CN111523568B - Antenna array fault diagnosis method based on deep neural network and radiation data compensation - Google Patents
Antenna array fault diagnosis method based on deep neural network and radiation data compensation Download PDFInfo
- Publication number
- CN111523568B CN111523568B CN202010259285.6A CN202010259285A CN111523568B CN 111523568 B CN111523568 B CN 111523568B CN 202010259285 A CN202010259285 A CN 202010259285A CN 111523568 B CN111523568 B CN 111523568B
- Authority
- CN
- China
- Prior art keywords
- array
- data
- neural network
- failure
- radiation
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 33
- 230000005855 radiation Effects 0.000 title claims abstract description 32
- 238000003745 diagnosis Methods 0.000 title abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 41
- 238000012360 testing method Methods 0.000 claims abstract description 28
- 239000000523 sample Substances 0.000 claims abstract description 16
- 238000005259 measurement Methods 0.000 claims description 21
- 239000013598 vector Substances 0.000 claims description 18
- 230000005284 excitation Effects 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 4
- 230000005684 electric field Effects 0.000 claims description 2
- 230000006870 function Effects 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 5
- 238000005070 sampling Methods 0.000 abstract description 3
- 238000013135 deep learning Methods 0.000 description 5
- 238000003491 array Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R29/00—Arrangements for measuring or indicating electric quantities not covered by groups G01R19/00 - G01R27/00
- G01R29/08—Measuring electromagnetic field characteristics
- G01R29/10—Radiation diagrams of antennas
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention discloses an antenna array fault diagnosis method based on a deep neural network and radiation data compensation, and belongs to the field of antenna signal processing and fault identification. Firstly, in a training stage before the application of an antenna, on a plurality of measuring points in a far field area, a probe is utilized to measure radiation data of an array under a plurality of fault scenes, then the measured data is used as the input of a neural network, the number and the positions of failure array elements are respectively used as the output, and two neural networks are trained; and secondly, in an application stage, measuring far-field radiation data of the array to be diagnosed on the same sampling point as the training stage, inputting the test data into a trained neural network, preliminarily determining the position of the most probable failed array element, and gradually determining the positions of all the probable failed array elements by using a radiation compensation method. The algorithm effectively reduces the sample data required to be collected in the training stage through a data compensation method, and is suitable for large-scale array fault diagnosis.
Description
Technical Field
The invention relates to the field of antenna array signal processing and deep learning, in particular to an antenna array fault diagnosis method based on deep learning.
Background
The array antenna is a special antenna formed by more than or equal to two antenna units which are arranged according to a certain rule or randomly, and compared with a single antenna, the array antenna has many advantages, mainly comprising: high antenna gain, strong directivity, narrow beam width, and beam scanning. At present, array antennas play an indispensable role in various fields such as radar detection, mobile communication, satellite remote sensing, and the like. There is a close relationship between the radiation performance of the array antenna and the number of the radiation elements, and generally, the more the radiation elements, the better the radiation performance of the antenna. Such as in large military phased arrays, typically contain hundreds or thousands of radiating elements. However, as the array size grows, the probability of array element failure also increases. When a certain number of array elements fail, the array structure can be seriously damaged, leading to main lobe broadening, side lobe level increase, gain reduction, target detection influence and the like. In order to correct the array radiation performance degradation caused by the array element failure in time, it is important to accurately detect the number and the positions of the failed array elements.
At present, array fault diagnosis methods are mainly classified into three categories: a very near field method, a source reconstruction method, and a parametric model method. The extreme near field method is to determine the position of a failure array element by measuring the radiation near field of the array to be diagnosed; the source reconstruction method is divided into two modes of equivalent source reconstruction and excitation source reconstruction, and the difference between the two modes is that the equivalent source reconstruction is an equivalent current source reconstruction, and then the position of a failure array element is determined according to the distribution of the equivalent current source; the later directly locates the failure array element by reconstructing the excitation of the array element. The parametric model method is generally divided into two stages, first a training stage: training the parametric model using known training data; the second is an application stage: inputting the characteristic data of the array to be diagnosed into the trained parameter model to determine the position of the failed array element. In the three methods, as part of the antenna application scenes are special, and accurate near-field data is difficult to obtain, the array fault diagnosis method based on near-field measurement has higher requirements on the antenna application scenes. Compared with a source reconstruction method based on far-field measurement data, the parametric model method has the defect of slow training process, but has the advantages of high diagnosis speed and capability of realizing real-time diagnosis in an application stage.
