CN114384487A - XGboost-based weather radar ground object echo identification method - Google Patents

XGboost-based weather radar ground object echo identification method Download PDF

Info

Publication number
CN114384487A
CN114384487A CN202111596988.9A CN202111596988A CN114384487A CN 114384487 A CN114384487 A CN 114384487A CN 202111596988 A CN202111596988 A CN 202111596988A CN 114384487 A CN114384487 A CN 114384487A
Authority
CN
China
Prior art keywords
xgboost
model
radar
optimization
ground object
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
Application number
CN202111596988.9A
Other languages
Chinese (zh)
Inventor
张舒娜
汪玲
朱岱寅
周晔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN202111596988.9A priority Critical patent/CN114384487A/en
Publication of CN114384487A publication Critical patent/CN114384487A/en
Pending legal-status Critical Current

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
    • 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
    • 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
    • G01S13/95—Radar or analogous systems specially adapted for specific applications for meteorological use
    • 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
    • G01S7/411—Identification of targets based on measurements of radar reflectivity
    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/29—Graphical models, e.g. Bayesian networks
    • 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)
  • Physics & Mathematics (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Electromagnetism (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a meteorological radar ground object echo identification method based on XGboost, which comprises the following steps: s1, selecting radar parameters input into the XGboost model, wherein the radar parameters comprise reflectivity factors, reflectivity factor textures, differential reflectivity factors, variation degree of echo intensity along the radial direction, regional average value of radial speed and correlation coefficients; step S2, initializing the XGboost model, and determining the parameters of the initial XGboost model; step S3, inputting a meteorological radar measured data sample set, and dividing the meteorological radar measured data sample set into a training set and a testing set; step S4, carrying out Bayesian optimization on the initial XGboost model by using a training set to obtain a final XGboost optimization model; and step S5, inputting the test set into the XGboost optimization model, performing iteration, and stopping iteration when the maximum iteration times is reached or the loss function of the XGboost optimization model reaches the optimal solution of the target function, so as to finish the final identification classification. The invention is suitable for various weather conditions and can more accurately identify the ground object echo.

Description

XGboost-based weather radar ground object echo identification method
Technical Field
The invention relates to the technical field of meteorological radar signal processing, in particular to a meteorological radar ground object echo identification method based on XGboost.
Background
The weather radar is an important means for weather detection, is widely applied to observation of weather phenomena such as cloud, rain, snow, thunderstorm, turbulence and the like, and analyzes a weather target through echo information. The variety of the disaster weather is numerous and frequent, high economic loss is easily caused, and the application of the weather radar to the research work of monitoring and forecasting the disaster weather has very important significance. The echoes obtained by radar detection contain a considerable part of ground object echoes, and if the ground object echoes are not identified, the quality of radar observation data is reduced, and the detection and the identification of meteorological targets are influenced. Therefore, in order to acquire accurate radar data and guarantee the quality of radar products, the research on the identification method of the ground object echo is very important.
Kessinger provides a radar echo classification algorithm, characteristic parameters are calculated through radar basic parameters for the first time, and weather echoes are identified based on the idea of fuzzy logic. Cluckie et al propose a method of combining a Bayesian recognition algorithm with fuzzy logic to perform recognition classification. On the basis of the fuzzy logic method proposed by Kessinger, Liuliping and the like propose a method for recognizing ground objects in a distributed mode, the same characteristic parameters of the Kessinger proposal method are adopted, and the echo recognition effect of the ground objects is improved. The banksia rose et al improve the problem of over-suppression by aiming at the Liuliping method, add the function of echo filling before processing the echo, change the ground feature discrimination threshold value from a fixed value into a function changing along with the distance, and reduce the over-suppression of the echo of the ground feature at a long distance. The methods are characterized in that a criterion or a discrimination factor which can distinguish meteorological targets from non-meteorological targets is searched for and is used as an input parameter of a fuzzy logic method, a large number of empirical values are relied on, membership functions in the method cannot be automatically adjusted, the adaptability to weather conditions is low, and the method has no universality.
Disclosure of Invention
The invention aims to solve the technical problem of overcoming the defects of the prior art and provides a meteorological radar ground object echo identification method based on XGboost.
