. In particular, assume x n (n is an integer) and x n+1 The separation distance therebetween is 1km, the rank of the covariance matrix Cov is p, and p =100, the eigenvalues in the eigenvalue matrix Λ are as shown in equation (8).
λ 1 ≥λ 2 ≥λ 3 …≥λ p ＞0 (8)
It should be noted that, as a result of principal component analysis performed on the eigenvalues in the eigenvalue matrix, a plurality of eigenvalues (for example, 5) exist in the eigenvalue matrix, and a ratio of a cumulative sum of the eigenvalues to a sum of all eigenvalues reaches a preset ratio (for example, 95%), and then the eigenvalues are regarded as principal eigenvalues (i.e., principal components). Similarly, when principal component analysis is performed on the eigenvectors in the eigenvector matrix P, the eigenvectors corresponding to the plurality of eigenvalues are regarded as principal eigenvectors.
Specifically, eigenvalues of 100 markov chains (each markov chain represents one horizontally distributed sample) shown in fig. 14 are extracted to obtain an eigenvalue matrix, and each eigenvalue in the eigenvalue matrix is shown in table 1. As can be seen from Table 2, the first 5 eigenvalues λ 1 、λ 2 、λ 3 、λ 4 And λ 5 The cumulative sum of (c) represents 97% of the sum of all characteristic values, significantly greater than 95%. Thus, will λ 1 、λ 2 、λ 3 、λ 4 And λ 5 As the main eigenvalue. Will be lambda 1 、λ 2 、λ 3 、λ 4 And λ 5 The horizontal distribution of the heights of the 20 evaporation waveguides was calculated by substituting the value of (b) into the formula (6), as shown in fig. 11 below.
TABLE 2
As shown in fig. 5, the specific implementation process of the K-means clustering algorithm includes the following steps:
s501: s radio wave propagation factor samples are selected from an experimental database and used as initial clustering centers of hidden layer radial basis functions of a preset RBF neural network
S502: selecting L electric wave propagation factor samples F from an experiment database l (r) (1 ≦ L) (different from the S wave propagation factor samples) as input samples of the predetermined RBF neural network.
S503: and calculating the Euclidean distance between the target input sample and the clustering center, and if the Euclidean distance between the target input sample and the clustering center is smaller than the preset distance, matching the target input sample with the clustering center.
Wherein the target input sample is any one of the L input samples. In the present embodiment, the expression for matching the target input sample with the cluster center is shown in formula (9), i.e. the target input sample F l (r) clustering centers classified in the j-th class.
S504: using equation (10), the value of the cluster center is updated.
In the formula (10), t represents the number of times of updating the value of the clustering center, η represents a preset learning step length of the RBF neural network, and the value range of η is 0 to 1. Under the condition that a preset RBF neural network inputs a new input sample, if the clustering center of the jth class changes, only one clustering center of a radial basis function is updated each time, and the original clustering center still remains unchanged.
S505: and judging whether the value of the clustering center is in a preset threshold range.
And if the value of the clustering center is within the preset threshold range, determining that the value of the clustering center is converged. Therefore, the value of the converged clustering center is used as an estimation value of the center of the radial basis function of the preset RBF neural network.
If the value of the cluster center is not within the preset threshold range, repeating the steps S502-S504 to converge the value of the cluster center.
As shown in fig. 6, the concrete implementation process of the steepest descent algorithm based on the error method includes the following steps:
s601: and defining the weight and the threshold of the output layer in the preset RBF neural network as parameters, and initializing the parameters.
After the weight and the threshold are initialized, the weight w (t = 0) is initially set at t =0 (t is the number of iterations) (in the estimation process, the threshold and the weight are combined into a matrix for estimation), as shown in formula (11).
When L wave propagation factor samples are used as the input of the preset RBF neural network, the output of the hidden layer is as shown in formula (12), and the actual output of the output layer is as shown in formula (13).
Therefore, the expected output of the output layer is as shown in equation (14).
S602: an error between the expected output of the output layer and the actual output of the output layer is calculated.