Because the parameter model method needs a large amount of data and has a large calculation amount in the training stage, the existing array fault diagnosis method based on the parameter model only stays in the fault diagnosis of a small-scale array with a small number of failure array elements. For example, Patnaik A et al uses an artificial neural network to realize fault diagnosis of a 16-array-element linear array, Vakula D et al assumes that the number of failed array elements in the 16-array-element linear array is at most two, and designs three neural networks to realize fault diagnosis of the array. However, to meet radiation performance requirements, large-scale array applications are more widespread. Therefore, in order to utilize the advantage of rapid diagnosis by the parametric model method, it is necessary to expand the array fault diagnosis method based on the parametric model method so as to be suitable for large-scale array fault diagnosis.
Disclosure of Invention
The invention provides an antenna array fault diagnosis algorithm based on deep learning, and is suitable for antenna arrays in any shapes. The method aims to determine the number and the positions of fault array elements according to measured radiation data of an array far field on the basis of completing diagnosis model training in a training stage before antenna application. Compared with the existing array fault diagnosis algorithm based on a nonparametric model, the algorithm is not limited by the shape of the array and has higher diagnosis speed; compared with an array fault diagnosis algorithm based on a parameter model method, the algorithm needs less training samples and is suitable for large-scale array fault diagnosis.
The solution of the invention is: firstly, in a training stage before the antenna is applied, on a plurality of measuring points in a far field area, the radiation data of the array under a plurality of fault scenes are measured by using a probe, then the measured data is used as the input of a neural network, the number and the positions of failure array elements are respectively used as the output, and two neural networks are trained. And secondly, an application stage, measuring far-field radiation data of the array to be diagnosed at the same sampling point as the training stage, inputting the test data into a trained neural network, preliminarily determining the position of the most probable failed array element, and gradually determining the positions of all the probable failed array elements by using a radiation compensation method. Therefore, the technical scheme of the invention is an antenna array fault diagnosis method based on deep neural network and radiation data compensation, which comprises the following steps:
a model training stage:
step 1: measuring electric field radiation data of a far field of the array under N fault scenes by using a probe at M measuring points with different angles on a far field measuring plane; the data measured by the probe at the m-th angle is expressed by the following formula:
where c is the speed of light, f is the working frequency of the array, L is the total number of array elements contained in the array, xiAnd diExcitation of the ith array element and position in the array coordinate system, thetamIs the angle between the m-th measurement point and the positive z-axis of the array reference frame, nmNoise for the mth measurement point;
step 2: and (3) forming the array excitation and the measurement data of M points in the step 1 into a column vector:and zn=[zn(θ1),zn(θ2),…,zn(θM)]TExcitation vectors and patterns for the nth measurement, respectively, where T represents transposition;
constructing array failure indication vector and failure array element number label k according to excitation vectornWherein, ifThen theIf not, then the mobile phone is started,knis equal to the indicator vector ynThe number of the (1) in the (1),
and step 3: forming a training set by the data obtained in the step 2 (z)1,k1),(z2,k2),...,(zN,kN) Since z is a complex number, it is represented as zn=[Re(zn),Im(zn)]Wherein Re (·) represents taking the real part, and Im (·) represents taking the imaginary part; respectively carrying out standardization processing on a real part and an imaginary part of the radiation data, wherein the standardization formula is as follows:
wherein, mumIs the mean, σ, of the m-dimensional radiation datamVariance of the mth dimension radiation data;