The invention adopts the following technical scheme for solving the technical problems:
the invention provides a meteorological radar ground object echo identification method based on XGboost, which comprises the following steps:
s1, selecting radar parameters input into the XGboost model, wherein the radar parameters comprise reflectivity factors, reflectivity factor textures, differential reflectivity factors, variation degree of echo intensity along the radial direction, regional average value of radial speed and correlation coefficients;
step S2, initializing the XGboost model, and determining the parameters of the initial XGboost model;
step S3, inputting a meteorological radar measured data sample set [ x ]i,yi]Wherein x isiIs the ith meteorological radar parameter, and the magnitude is a vector in m dimensions, yiDividing a meteorological radar actual measurement data sample set into a training set and a testing set for the tag of the ith meteorological data sample;
step S4, carrying out Bayesian optimization on the initial XGboost model by using a training set to obtain a final XGboost optimization model;
and step S5, inputting the test set into the XGboost optimization model, performing iteration, and stopping iteration when the maximum iteration times is reached or the loss function of the XGboost optimization model reaches the optimal solution of the target function, so as to finish the final identification classification.
As a further optimization scheme of the method for identifying the weather radar ground object echo based on the XGBoost, the bayesian optimization in step S4 is specifically as follows:
step S41, initializing a Bayes optimization model, inputting parameters to be optimized in the XGboost model into the Bayes model, initializing a population P (0), and enabling an evolution algebra t to be 0;
step S42, selecting a candidate solution S (t) of the t iteration from the population P (t) of the t iteration;
step S43, establishing a bayesian network B (μ, σ) from the solution candidates S (t), the formula being:
Figure BDA0003431565180000021
wherein x istSampling position for next XGboost model
Figure BDA0003431565180000022
As a weight parameter, mut-1(x) In order to predict the mean value of the mean,
Figure BDA0003431565180000023
the variance is obtained, x is a super parameter of a meteorological radar ground object echo identification XGboost model, and D is a super parameter data set of the meteorological radar ground object echo identification XGboost model;
step S44, generating a new solution O (t) according to the joint distribution function of the Bayesian network B (mu, sigma);
step S45, replacing partial solutions in p (t) with o (t), forming a new population, making t ═ t +1, returning to step S42 until x is foundtAnd obtaining a final XGboost optimization model.
As a further optimization scheme of the method for identifying the weather radar ground object echo based on the XGBoost, in step S2, the parameters are set as: the maximum tree depth Max _ depth is 4, the child node minimum sample weight min _ child _ weight is 1, the minimum loss reduction amount gamma is 0, the learning rate eta is 0.1, the regularization parameter lamda is 0, and aplha is 0.
As a further optimization scheme of the XGboost-based meteorological radar ground object echo identification method, in step S4, a Bayesian optimization method is applied to optimize an XGboost model; the Bayesian optimization method comprises the steps of firstly, randomly selecting an initial population and a candidate solution thereof by using an evolutionary algorithm, constructing a new Bayesian network model, updating the candidate solution, substituting the updated candidate solution into the original population, and repeating the process until a function maximum value is found.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
(1) compared with the traditional fuzzy logic method, the method can adaptively train data, adapt to various weather conditions and more accurately identify the ground object echo;
(2) the XGboost network model is optimized by using a Bayesian optimization algorithm without manual parameter adjustment, and the recognition rate is higher than that of other optimization algorithms such as grid search, random search and the like;
(3) in addition, the method provided by the invention has the advantages that the performance of identifying the ground object echo by using the discrete meteorological radar parameters is better, the learning is more excellent compared with other machines, and the ground object echo identification rate is higher.
Drawings
FIG. 1 is a framework diagram of the XGBoost tree model.
Fig. 2 is an overall algorithm flow diagram.
FIG. 3 is a network training process for a data set.