The calculation process of the error is shown in formula (15).
e(t)＝Y′-Y(t) (15)
S603: adjusting the weight and the threshold value to obtain w (t + 1) T Decreasing in the direction of minimum gradient.
Wherein, w (t + 1) T The calculation process of (c) is shown in equation (16).
w(t+1) T ＝w(t) T +η 1 Φ T e(t) (16)
In equation (16), t is used to indicate the number of adjustments, η 1 And the learning step length is used for indicating the preset RBF neural network.
S604: and substituting the adjusted weight and the threshold value into a preset RBF neural network, calculating the actual output and the expected output of the output layer when the L-th input sample is input, substituting the actual output and the expected output of the output layer into a Mean Square Error calculation formula, calculating to obtain a Mean Square Error (MSE), and judging whether the Mean Square Error is smaller than the preset threshold value.
Wherein the mean square error represents the deviation between the actual output and the predicted output of the RBF neural network, and the mean square error is calculated as shown in equation (17).
In the formula (17), y l (x k ) Representing the actual output of the output layer at the input of the L-th sample, y' l (x k ) Representing the expected output of the output layer at the input of the lth input sample.
It should be noted that, if the value of MSE is smaller than the preset threshold, the adjusted weight and threshold are determined to converge. Therefore, the weight obtained by the convergence is used as an estimated value of the weight of the output layer, and the threshold obtained by the convergence is used as an estimated value of the threshold of the output layer.
If the value of the MSE is not less than the preset threshold, S602-603 is repeatedly executed, so that the adjusted weight and threshold are continuously converged until the MSE can be less than the preset threshold.
As shown in fig. 7, the specific process of estimating the number of nodes of the input layer, the output layer, and the hidden layer includes the following steps:
s701: and calculating the sum of the propagation distance of the evaporation waveguide and the sampling interval distance of the height of the evaporation waveguide, and taking the sum as the node number of the input layer.
Specifically, assuming that the horizontal distance of the propagation path of the evaporation waveguide is in the range of 10-40km, the propagation distance of the evaporation waveguide is 30km, and the sampling interval distance of the height of the evaporation waveguide is 1km, the number N of nodes of the input layer should be set to 31.
It should be noted that the above specific implementation process is only for illustration.
S702: and calculating the maximum value of the height of the evaporation waveguide and the quotient of the sampling interval distance of the height of the evaporation waveguide, and taking the quotient as the number of nodes of the output layer.
Specifically, assuming that the evaporation waveguide refractive index profile sample indicates that the height of the evaporation waveguide ranges from 0km to 40km, the maximum value of the height of the evaporation waveguide is 40, and the sampling interval distance of the height of the evaporation waveguide is 1km, the number K of nodes of the output layer should be set to 40.
It should be noted that the above specific implementation process is only for illustration.
S703: and setting the number of the nodes of the hidden layer according to the corresponding relation between the preset mean square error of the RBF neural network and the number of the nodes of the hidden layer.
In the training process of the preset RBF neural network, if the mean square error (that is, if the MES shown in the formula (17) has a larger value, it indicates that the prediction effect of the preset RBF neural network is worse), the number of nodes of the hidden layer will also be reduced, and if the mean square error increases, the number of nodes of the hidden layer will also be increased.
Specifically, assuming that the MES initial value of the preset RBF neural network is 0.1, the training sample is input into the preset RBF neural network, and the preset RBF neural network is trained, and in the training process of the RBF neural network, the number of nodes of the hidden layer is a change curve adjusted along with the change of the MES, as shown in fig. 15. Therefore, the spatial distribution prediction network of the evaporation waveguide based on the RBF neural network needs a plurality of linear mappings to satisfy, that is, the larger the number of nodes of the hidden layer is, the better the prediction effect of the spatial distribution prediction network of the evaporation waveguide is.
Specifically, a specific structure of an obtained spatial distribution prediction network of the evaporation waveguide by training a preset RBF neural network is shown in fig. 16. In fig. 16, the number of nodes of the input layer is 31, the number of nodes of the hidden layer is 40, and the number of nodes of the output layer is 40.