and 4, step 4: training a single-label neural network f for learning the mapping relation between far-field measurement data and the number of failure array elements by using the training set1:kn=f1([Re(zn),Im(zn)]T) (ii) a By means of neural networks f1The output result of (2) and the training set obtained in step (z)1,y1),(z2,y2),...,(zN,yN) Training a multi-label neural network f for learning the mapping relation between far-field measurement data and the positions of failed array elements2:yn=f2([Re(zn),Im(zn)]T) (ii) a Neural network f1And f2Respectively taking the multi-class logarithmic cross entropy and the binary-class logarithmic cross entropy as loss functions;
and (3) a model application stage:
and 5: for the test array, inIn the model training phase, far-field radiation data acquisition is performed on the same M measurement points in step 1, and test data z ═ z (θ (theta)) can be obtained1),z(θ2),...,z(θM)]TThe data is input into the neural network f trained in step 4 after being standardized1In, the number of predicted failure array elements is
Step 6: inputting the test data in the step 5 into the neural network f trained in the step 42In the method, array element failure probability vector p ═ p is obtained1,p2,...,pL]Wherein p isi,i∈[1,L]Indicating the possibility of the failure of the ith array element; setting u and v as superscripts of a maximum value and a second maximum value in the probability vector p respectively and considering that one array element in u and v is invalid;
and 7: the test data z ═ z (θ) in step 5 is compensated separately using u and v1),z(θ2),...,z(θM)]T;For gaussian white noise, the m-th test data after compensation are respectively:
and 8: order toThe two groups of test data compensated in the step 7 are input into the neural network f after being standardized to replace the original test data2In (1), two probability vectors p are obtained[u]=[p[u] 1,p[u] 2,...,p[u] L]And p[v]=[p[v] 1,p[v] 2,...,p[v] L]If p is[u]Maximum value greater than p[v]If the maximum value is less than the maximum value, judging that the u-th array element is invalid, and updating u and v; otherwise, judging that the v-th array element is invalid, and updating u and v;
and step 9: repeating the steps 7-8 untilThe invention discloses an antenna array fault diagnosis method based on deep learning. Firstly, in a model training stage, two diagnostic networks are trained by using data acquired in a far field area under various failure scenarios. And then, in the actual application stage, acquiring data at the same sampling point as the training stage by using a probe, inputting the data into a trained model, preliminarily determining the number and the position of the failure array elements, circularly compensating the measurement data according to a data compensation formula, and sequentially judging the positions of the failure array elements. Compared with the traditional antenna diagnosis algorithm, the algorithm has the advantage of high diagnosis speed in the practical application stage. Compared with the existing array fault diagnosis method based on the parameter model, the algorithm effectively reduces the sample data required to be collected in the training stage through a data compensation method, and is suitable for large-scale array fault diagnosis.
Drawings
FIG. 1 is a block flow diagram of the present algorithm;
FIG. 2 shows the diagnosis result of the position of the failed array element of the failure array under different numbers of the failed array elements when the SNR is equal to 30 dB;
fig. 3 shows the diagnosis result of the position of the fault array element under different signal-to-noise ratios.
Detailed Description
The fault array considered in this embodiment is a uniform linear array comprising 50 array elements, with an array element spacing of 0.5 λ, where λ is the wavelength. In practice, the number of failed array elements is usually small, and therefore, it is assumed that the number of failed array elements in the array is at most 7.
A model training stage:
step 1: the embodiment is in pitchAngle theta e [ -90 DEG, 90 DEG]The amplitude and phase data of the radiation field of the fault array are measured at 2 ° intervals, and each sample data contains 91 measurement points. Working array element excitation x i1, failing array element excitation xiAnd (0) in the far field region, under the condition of determining the number of array element failures, respectively measuring the radiation data in the random failure scene of 700 times of array element positions and carrying out standardization processing.
Step 2: and (3) forming eight training sets according to the number of the invalid array elements by using all the measurement data z [ Re (z), im (z) ] in the step 1, wherein each training set comprises 700 training samples, and sequentially constructing an array invalid indication vector y and an invalid array element number label k of the sample according to the sample excitation x.