FIG. 4 is a PPI diagram of the feature echo identification result of the KVTX radar in the United states; wherein, (a) is a classification result reference diagram given by a hydrogel classification algorithm used by a WDR-88D radar in the United states, and (b) is a terrain echo and meteorological target classification diagram completed by an XGboost algorithm.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
Fig. 1 is a framework diagram of the XGboost tree model, and fig. 2 is an overall algorithm flow diagram. The method comprises the following steps:
s1, selecting radar parameters input into the XGboost model, wherein the radar parameters comprise reflectivity factors, reflectivity factor textures, differential reflectivity factors, variation degree of echo intensity along the radial direction, regional average value of radial speed and correlation coefficients;
step S2, initializing the XGboost model, and determining the parameters of the initial XGboost model;
step S3, inputting a meteorological radar measured data sample set [ x ]i,yi]Wherein x isiIs the ith meteorological radar parameter, and the magnitude is a vector in m dimensions, yiDividing a meteorological radar actual measurement data sample set into a training set and a testing set for the tag of the ith meteorological data sample;
step S4, carrying out Bayesian optimization on the initial XGboost model by using a training set to obtain a final XGboost optimization model;
and step S5, inputting the test set into the XGboost optimization model, performing iteration, and stopping iteration when the maximum iteration times is reached or the loss function of the XGboost optimization model reaches the optimal solution of the target function, so as to finish the final identification classification.
The bayesian optimization in step S4 is specifically as follows:
step S41, initializing a Bayes optimization model, inputting parameters needing to be optimized in XGboost into the Bayes model, initializing a population P (0), and making an evolution algebra t equal to 0;
step S42, selecting a candidate solution S (t) of the t iteration from the population P (t) of the t iteration;
step S43, establishing a bayesian network B (μ, σ) from the solution candidates S (t), the formula being:
Figure BDA0003431565180000041
wherein x istSampling locations for the next model
Figure BDA0003431565180000042
As a weight parameter, mut-1(x) In order to predict the mean value of the mean,
Figure BDA0003431565180000043
the variance is obtained, x is a super parameter of a meteorological radar ground object echo identification XGboost model, and D is a super parameter data set of the meteorological radar ground object echo identification XGboost model;
step S44, generating a new solution O (t) according to the joint distribution function of the Bayesian network B (mu, sigma);
step S45, replacing partial solutions in p (t) with o (t), forming a new population, making t ═ t +1, returning to step S42 until x is foundtAnd obtaining a final XGboost optimization model.
The method applies three basic parameter reflectivity factors, differential reflectivity factors and correlation coefficients of the dual-polarization radar, and three characteristic parameter reflectivity factor textures, the radial variation degree of the reflectivity factors and the area average value of radial speed, and comprises the following specific contents:
2 characteristic parameters are extracted from the reflectivity factor: reflectivity factor texture TDBZDegree of change S of the reflectivity factor in the radial directionPINThe definition is as follows:
Figure BDA0003431565180000044
Figure BDA0003431565180000045
Figure BDA0003431565180000046
2 characteristic parameters are extracted from the radial velocity: regional average value M of radial velocityDVEVariance of radial velocity SDVEThe parameters are defined as follows:
MDVE＝med{VEi,j,i∈Nbeams,j∈Ngates} (3)
wherein N isgates、NbeamsRepresenting the number of points of computation defined in the distance and azimuth directions, Zi,jEcho intensity, T, at coordinate point (i, j)DBZThe variance between the adjacent distance libraries of the echo intensity mainly reflects the change of the echo intensity within a certain range; zlowAnd ZupEcho intensities scanned for the corresponding low and high elevation PPI; zthreshFor the threshold value of the change of the adjacent echo intensity, 2-5dB, S can be takenPINThe consistency of the change of the echo intensity along the radial direction is reflected; mDVEAnd representing the radial velocity value of any point subjected to the median filtering processing.
The classification method based on the XGboost provided by the invention constructs a target identification network architecture taking radar characteristic parameters as input, and comprises the following specific contents:
when the XGboost is used for classification, model training is carried out, and in order to enable the interval of the predicted values of the classification problem result to be between [0 and 1], the XGboost introduces a log (odds) function, similar to logistic regression:
Figure BDA0003431565180000051
converting the predicted value into log (odds) when the default initial predicted value is set
Figure BDA0003431565180000052
At 0.5, log (odds) is 0, and in Gradient Boosting we use the loss function as:
L(yi,pi)＝-[yilog(pi)+(1-yi)log(1-pi)] (7)
the loss function for a single sample of XGBoost becomes:
L(yi,log(odds))＝-yilog(odds)+log(1+elog(odds)) (8)
the first derivative and the second derivative are respectively:
Figure BDA0003431565180000053
in the classification model, the weight O is predictedjComprises the following steps:
Figure BDA0003431565180000054
in the formula, we can
Figure BDA0003431565180000055
Viewed as the sum of residuals, pi×(1-pi) Then the product of the predicted value and 1-predicted value, then the weight in the classification model, OvalueCan be converted into:
Figure BDA0003431565180000056
the objective function Obj is:
Figure BDA0003431565180000061
the predicted values for the current tree are:
Figure BDA0003431565180000062
the sample set selected in step S3 includes the following specific contents:
the classification targets are ground object echo and meteorological targets and are classified into two categories. Selecting KVTX radar data of the United states as a data set, wherein the data set comprises 20107 samples, the number of non-meteorological target data is 6028, the number of meteorological target data is 14079, and the data are divided into a training set and a testing set according to the ratio of 8: 2. Since it belongs to the case of two classes and we are more concerned with the classification of the clutter, non-meteorological targets and meteorological targets are labeled as positive and negative classes, 1 and 0, respectively. The data is pre-processed before being processed by the algorithm.
