CN112560342A - DNN-based atmospheric waveguide parameter estimation method - Google Patents
DNN-based atmospheric waveguide parameter estimation method Download PDFInfo
- Publication number
- CN112560342A CN112560342A CN202011481232.5A CN202011481232A CN112560342A CN 112560342 A CN112560342 A CN 112560342A CN 202011481232 A CN202011481232 A CN 202011481232A CN 112560342 A CN112560342 A CN 112560342A
- Authority
- CN
- China
- Prior art keywords
- dnn
- atmospheric waveguide
- layer
- training
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Measurement Of Resistance Or Impedance (AREA)
Abstract
The invention discloses an atmospheric waveguide parameter estimation method based on DNN, which utilizes a DNN network to extract attenuation characteristics of signals in an atmospheric waveguide, converts a parameter estimation problem into a DNN network characteristic extraction problem and finds out a mapping relation between radio wave signal attenuation and atmospheric waveguide parameters at different distances. And establishing a propagation model of the X-band signal in the atmospheric waveguide, taking power sampling points at different distances as input, and performing atmospheric waveguide parameter estimation through the radio wave signal of the sparse frequency point, thereby achieving atmospheric waveguide parameter estimation of continuous frequency. The invention can solve the defects of low precision and long time consumption of numerical calculation, can accurately predict results at a longer distance, and further can obtain continuous frequency estimation results by utilizing sparse point frequency training.
Description
Technical Field
The invention relates to a method for estimating air waveguide parameters, in particular to a method for estimating atmospheric waveguide parameters based on DNN (deep numerical network), belonging to the field of signal processing.
Background
Atmospheric waveguides are a typical event in a near-earth atmospheric environment. They represent an abnormal state of the atmosphere. The waveguide traps electromagnetic waves in a specific atmosphere and transmits the waves over a distance of beyond visual range with low propagation loss, which greatly enhances the propagation distance of radio waves. It also causes long-distance co-channel interference in the TD-LTE network system, which can seriously affect the urban communication system.
The atmospheric waveguide forecast information can also be obtained based on the numerical weather forecast technology, and the technology is restricted by the development of the numerical weather forecast technology and is still in a starting stage. The detection of the atmospheric waveguide can be divided into two types, namely contact measurement and remote sensing inversion, wherein the contact measurement and the remote sensing inversion are the traditional atmospheric waveguide detection technologies, including microwave refractometer measurement, meteorological sounding, rocket sounding and the like; the remote sensing inversion of the atmospheric waveguide is a hotspot and a difficulty in recent research, and technologies such as radar sea clutter inversion of the atmospheric waveguide, GNSS scattered signal monitoring inversion of the atmospheric waveguide, laser radar and microwave radiometer detection of the atmospheric waveguide are successively provided.
In recent years, with continuous research and study in the field of deep learning, great success has been achieved in the fields of voice recognition, image recognition, and natural language. The deep learning can be applied to small sample atmospheric waveguide parameter estimation to replace the original detection technology, so that how to improve the precision optimization algorithm through the deep learning is a problem to be solved.
Disclosure of Invention
Aiming at the prior art, the invention aims to provide an atmospheric waveguide parameter estimation method based on DNN (deep numerical network), and solve the problems of high operation difficulty, complex flow and the like of the traditional mathematical model for atmospheric waveguide parameter estimation at present.