In order to verify the prediction effect of the spatial distribution prediction network of the evaporation waveguide, a radio wave propagation factor sample shown in fig. 17b is used (specifically, radar parameters are: radio wave frequency f =10GHZ, and transmitting antenna height h) t =5 m) as a prediction target, and an output result (a vertical distribution of the height of the evaporation waveguide is not considered because the height of the evaporation waveguide is acquired in real time, only a horizontal distribution of the degree is considered, and only a horizontal distribution of the height is considered in a comparison process described below) is obtained as shown in fig. 18 a. Comparing the output results shown in fig. 18a with the error between the horizontal distribution samples of the heights of the evaporation waveguides shown in fig. 17a, the error results obtained by the comparison are shown in fig. 18 b.
The height of the evaporation waveguide shown in fig. 18a is substituted into the formula (3), and the radio wave propagation factor is calculated (the radar parameters are the same as those of the radio wave propagation factor sample shown in fig. 17b, and both are the radio wave frequency f =10GHZ, and the transmitting antenna height h is calculated t =5 m), the calculation result is shown in fig. 19 a. The absolute error between the radio wave propagation factor shown in fig. 19a and the sample of the radio wave propagation factor shown in fig. 17b is compared, and the result of the absolute error is shown in fig. 19 b.
Using the RFC inversion model, the radio wave propagation factor sample shown in fig. 17b was inverted using the RFC inversion model to obtain a prediction result of the horizontal distribution of the height of the evaporation waveguide, which is shown in fig. 20 a. Comparing the error between the horizontal distribution of the height of the evaporation waveguide shown in fig. 20a and the horizontal distribution sample of the height of the evaporation waveguide shown in fig. 17a, the error result obtained by the comparison is shown in fig. 20 b.
The height of the evaporation waveguide shown in fig. 20a is substituted into the formula (3), and the radio wave propagation factor is calculated (the radar parameters are the same as those of the radio wave propagation factor sample shown in fig. 17b, and the radio wave frequency f =10GHZ, and the transmitting antenna height h t =5 m), the calculation result is shown in fig. 21 a. Compare the electricity shown in FIG. 21aFig. 21b shows the absolute error result obtained by comparing the absolute error between the wave propagation factor and the sample of the wave propagation factor shown in fig. 17 b.
By comparing fig. 18b and fig. 20b, the following conclusions can be drawn:
the first error is less than the second error. Wherein, the first error specifically refers to: the spatial distribution of the evaporation waveguide predicts the error between the output of the network and the horizontally distributed sample of the height of the evaporation waveguide. The second error specifically refers to: error between the output of the RFC inversion model and the horizontally distributed sample of the height of the evaporation waveguide.
By comparing fig. 19b and fig. 21b, the following conclusions can be drawn:
the first absolute error is less than the second absolute error. Wherein, the first absolute error specifically means: and (3) substituting the output result (namely the height of the evaporation waveguide) of the spatial distribution prediction network of the evaporation waveguide into the formula (3), and calculating the obtained radio wave propagation factor, wherein the absolute error between the radio wave propagation factor and the radio wave propagation factor sample. The second absolute error specifically means: substituting the output result of the RFC inversion model (namely the height of the evaporation waveguide) into the formula (3), and calculating the obtained radio wave propagation factor and the absolute error between the radio wave propagation factor and the radio wave propagation factor sample.
In conclusion, the prediction effect of the spatial distribution prediction network of the evaporation waveguide is superior to that of the RFC inversion model.
Corresponding to the inversion method of the propagation characteristic of the evaporation waveguide shown in fig. 1, as shown in fig. 22, an architecture diagram of an inversion apparatus of the propagation characteristic of the evaporation waveguide provided in an embodiment of the present application includes:
the calculation unit 100 is configured to obtain the height of the evaporation waveguide acquired by the radar, and calculate a radio wave propagation factor according to the radar parameter of the radar and the height of the evaporation waveguide, where the radio wave propagation factor is used to represent a first mapping relationship between an atmospheric environment factor and an evaporation waveguide propagation characteristic.