And step 3: and (4) carrying out standardization processing on the data characteristics of the training set in the step 2 by using a formula (3).
And 4, step 4: combining and randomly ordering the samples in the eight training sets in the step 3 intoThen dividing the training data into two training setsAndrespectively for training the neural network f1And f2Wherein f is1And f2The network hyper-parameters are shown in table 1.
And (3) a model application stage:
and 5: aiming at 2400 random failure scenes of the AUT (300 tests under the condition of determining the number of failure array elements), far-field radiation data acquisition is carried out on 91 measurement points which are the same as those in the step 1 of the model training stage, and test data Z ═ Z can be obtained1,z2,...,z2400]TThe data is input into the neural network f trained in step 3 after being standardized1In, the number of output failure array elements is
Step 6: inputting the test data in the step 5 into the neural network f trained in the step 32In, output array element failure probability matrix P ═ P1,p2,…p2400]TWherein p isijIndicating the possibility of failure of the jth array element in the ith test sample. ByLet uiAnd viRespectively the ith test sample probability vector piThe upper and lower maximum values are labeled and are designated as uiAnd viIf one array element fails, u-u is determined for all test samples1,u2,…u2400]TAnd v ═ v1,v2,…v2400]T。
And 7: the test data Z in step 5 are compensated separately using u and v.And if the signal is Gaussian white noise, the m test data of the compensated i sample are respectively as follows:
and 8: order toStandardizing the two groups of test data compensated in the step 7 and inputting the standardized test data to the neural network f instead of the original test data2In (1), two probability matrices P are obtained[u]=[p1[u],p2[u],…,p2400[u]]TAnd P[v]=[p1[v],p2[v],…,p2400[v]]T. For the ith sample, if pi[u]Maximum value of greater than pi[v]Maximum value of (1), uiUpdating u, v when each array element fails; otherwise, viAnd (5) failing each array element, and updating u and v.
And step 9: repeating the steps 7-8 untilThe antenna array fault diagnosis algorithm based on deep learning provided by the invention is applied to a uniform linear array, the array comprises 50 array elements in total, the array element spacing is 0.5 lambda, and lambda is the wavelength. 8000 array samples were randomly generated, of which 5600 were used for training the neural network and 2400 for diagnostic testing. In the far field region, the pitch angle theta is ∈ [ -90 DEG, 90 DEG]In the range of (2), 91 measurement field points are sampled at 2 ° intervals. The noise added is complex white gaussian noise that is expected to be zero, and the signal-to-noise ratio SNR is 30 dB. The result of the accuracy of the diagnosis of the position of the failed array element of 2400 test samples is shown in fig. 2, and the above experimental process is repeated under the conditions of different signal to noise ratios, and the diagnosis result is shown in fig. 3.
Table 1: neural network f1And f2Is a hyper-parameter of
Claims (1)
1. A method for diagnosing faults of an antenna array based on a deep neural network and radiation data compensation comprises the following steps:
a model training stage:
step 1: measuring electric field radiation data of a far field of the array under N fault scenes by using a probe at M measuring points with different angles on a far field measuring plane; the data measured by the probe at the m-th angle is expressed by the following formula:
where c is the speed of light, f is the working frequency of the array, L is the total number of array elements contained in the array, xiAnd diExcitation of the ith array element and position in the array coordinate system, thetamIs the angle between the m-th measurement point and the positive z-axis of the array reference frame, nmNoise for the mth measurement point;
step 2: and (3) forming the array excitation and the measurement data of M points in the step 1 into a column vector:and zn=[zn(θ1),zn(θ2),...,zn(θM)]TExcitation vectors and patterns for the nth measurement, respectively, where T represents transposition;
constructing array failure indication vector and failure array element number label k according to excitation vectornWherein, ifThenIf not, then,knis equal to the indicator vector ynThe number of the (1) in the (1),