In order to solve the technical problem, the method for estimating the atmospheric waveguide parameters based on DNN comprises the following steps:
step 1: the electromagnetic wave propagation loss model is modeled by adopting a narrow-angle parabolic equation, and specifically comprises the following steps:
wherein, the horizontal propagation direction of the electromagnetic wave is an x axis, the vertical direction is a z axis, n (x, z) is a refractive index, k0Representing the propagation constant in vacuum, u (x, z) satisfies:
Ψ (x, z) represents an electric or magnetic field component;
calculating the propagation loss L of the electromagnetic wave in the waveguide, specifically:
L＝Lf-LA＝32.45+20lg f+20lg r-20lg F
wherein L isfFor electromagnetic wave free space propagation loss, LAThe medium propagation loss is represented by r, the distance between a transmitting source and a receiving point is represented by F, the propagation factor in the waveguide is represented by F, and the frequency is represented by F;
step 2: establishing a network training set, selecting a point at set intervals within a selected horizontal range for sampling to obtain electromagnetic wave power, calculating a propagation loss value relative to a transmitting point according to the step 1 to form a group of propagation loss sequence values as input of the network, and selecting a single-carrier frequency sinusoidal signal s (t) by an antenna signal, wherein the method specifically comprises the following steps:
s(t)＝sin(2πfi)
wherein f isiDividing the frequency range selected by the training set into a plurality of small training sets according to the frequency range for the frequency of the signal, setting the signal frequency of each small training set to be the same as the height of the atmospheric waveguide, selecting a point at a given height within the height range of the atmospheric waveguide to simulate to obtain a plurality of groups of data, and setting the heights of the antenna and the receiver;
and step 3: constructing a DNN branch, and defining layers and parameters of DNN, wherein the method specifically comprises the following steps: the DNN has five layers in total, including an input layer, an output layer, and three hidden layers, the output of each layer being defined as:
assuming that there are m neurons in layer l-1 and n neurons in layer l, the linear coefficients W in layer l form an n × m matrix WlThe bias value b of the l-th layer constitutes an n x l vector blThe output a of layer l-1 constitutes an m x l vector al-1The linear output z of the l-th layer before being activated constitutes an n x l vector zlI.e. the output of the l layer;
the activation functions are all ReLU functions, and the ReLU activation functions are specifically:
and 4, step 4: training the model by using a training set, evaluating the model by using a verification set, and continuously adjusting the model parameters until the model parameters meet the precision requirement to obtain a neural network model, which specifically comprises the following steps: the loss function adopts a quadratic loss function as follows:
wherein y represents a true value, y' represents a predicted value, and the training batch number and the testing batch number are set;
setting an initial value of a learning rate, adopting an exponential decay automatic learning rate adjusting mode, calculating signal and channel data through a DNN network, and outputting an original phase difference of antenna signals; after multiple iterative training, adjusting the weight of each neuron of the DNN model through the error of the estimation result until the error meets the precision requirement, and storing the training model;
and 5: and performing atmospheric waveguide parameter estimation on the trained model by using the test set, and outputting an atmospheric waveguide height estimation result.
The invention has the beneficial effects that: the invention provides a parameter estimation method based on a deep neural network, which is based on the refractive index distribution of an evaporation waveguide and is combined with deep learning to establish a network mapping model between the power attenuation of a radio wave signal and atmospheric waveguide parameters. The model is applied to the atmospheric altitude estimation inversion problem, the inversion result is analyzed, the feasibility of deep learning in the inversion problem is verified, and a network is trained by using a sparse frequency point radio wave signal of a small sample, so that the estimation under the full frequency radio wave signal is achieved.
The present invention will replace the traditional numerical calculation method with a DNN network. The network is used for replacing a numerical calculation method, so that the defects of low precision and long time consumption of numerical calculation can be overcome, accurate prediction results can be obtained at a longer distance, and furthermore, the results of continuous frequency estimation can be obtained by using sparse point frequency training.
Drawings
FIG. 1 is a schematic diagram of atmospheric waveguide parameter estimation experiment data acquisition in accordance with the present invention;
FIG. 2 is a block diagram of the overall system architecture of the present invention;
FIG. 3 is an experimental flow chart of the present invention;
FIG. 4 is a graph of the gradient of the minimum mean square error of the present invention as a function of training steps;
FIG. 5 is a result graph of error in the test experiment network output and true values of the present invention;
Detailed Description
The following further describes the embodiments of the present invention with reference to the drawings.
The invention provides an atmospheric waveguide parameter method based on a deep learning DNN network, which is characterized in that a DNN network is utilized to extract attenuation characteristics of signals in an atmospheric waveguide, a parameter estimation problem is converted into a DNN network characteristic extraction problem, and a mapping relation between radio wave signal attenuation and atmospheric waveguide parameters at different distances is found out. And establishing a propagation model of the X-band signal in the atmospheric waveguide, taking power sampling points at different distances as input, and performing atmospheric waveguide parameter estimation through the radio wave signal of the sparse frequency point, thereby achieving atmospheric waveguide parameter estimation of continuous frequency. The invention does not specially research the DNN network, but modifies the structure and parameters of the mature neural network to make it suitable for training the signal received by the receiver.