The prediction unit 200 is configured to input the radio wave propagation factor into a pre-constructed spatial distribution prediction network of the evaporation waveguide to obtain an evaporation waveguide refractive index profile of the evaporation waveguide, where the spatial distribution prediction network of the evaporation waveguide is configured to represent a second mapping relationship between the radio wave propagation factor and the evaporation waveguide propagation characteristic, convert a nonlinear indifference of the second mapping relationship in a low-latitude space into a linear separable in a high-dimension space, and use the evaporation waveguide refractive index profile to represent the propagation characteristic of the evaporation waveguide.
The construction process of the spatial distribution prediction network of the evaporation waveguide comprises the following steps: and inputting the electric wave propagation factor samples in a preset experiment database as training samples of a preset radial basis function neural network to obtain an output result. Training the radial basis function neural network based on the radio wave propagation factor sample, the evaporation waveguide refractive index profile sample corresponding to the radio wave propagation factor sample and the output result until the radial basis function neural network outputs the evaporation waveguide refractive index profile corresponding to the radio wave propagation factor sample, and determining that the radial basis function neural network obtained by current training is a spatial distribution prediction network of the evaporation waveguide. The experimental database comprises a preset corresponding relation between an evaporation waveguide refractive index profile sample and an electric wave propagation factor sample.
The preset process of the experiment database comprises the following steps: and taking the height sample of the evaporation waveguide obtained by collection as a parameter, respectively modeling the vertical distribution of the height of the evaporation waveguide and the horizontal distribution of the height of the evaporation waveguide, and taking the combination of the vertical distribution model and the horizontal distribution model of the height of the evaporation waveguide as a refractive index model of the evaporation waveguide. And outputting a sample of the refractive index profile of the evaporation waveguide by using the refractive index model of the evaporation waveguide. And inputting the height indicated by the evaporation waveguide refractive index profile sample and a preset radar parameter sample into a preset electric wave propagation factor model to obtain an electric wave propagation factor sample. And recording the corresponding relation between the evaporation waveguide refractive index profile sample and the electric wave propagation factor sample.
A process for modeling the horizontal distribution of height of an evaporation waveguide, comprising: and generating a Markov chain matrix by using a Gaussian-Markov process. And extracting the eigenvalue and the eigenvector of the Markov chain matrix by using a principal component analysis method of K-L transformation to obtain a horizontal distribution model of the height of the evaporation waveguide. Wherein the Markov chain matrix is used for representing a horizontal distribution sample of the height of the evaporation waveguide.
A vertical distribution model of the height of the evaporation waveguide, including a P-J model.
The process of constructing the wave propagation factor model includes: and establishing a calculation formula of the electric wave propagation factor by taking a parabolic equation of electric wave propagation and electric field distribution of the electric wave in the evaporation waveguide environment as parameters.
The training process of the radial basis function neural network comprises the following steps: the center and standard deviation of the radial basis functions of the hidden layer in the radial basis function neural network are estimated. And estimating the weight and the threshold of an output layer in the radial basis function neural network. And estimating the respective node numbers of an input layer, a hidden layer and an output layer in the radial basis function neural network.
In conclusion, the height of the evaporation waveguide is obtained through radar acquisition, and the radio wave propagation factor is calculated according to the radar parameters of the radar and the height of the evaporation waveguide. And inputting the electric wave propagation factor into a pre-constructed space distribution prediction network of the evaporation waveguide to obtain the evaporation waveguide refractive index profile of the evaporation waveguide. Compared with the prior art, the first mapping relation between the atmospheric environmental factor and the evaporation waveguide propagation characteristic is represented by the electric wave propagation factor in the embodiment, so that the atmospheric environmental factor is not used as a reference basis in the process of inverting the evaporation waveguide propagation characteristic, and the spatial distribution prediction network of the evaporation waveguide is used for representing the second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, that is, the prediction result of the spatial distribution prediction network of the evaporation waveguide is not influenced by the atmospheric environmental factor, so that the interference of the atmospheric environmental factor on the evaporation waveguide propagation characteristic inversion process can be avoided. In addition, the spatial distribution prediction network of the evaporation waveguide can convert the nonlinear irreparable of the second mapping relation in the low-latitude space into the linear separable of the high-dimensional space, so that the inversion process of the propagation characteristic of the evaporation waveguide can be efficient and accurate.