and step 3: forming a training set by the data obtained in the step 2 (z)1,k1),(z2,k2),...,(zN,kN) Since z is a complex number, it is represented as zn=[Re(zn),Im(zn)]Wherein Re (·) represents taking the real part, and Im (·) represents taking the imaginary part; separately normalizing the real and imaginary components of radiation dataProcessing, the normalized formula is:
wherein, mumIs the mean, σ, of the m-dimensional radiation datamVariance of the mth dimension radiation data;
and 4, step 4: training a single-label neural network f for learning the mapping relation between far-field measurement data and the number of failure array elements by using the training set1:kn=f1([Re(zn),Im(zn)]T) (ii) a By means of neural networks f1The output result of (2) and the training set obtained in step (z)1,y1),(z2,y2),...,(zN,yN) Training a multi-label neural network f for learning the mapping relation between far-field measurement data and the positions of failed array elements2:yn=f2([Re(zn),Im(zn)]T) (ii) a Neural network f1And f2Respectively taking the multi-class logarithmic cross entropy and the binary-class logarithmic cross entropy as loss functions;
and (3) a model application stage:
and 5: for the test array, far-field radiation data acquisition is performed on the same M measurement points as in step 1 of the model training phase, and test data z ═ z (θ) can be obtained1),z(θ2),...,z(θM)]TThe data is input into the neural network f trained in step 4 after being standardized1In, the number of predicted failure array elements is
Step 6: inputting the test data in the step 5 into the neural network f trained in the step 42In the method, an array element failure probability vector p ═ p is obtained1,p2,...,pL]Wherein p isi,i∈[1,L]Indicating the possibility of the failure of the ith array element; let u and v be in the probability vector p, respectivelyThe superscript of the maximum value and the second maximum value considers that one array element in u and v is invalid;
and 7: the test data z ═ z (θ) in step 5 is compensated separately using u and v1),z(θ2),...,z(θM)]T;And the m-th test data after compensation is white gaussian noise, respectively:
and 8: order toThe two groups of test data compensated in the step 7 are input into the neural network f after being standardized to replace the original test data2In (1), two probability vectors p are obtained[u]=[p[u] 1,p[u] 2,...,p[u] L]And p[v]=[p[v] 1,p[v] 2,...,p[v] L]If p is[u]Maximum value greater than p[v]If the maximum value is less than the maximum value, judging that the u-th array element is invalid, and updating u and v; otherwise, judging that the v-th array element is invalid, and updating u and v;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010259285.6A CN111523568B (en) | 2020-04-03 | 2020-04-03 | Antenna array fault diagnosis method based on deep neural network and radiation data compensation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010259285.6A CN111523568B (en) | 2020-04-03 | 2020-04-03 | Antenna array fault diagnosis method based on deep neural network and radiation data compensation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111523568A CN111523568A (en) | 2020-08-11 |
CN111523568B true CN111523568B (en) | 2022-05-03 |
Family
ID=71902283
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010259285.6A Expired - Fee Related CN111523568B (en) | 2020-04-03 | 2020-04-03 | Antenna array fault diagnosis method based on deep neural network and radiation data compensation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111523568B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FI129476B (en) | 2020-12-29 | 2022-03-15 | Sidus Innovations Oy | Antenna radiation pattern extraction using sparse field measurements |
CN113552450A (en) * | 2021-07-07 | 2021-10-26 | 上海交通大学 | Partial discharge source positioning method based on virtual sensor |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104579511A (en) * | 2015-01-05 | 2015-04-29 | 电子科技大学 | Method for obtaining array antenna receiving signal power error range |
CN107015066A (en) * | 2017-03-27 | 2017-08-04 | 电子科技大学 | A kind of aerial array method for diagnosing faults based on management loading |