The technical scheme of the invention is an atmospheric waveguide parameter estimation method based on DNN, and the method comprises the following steps:
the method comprises the following steps: here the parabolic equation method is used for modeling the signal attenuation under the influence of the atmospheric waveguide. Since the trapping angle of the electromagnetic wave in the waveguide is relatively small, a narrow-angle parabolic equation is used here. The horizontal propagation direction of the electromagnetic wave is set as an x axis, and the vertical direction is set as a z axis. When the propagation medium is isotropic and the refractive index is n (x, z), a two-dimensional wave equation is obtained by transforming Maxwell equation system, as shown in (1):
wherein k is0Representing the propagation constant in vacuum, the parameter Ψ represents either the electric field component or the magnetic field component, introducing a simplified function, as follows:
where i is an imaginary number. The function describes only amplitude variations of the electric or magnetic field components, and is independent of phase variations. Bringing it into (1) gives:
factorization of the above formula yields:
formula (4) can be decomposed into:
equation (6) is a forward parabolic equation, and equation (7) is a backward parabolic equation, which is applied to the electromagnetic wave propagating forward or backward, respectively. Only the solution of forward parabolic equation (6) is studied here. And (3) carrying out Taylor expansion on the differential operator Q:
substituting (8) into (6) yields:
equation (9) is the narrow angle parabolic equation used in calculating the propagation characteristic at small elevation angles.
The propagation loss L of the electromagnetic wave in the waveguide is composed of two parts, wherein one part is the free-space propagation loss L of the electromagnetic wavef(ii) a Another part is the loss caused by the reflection and absorption of electromagnetic waves by the propagation medium and obstacles, called medium propagation loss LA。
LfTransmitting power P through antennatAnd the received power PrIs defined as follows in the logarithmic domain:
wherein P istAnd PrThere is also the following relationship, r representing the distance between the transmission source and the reception point:
LAthen, with respect to the propagation factor F in the waveguide, the relationship is as follows:
u (x, z) is calculated according to the parabolic equation (9), and x represents the horizontal propagation distance.
According to the above analysis, combining the two parts, the final propagation loss L can be expressed as:
L＝Lf-LA＝32.45+20lg f+20lg r-20lg F (13)
where f denotes frequency in MHz and r in km.
Step two: the training set of the network is explained with reference to fig. 2. The electromagnetic wave propagation loss model selects the parabolic equation model explained above, the input transmission loss data is calculated by formula (13), within the horizontal range of 3km to 53km, every 200m, a point is selected for sampling to obtain the electromagnetic wave power and the propagation loss value relative to the emission point is calculated, finally, 250 values are selected to form a group of propagation loss sequence values as the input of the network, and the antenna signal selects a simple single-carrier frequency sinusoidal signal s (t) as follows:
s(t)＝sin(2πfi) (14)
fifor the frequency of signals, the frequency band of the training set is selected to be 1 to 10GHz, every 1GHz is used as a small training set, the signal frequency of each small training set is the same, the height of the atmospheric waveguide is set to be 1 to 40m, 20 groups of data are obtained by selecting one point at intervals of 1m through simulation, and the heights of the antenna and the receiver are both 15 m.
Table 1: key parameters of experiment
Step three: constructing DNN branches, defining layers and parameters of DNN
DNN is defined as having five layers including an input layer, an output layer and three hidden layers, wherein the output of each layer is defined as
Assuming that layer l-1 has m neurons and layer l has n neurons, the linear coefficients W of layer l form an n × m matrix WlThe bias value b of the l-th layer constitutes an n x l vector blThe output a of layer l-1 constitutes an m x l vector al-1The linear output z of the l-th layer before being activated constitutes an n x l vector zlI.e. the output of the l-th layer.
The activation functions are all ReLU functions, and compared with sigmoid functions and tanh functions, the method can overcome the problem of gradient disappearance, accelerate training speed and activate sparsity of the functions. The ReLU activation function is shown in the following graph:
step four: and training the model by using a training set, evaluating the model by using a verification set, and continuously adjusting the model parameters so as to obtain the optimal neural network model. The loss function adopts a quadratic loss function, the formula is shown as formula (17), y represents a true value, and y' represents a predicted value. The training batch size is set to 512 and the test batch size is 256.