The present application also provides a computer storage medium comprising a stored program, wherein the program performs the method of inversion of evaporation waveguide propagation characteristics as provided herein above.
The present application further provides an electronic device, including: a processor, memory, and a bus. The processor is connected with the memory through a bus, the memory is used for storing programs, and the processor is used for running the programs, wherein the program runs to execute the inversion method of the propagation characteristics of the evaporation waveguide provided by the application.
The functions described in the method of the embodiment of the present application, if implemented in the form of software functional units and sold or used as independent products, may be stored in a storage medium readable by a computing device. Based on such understanding, part of the contribution to the prior art of the embodiments of the present application or part of the technical solution may be embodied in the form of a software product stored in a storage medium and including several instructions for causing a computing device (which may be a personal computer, a server, a mobile computing device or a network device) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In the present specification, the embodiments are described in a progressive manner, and each embodiment focuses on differences from other embodiments, and the same or similar parts between the embodiments are referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method for inverting propagation characteristics of an evaporation waveguide, comprising:
the method comprises the steps of obtaining the height of an evaporation waveguide acquired by a radar, and calculating a radio wave propagation factor according to radar parameters of the radar and the height of the evaporation waveguide, wherein the radio wave propagation factor is used for representing a first mapping relation between atmospheric environmental factors and evaporation waveguide propagation characteristics;
inputting the electric wave propagation factor into a pre-constructed spatial distribution prediction network of the evaporation waveguide to obtain an evaporation waveguide refractive index profile of the evaporation waveguide, wherein the spatial distribution prediction network of the evaporation waveguide is used for representing a second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, and converting the nonlinear indifference of the second mapping relation in a low-latitude space into linear divisible of a high-dimensional space, and the evaporation waveguide refractive index profile is used for representing the propagation characteristic of the evaporation waveguide.
2. The method according to claim 1, wherein the construction process of the spatial distribution prediction network of the evaporation waveguide comprises:
inputting a radio wave propagation factor sample in a preset experiment database as a training sample of a preset radial basis function neural network to obtain an output result;
training the radial basis function neural network based on the electric wave propagation factor sample, the evaporation waveguide refractive index profile sample corresponding to the electric wave propagation factor sample and the output result until the radial basis function neural network outputs the evaporation waveguide refractive index profile corresponding to the electric wave propagation factor sample;
and determining the trained radial basis function neural network as a spatial distribution prediction network of the evaporation waveguide.
3. The method of claim 2, wherein the pre-set procedure of the experiment database comprises:
respectively modeling the vertical distribution of the heights of the evaporation waveguides and the horizontal distribution of the heights of the evaporation waveguides by taking the height samples of the evaporation waveguides obtained by collection as parameters, and taking the combination of the vertical distribution model and the horizontal distribution model of the heights of the evaporation waveguides as an evaporation waveguide refractive index model;
outputting an evaporation waveguide refractive index profile sample by using the evaporation waveguide refractive index model;
inputting the height indicated by the evaporation waveguide refractive index profile sample and a preset radar parameter sample into a preset radio wave propagation factor model to obtain a radio wave propagation factor sample;
and recording the corresponding relation between the evaporation waveguide refractive index profile sample and the electric wave propagation factor sample.
4. The method of claim 3, wherein the process of modeling the horizontal distribution of the height of the evaporation waveguide comprises:
generating a Markov chain matrix by using a Gauss-Markov process;
extracting characteristic values and characteristic vectors of the Markov chain matrix by using a principal component analysis method of K-L transformation to obtain the horizontal distribution model of the height of the evaporation waveguide;
wherein the Markov chain matrix is used to characterize a horizontally distributed sample of the height of the evaporation waveguide.