CN108446503A (en) * | 2018-03-22 | 2018-08-24 | 电子科技大学 | Near field bare cloth aerial array optimization method based on compressed sensing Yu convex optimization |
CN108932381A (en) * | 2018-06-27 | 2018-12-04 | 电子科技大学 | A kind of aerial array method for diagnosing faults considering array error |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10243276B2 (en) * | 2015-10-12 | 2019-03-26 | The Boeing Company | Phased array antenna system including a modular control and monitoring architecture |
-
2020
- 2020-04-03 CN CN202010259285.6A patent/CN111523568B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104579511A (en) * | 2015-01-05 | 2015-04-29 | 电子科技大学 | Method for obtaining array antenna receiving signal power error range |
CN107015066A (en) * | 2017-03-27 | 2017-08-04 | 电子科技大学 | A kind of aerial array method for diagnosing faults based on management loading |
CN108446503A (en) * | 2018-03-22 | 2018-08-24 | 电子科技大学 | Near field bare cloth aerial array optimization method based on compressed sensing Yu convex optimization |
CN108932381A (en) * | 2018-06-27 | 2018-12-04 | 电子科技大学 | A kind of aerial array method for diagnosing faults considering array error |
Non-Patent Citations (5)
Title |
---|
《A Fault Diagnosis Algorithm of Antenna Arrays in the Presence of Array Mismatch》;Ying Zhang等;《 2019 IEEE International Conference on Computational Electromagnetics (ICCEM)》;20190729;第1-3页 * |
《利用一维像序列时域差分估计目标进动频率》;周代英等;《航空学报》;20181231;第39卷(第S1期);第69-74页 * |
《宽带水声参量阵指向性设计与实验验证》;张富东等;《声学技术》;20110430;第30卷(第2期);第140-143页 * |
《电离层环境对地空信息系统的影响效应》;冯健等;《装备环境工程》;20170731;第14卷(第7期);第7-11页 * |
《统计模式识别在相控阵天线故障诊断中的应用研究》;韦哲等;《信号处理》;20140831;第30卷(第8期);第987-992页 * |
Also Published As
Publication number | Publication date |
---|---|
CN111523568A (en) | 2020-08-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111523568B (en) | Antenna array fault diagnosis method based on deep neural network and radiation data compensation | |
CN105277826B (en) | Array antenna fault diagnosis system | |
CN108932381B (en) | Antenna array fault diagnosis method considering array errors | |
CN109991519B (en) | Partial discharge direction-finding method and system based on neural network and wireless sensor array | |
CN111132007A (en) | NB terminal high-precision positioning algorithm based on 5G/NB-IoT cluster node information fusion | |
CN113835063B (en) | Unmanned aerial vehicle array amplitude and phase error and signal DOA joint estimation method | |
CN113759303A (en) | Non-grid DOA (angle of arrival) estimation method based on particle swarm optimization | |
CN117031509A (en) | Soil humidity inversion method and device integrating isolated forest and deep learning | |
RU2744320C1 (en) | Method and system for testing of antenna with multiple radiation elements | |
CN109839543B (en) | System and method for testing amplitude-phase consistency of antenna | |
CN115561530A (en) | Antenna array fault diagnosis method based on data tolerance analysis extension | |
Kuznetsov et al. | Modifications of active phased antenna arrays near-field diagnosis method based on compressive sensing | |
Sun et al. | A fast failure diagnosis method based on structure features of planar arrays | |
US11899098B2 (en) | System and method for testing a radar under test | |
CN110018460B (en) | Far field measurement method for phase difference between whole satellite phases of satellite-borne synthetic aperture radar | |
Salucci et al. | Antenna measurements by design: A novel paradigm for antenna testing | |
CN116990746B (en) | Direction finding system and method for radio monitoring | |
Chi et al. | a New Distributed Target Extraction Method for Polarimetric SAR Calibration | |
CN115017936A (en) | Phased array radar array element fault diagnosis method based on double-current DCNN | |
Li et al. | Shifted base mode character of array antenna with failed elements | |
CN114994627B (en) | Direction finding method based on deep neural network and subspace principle | |
CN116008679B (en) | Near-field amplitude-phase data extraction circuit and near-field amplitude-phase data extraction method | |
RU2798753C1 (en) | Method for diagnosing an antenna array | |
Güngör et al. | Hardware-Efficient Direction of Arrival Estimation using Compressive Sensing | |
Xiong et al. | Diagnose element failures of phased antenna arrays with a single fixed receiving probe |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220503 |