The initial value of the learning rate is 0.000001, and an exponential decay automatic learning rate adjusting mode is adopted. Calculating the signal plus channel data through a DNN network, and outputting the original phase difference of the antenna signals; and after multiple times of iterative training, adjusting the weight of each neuron of the DNN model through the error of the estimation result, and storing the training model.
Step five: and performing atmospheric waveguide parameter estimation on the trained model by using the test set, and outputting an atmospheric waveguide height estimation result. Test set signal selection is as in equation (14), fiThe test is performed at 1 to 10GHz, with values taken every 100MHz, waveguide heights of 1 to 40m, taken every 1 m. And finally, taking random data of 100 test sets as a round of test, and solving the standard deviation of the output phase difference to obtain a result.
The formula (18) is a formula of the standard deviation, the same formula (17) y represents a true value, and y' represents a predicted value.
The performance of the network is further verified in simulation with fig. 3:
1. an experimental scene is as follows: the experiment adopts python language to build a network, and the model is trained based on tensierflow. And the simulation platform uses Spyder software to perform performance verification. The training set is G, the signal frequency is set by selecting a point frequency point every 1GHz from 1 to 10GHz, the heights of the antenna and the receiver are both 15m, the height of the waveguide is 1 to 40m, the signal frequency is selected every 1m, each sample is repeated for 50 times, and 20000 groups of training data are obtained in total. After the deep neural network is used for training, variables such as network weight deviation and the like are saved, and a test set is used for testing. The test set G (m) is that the signal frequency is 1 to 10GHz, a value is taken every 100MHz, the waveguide height is 1 to 40m, the test data is selected every 1m, 4000 groups of test data are totally obtained, finally, the random data of 100 test sets are taken as one round of test, and the network is put into the network for testing in the network training process to obtain the error of the network output and the true value, so that the change of the network and the final network effect can be seen.
2. Analysis of Experimental content
From fig. 4, it can be seen that the MSE varies with the number of training steps, and it can be seen that the MSE for both the last training and testing is close to 0; fig. 5 shows that the test data is put into the network during the training process, and the error analysis is performed on the network and the atmospheric waveguide with the true value, so that the network output error is rapidly reduced with the increase of the training times, and finally, the network output error is stabilized within 0.5, which indicates that the network effect is good.
Claims (1)
1. A DNN-based atmospheric waveguide parameter estimation method is characterized by comprising the following steps:
step 1: the electromagnetic wave propagation loss model is modeled by adopting a narrow-angle parabolic equation, and specifically comprises the following steps:
wherein, the horizontal propagation direction of the electromagnetic wave is an x axis, the vertical direction is a z axis, n (x, z) is a refractive index, k0Representing the propagation constant in vacuum, u (x, z) satisfies:
Ψ (x, z) represents an electric or magnetic field component;
calculating the propagation loss L of the electromagnetic wave in the waveguide, specifically:
L＝Lf-LA＝32.45+20lgf+20lgr-20lgF
wherein L isfFor electromagnetic wave free space propagation loss, LAThe medium propagation loss is represented by r, the distance between a transmitting source and a receiving point is represented by F, the propagation factor in the waveguide is represented by F, and the frequency is represented by F;
step 2: establishing a network training set, selecting a point at set intervals within a selected horizontal range for sampling to obtain electromagnetic wave power, calculating a propagation loss value relative to a transmitting point according to the step 1 to form a group of propagation loss sequence values as input of the network, and selecting a single-carrier frequency sinusoidal signal s (t) by an antenna signal, wherein the method specifically comprises the following steps:
s(t)＝sin(2πfi)