5. The method of claim 3, wherein the vertical distribution model of the height of the evaporation waveguide comprises:
P-J model.
6. The method according to claim 3, wherein the constructing process of the wave propagation factor model includes:
and establishing a calculation formula of the electric wave propagation factor by taking a parabolic equation of electric wave propagation and electric field distribution of the electric wave in the evaporation waveguide environment as parameters.
7. The method of claim 2, wherein the training process of the radial basis function neural network comprises:
estimating the center and standard deviation of the radial basis function of the hidden layer in the radial basis function neural network;
estimating the weight and the threshold of an output layer in the radial basis function neural network;
estimating respective numbers of nodes of an input layer, the hidden layer, and the output layer in the radial basis function neural network.
8. An apparatus for inverting propagation characteristics of an evaporation waveguide, comprising:
the calculation unit is used for acquiring the height of the evaporation waveguide acquired by the radar, and calculating a radio wave propagation factor according to the radar parameter of the radar and the height of the evaporation waveguide, wherein the radio wave propagation factor is used for representing a first mapping relation between an atmospheric environment factor and the propagation characteristic of the evaporation waveguide;
the prediction unit is used for inputting the electric wave propagation factor into a pre-constructed spatial distribution prediction network of the evaporation waveguide to obtain an evaporation waveguide refractive index profile of the evaporation waveguide, the spatial distribution prediction network of the evaporation waveguide is used for representing a second mapping relation between the electric wave propagation factor and the evaporation waveguide propagation characteristic, and converting the nonlinear indifference of the second mapping relation in a low-latitude space into linear divisible of a high-dimensional space, and the evaporation waveguide refractive index profile is used for representing the propagation characteristic of the evaporation waveguide.
9. A computer storage medium, characterized in that the computer storage medium comprises a stored program, wherein the program performs the method of inversion of propagation characteristics of an evaporation waveguide according to any one of claims 1 to 7.
10. An electronic device, comprising: a processor, memory, and a bus; the processor and the memory are connected through the bus;
the memory is used for storing a program and the processor is used for running the program, wherein the program is run for executing the inversion method of the propagation characteristic of the evaporation waveguide according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010429593.9A CN111610514B (en) | 2020-05-20 | 2020-05-20 | Inversion method and device for propagation characteristics of evaporation waveguide |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010429593.9A CN111610514B (en) | 2020-05-20 | 2020-05-20 | Inversion method and device for propagation characteristics of evaporation waveguide |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111610514A CN111610514A (en) | 2020-09-01 |
| CN111610514B true CN111610514B (en) | 2023-01-31 |
Family
ID=72195091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010429593.9A Active CN111610514B (en) | 2020-05-20 | 2020-05-20 | Inversion method and device for propagation characteristics of evaporation waveguide |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111610514B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114065607B (en) * | 2021-10-19 | 2022-11-29 | 中国电波传播研究所（中国电子科技集团公司第二十二研究所） | Offshore non-uniform evaporation waveguide profile inversion method based on deep convolutional network |
| CN114035172B (en) * | 2021-11-12 | 2024-07-23 | 中国人民解放军国防科技大学 | Method for inverting atmospheric waveguide by radar echo data based on particle filtering |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102540162A (en) * | 2011-12-12 | 2012-07-04 | 中国船舶重工集团公司第七二四研究所 | Method for estimating low-altitude electromagnetic wave propagation characteristic on basis of sea clutter |
| CN106354979A (en) * | 2016-10-08 | 2017-01-25 | 西安电子科技大学 | Method for inverting evaporation waveguide of radar sea clutters based on quantum genetic algorithm |