wherein f isiDividing the frequency range selected by the training set into a plurality of small training sets according to the frequency range for the frequency of the signal, setting the signal frequency of each small training set to be the same as the height of the atmospheric waveguide, selecting a point at a given height within the height range of the atmospheric waveguide to simulate to obtain a plurality of groups of data, and setting the heights of the antenna and the receiver;
and step 3: constructing a DNN branch, and defining layers and parameters of DNN, wherein the method specifically comprises the following steps: the DNN has five layers in total, including an input layer, an output layer, and three hidden layers, the output of each layer being defined as:
assuming that there are m neurons in layer l-1 and n neurons in layer l, the linear coefficients W in layer l form an n × m matrix WlThe bias value b of the l-th layer constitutes an n x l vector blThe output a of layer l-1 constitutes an m x l vector al-1The linear output z of the l-th layer before being activated constitutes an n x l vector zlI.e. the output of the l layer;
the activation functions are all ReLU functions, and the ReLU activation functions are specifically:
and 4, step 4: training the model by using a training set, evaluating the model by using a verification set, and continuously adjusting the model parameters until the model parameters meet the precision requirement to obtain a neural network model, which specifically comprises the following steps: the loss function adopts a quadratic loss function as follows:
wherein y represents a true value, y' represents a predicted value, and the training batch number and the testing batch number are set;
setting an initial value of a learning rate, adopting an exponential decay automatic learning rate adjusting mode, calculating signal and channel data through a DNN network, and outputting an original phase difference of antenna signals; after multiple iterative training, adjusting the weight of each neuron of the DNN model through the error of the estimation result until the error meets the precision requirement, and storing the training model;
and 5: and performing atmospheric waveguide parameter estimation on the trained model by using the test set, and outputting an atmospheric waveguide height estimation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011481232.5A CN112560342A (en) | 2020-12-15 | 2020-12-15 | DNN-based atmospheric waveguide parameter estimation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011481232.5A CN112560342A (en) | 2020-12-15 | 2020-12-15 | DNN-based atmospheric waveguide parameter estimation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112560342A true CN112560342A (en) | 2021-03-26 |
Family
ID=75063934
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011481232.5A Pending CN112560342A (en) | 2020-12-15 | 2020-12-15 | DNN-based atmospheric waveguide parameter estimation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112560342A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113468466A (en) * | 2021-07-23 | 2021-10-01 | 哈尔滨工业大学 | Multi-working-condition one-dimensional wave equation solving method based on neural network |
| CN114325613A (en) * | 2021-12-24 | 2022-04-12 | 西南交通大学 | Radar detection power prediction method under regional non-uniform evaporation waveguide |
| CN114362850A (en) * | 2021-12-30 | 2022-04-15 | 国网江苏省电力有限公司电力科学研究院 | Method and system for determining wireless signal attenuation condition of underground environment |
| CN117033873A (en) * | 2023-10-10 | 2023-11-10 | 宁波麦思捷科技有限公司武汉分公司 | Low-altitude atmosphere waveguide measurement method and system based on neural network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104698501A (en) * | 2015-02-15 | 2015-06-10 | 山东大学 | Single-hole directional detection radar antenna for tunnel advanced geological prediction |
| GB201709904D0 (en) * | 2017-06-21 | 2017-08-02 | Soliton Holdings Corp Delaware Corp | Absorption of electromagnetic energy |
| CN107545104A (en) * | 2017-08-21 | 2018-01-05 | 西安电子科技大学 | Irregular terrain profiles radio wave propagation factor prediction method based on three dimensional parabolic equation |
| JP2019153830A (en) * | 2018-02-28 | 2019-09-12 | 株式会社東芝 | Wireless system and wireless communication method |
| CN110325929A (en) * | 2016-12-07 | 2019-10-11 | 阿瑞路资讯安全科技股份有限公司 | System and method for detecting the waveform analysis of cable network variation |
| CN111310889A (en) * | 2020-01-16 | 2020-06-19 | 西北工业大学 | Evaporation waveguide profile estimation method based on deep neural network |
-
2020
- 2020-12-15 CN CN202011481232.5A patent/CN112560342A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104698501A (en) * | 2015-02-15 | 2015-06-10 | 山东大学 | Single-hole directional detection radar antenna for tunnel advanced geological prediction |
| CN110325929A (en) * | 2016-12-07 | 2019-10-11 | 阿瑞路资讯安全科技股份有限公司 | System and method for detecting the waveform analysis of cable network variation |
| GB201709904D0 (en) * | 2017-06-21 | 2017-08-02 | Soliton Holdings Corp Delaware Corp | Absorption of electromagnetic energy |
| CN107545104A (en) * | 2017-08-21 | 2018-01-05 | 西安电子科技大学 | Irregular terrain profiles radio wave propagation factor prediction method based on three dimensional parabolic equation |
| JP2019153830A (en) * | 2018-02-28 | 2019-09-12 | 株式会社東芝 | Wireless system and wireless communication method |
| CN111310889A (en) * | 2020-01-16 | 2020-06-19 | 西北工业大学 | Evaporation waveguide profile estimation method based on deep neural network |
Non-Patent Citations (1)
| Title |
|---|
| 郭晓薇: "对流层波导特性及其基于深度学习的反演方法研究", 《CNKI优秀硕士学位论文全文库 基础科学辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113468466A (en) * | 2021-07-23 | 2021-10-01 | 哈尔滨工业大学 | Multi-working-condition one-dimensional wave equation solving method based on neural network |
| CN114325613A (en) * | 2021-12-24 | 2022-04-12 | 西南交通大学 | Radar detection power prediction method under regional non-uniform evaporation waveguide |
| CN114362850A (en) * | 2021-12-30 | 2022-04-15 | 国网江苏省电力有限公司电力科学研究院 | Method and system for determining wireless signal attenuation condition of underground environment |
| CN114362850B (en) * | 2021-12-30 | 2024-05-14 | 国网江苏省电力有限公司电力科学研究院 | Method and system for determining attenuation condition of wireless signal in underground environment |
| CN117033873A (en) * | 2023-10-10 | 2023-11-10 | 宁波麦思捷科技有限公司武汉分公司 | Low-altitude atmosphere waveguide measurement method and system based on neural network |
| CN117033873B (en) * | 2023-10-10 | 2023-12-19 | 宁波麦思捷科技有限公司武汉分公司 | Low-altitude atmosphere waveguide measurement method and system based on neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112560342A (en) | DNN-based atmospheric waveguide parameter estimation method | |
| CN109993280B (en) | Underwater sound source positioning method based on deep learning | |
| CN108845306B (en) | Laser radar echo signal denoising method based on variational modal decomposition | |
| CN110222748B (en) | OFDM radar signal identification method based on 1D-CNN multi-domain feature fusion | |
| CN107192991B (en) | A kind of radio-frequency radiation formula emulation wideband pulse radar imagery equivalent simulation method | |
| CN112036239B (en) | Radar signal working mode identification method and system based on deep learning network | |
| CN112859014A (en) | Radar interference suppression method, device and medium based on radar signal sorting | |
| CN114970378B (en) | Sea clutter sample library construction method based on GAN network | |
| CN115356702A (en) | Evaporation waveguide deep learning joint inversion method based on marine multi-source radar echo | |
| CN114065607B (en) | Offshore non-uniform evaporation waveguide profile inversion method based on deep convolutional network | |
| CN107833180A (en) | A kind of method using complex field neutral net rapid solving nonlinear electromagnetic inverse Problem | |
| CN110764084A (en) | Radar detection method for shielding target under complex terrain | |
| CN111007457A (en) | Radiation source direct positioning method based on block sparse Bayesian model | |
| CN116956737A (en) | Method for inverting electron density of low ionosphere based on neural network and genetic algorithm | |
| CN114358046B (en) | Multi-complexity-level complex electromagnetic interference environment simulation generation method and system | |
| CN115932773A (en) | Target angle detection method, device, equipment and medium based on spectrum shape characteristics | |
| CN115598714A (en) | Time-space coupling neural network-based ground penetrating radar electromagnetic wave impedance inversion method | |
| Nagao et al. | Fine-tuning for propagation modeling of different frequencies with few data | |
| Wu et al. | Enhanced path loss model by image-based environmental characterization | |
| CN116131979A (en) | Wireless channel prediction method and system under remote interference, storage medium and terminal | |
| CN110907912B (en) | Pole characteristic solving method in target narrow-band mode of resonance region | |
| CN115390031B (en) | High-resolution sea clutter modeling and simulation method | |
| Saiko et al. | Models of Improving the Efficiency of Radio Communication Systems Using the Terrahert Range | |
| Wu et al. | Environmental Knowledge-Driven Over-the-Horizon Propagation Loss Prediction Based on Short-and Long-Term Parallel Double-Flow TrellisNets | |
| Zhou et al. | Field strength prediction of high altitude drop mountain area based on deep learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210326 |
|
| RJ01 | Rejection of invention patent application after publication |