| CN106772300A (en) * | 2016-12-02 | 2017-05-31 | 中国电波传播研究所（中国电子科技集团公司第二十二研究所） | A kind of microwave over-the-horizon radar reflectogram computational methods |
| CN106772386A (en) * | 2016-12-13 | 2017-05-31 | 中国人民解放军理工大学 | One kind is using LPSO algorithms by radar return inverting atmospheric duct method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190128805A1 (en) * | 2015-03-16 | 2019-05-02 | Fredrick S. Solheim | Characterizing tropospheric boundary layer thermodynamic and refractivity profiles utilizing selected waveband infrared observations |
-
2020
- 2020-05-20 CN CN202010429593.9A patent/CN111610514B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102540162A (en) * | 2011-12-12 | 2012-07-04 | 中国船舶重工集团公司第七二四研究所 | Method for estimating low-altitude electromagnetic wave propagation characteristic on basis of sea clutter |
| CN106354979A (en) * | 2016-10-08 | 2017-01-25 | 西安电子科技大学 | Method for inverting evaporation waveguide of radar sea clutters based on quantum genetic algorithm |
| CN106772300A (en) * | 2016-12-02 | 2017-05-31 | 中国电波传播研究所（中国电子科技集团公司第二十二研究所） | A kind of microwave over-the-horizon radar reflectogram computational methods |
| CN106772386A (en) * | 2016-12-13 | 2017-05-31 | 中国人民解放军理工大学 | One kind is using LPSO algorithms by radar return inverting atmospheric duct method |
Non-Patent Citations (2)
| Title |
|---|
| CASPER West Evaporation Duct Height Inversion Using LATPROP-Radar;Joshua Compalec 等;《2018 IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting》;20180713;第881-882页 * |
| 基于雷达海杂波与蒸发波导模型参数化的反演方法研究;左雷等;《舰船电子工程》;20170331;第37卷(第03期);第78-82页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111610514A (en) | 2020-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106600059B (en) | Intelligent power grid short-term load prediction method based on improved RBF neural network | |
| CN109063911B (en) | Load aggregation grouping prediction method based on gated cycle unit network | |
| CN111222549B (en) | Unmanned aerial vehicle fault prediction method based on deep neural network | |
| US10387768B2 (en) | Enhanced restricted boltzmann machine with prognosibility regularization for prognostics and health assessment | |
| CN108520111B (en) | Soft measurement method based on optimal selection and optimal regression of orthogonal components | |
| CN111610514B (en) | Inversion method and device for propagation characteristics of evaporation waveguide | |
| CN102185735A (en) | Network security situation prediction method | |
| CN111768000A (en) | Industrial process data modeling method for online adaptive fine-tuning deep learning | |
| CN106778838A (en) | A kind of method for predicting air quality | |
| CN112367130B (en) | Wireless network signal transmission intensity calculation method and computer storage medium | |
| CN115270634B (en) | Counterfactual interpretation generation method and system suitable for autonomous air combat field | |
| CN113743504A (en) | Voltage sag source identification model construction method and device, terminal and medium | |
| CN112149883A (en) | Photovoltaic power prediction method based on FWA-BP neural network | |
| CN104795063A (en) | Acoustic model building method based on nonlinear manifold structure of acoustic space | |
| CN112001115A (en) | Soft measurement modeling method of semi-supervised dynamic soft measurement network | |
| Han et al. | A Hybrid BPNN-GARF-SVR PredictionModel Based on EEMD for Ship Motion. | |
| Li et al. | Multiple-input multiple-output soft sensors based on KPCA and MKLS-SVM for quality prediction in atmospheric distillation column | |
| CN117350170A (en) | Nonlinear oscillation analysis method based on KOOPHAN deep neural network | |
| CN110866838A (en) | Network representation learning algorithm based on transition probability preprocessing | |
| CN115526376A (en) | Multi-feature fusion generation countermeasure network ultra-short-term wind power prediction method | |
| CN110728024B (en) | Vine copula-based soft measurement method and system | |
| Wu et al. | The forecasting model based on fuzzy novel ν-support vector machine | |
| Wu et al. | Parametric Model for Coaxial Cavity Filter with Combined KCCA and MLSSVR | |
| CN117688367B (en) | Wind power generation ultra-short term power prediction method and device based on instant learning | |
| CN116595883B (en) | Real-time online system state correction method for numerical reactor |
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 |