First is default value processing. All samples are counted, and the 6 input parameters share bar 1222 default data, where the greatest default is the regional average M of radial velocityDVEWhereas the parameter related to the echo intensity has no default value. Since the number of default values is small in the total number of samples, the default values are directly removed, and the default values are simpler and do not need to be specially processed. So 18885 samples remain finally, of which 13508 meteorological object data and 5377 non-meteorological objects.
Then a normalization process follows. According to different observation times or weather conditions, the data may have a large difference, in algorithm learning, when the difference of different characteristic values is large, a prediction direction may be wrong, and in order to reduce the influence of the difference of different characteristic values on training, all data are normalized. Normalization is performed herein using a linear function:
Figure BDA0003431565180000063
in the formula, XmaxAnd XminRespectively, a maximum value and a minimum value for each characteristic parameter.
Substituting the training set into the initialized model in the step S4 to train to obtain an optimal training model, and stopping training when the optimal solution appears or the training frequency reaches the highest value; and respectively carrying out classification and identification on the test set by using the trained learning model.
Examples of the embodiments
In order to verify the effectiveness of the XGboost-based classification algorithm, the data of the United states X-band KVTX meteorological radar is adopted for experimental verification. In addition, in order to illustrate the advantages of the method for identifying the weather radar ground object echo based on the XGboost, the accuracy of various optimization models is firstly compared, and then the target identification result obtained by the method is compared with other machine learning identification results.
Fig. 3 is a network training process of a data set, a solid line is a training set loss function curve, a dotted line is a test set loss function curve, and a red point is a test set loss function minimum point, i.e., an optimal solution of the entire model.
FIG. 4 is a PPI map of terrain echo identification. Comparing the recognition result graph in fig. 4 (b) with the reference result in fig. 4 (a), the recognition effect at the short distance of the radar is good, the recognition effect is basically consistent with the reference classification, and the overall recognition effect is good under the condition that the weather echo is misjudged in the long-distance southwest direction of the radar.
TABLE 1 XGboost different optimization method identification accuracy
Rate of accuracy
Initial model 0.9370
Post-grid search model 0.9423
Bayesian optimized post-model 0.9632
Table 1 shows comparison of recognition rates of different optimization methods of XGBoost, where the accuracy of the initial model is the lowest under the default parameters, because the learning rate is high, the grid search needs manual parameter adjustment to reduce the learning rate, and the obtained model needs to be subjected to multiple experiments to find out the optimal solution. The Bayes method selects the optimal solution in a self-adaptive manner through an optimization function, the selection result is more accurate, and the recognition rate is higher than that of the other two models, so that Bayes optimization is selected as the optimization algorithm of the XGboost model.
TABLE 2 confusion matrix of surface feature echo and meteorological target identification results
Figure BDA0003431565180000071
Table 2 shows the finally obtained two-class confusion matrix, and it can be seen that the overall recognition rate is 95.99%, wherein the recognition rate of the ground object echo is 94.92%, the recognition rate of the weather echo is 96.43%, and the recognition accuracy of the ground object echo is lower than that of the weather echo. In addition, the ground feature echo recognition result based on the XGboost is compared with the target recognition results of other methods, as shown in table 3, the target recognition rates of different methods are given in table 3, and as can be seen from table 3, the classification accuracy of the XGboost network model optimized by Bayes is slightly superior to that of other traditional machine learning methods, and meanwhile, the performance of the method on discrete feature training is higher than that of a neural network model, and overfitting can be well prevented.
TABLE 3 comparison of recognition rates for different methods
Name of method Rate of accuracy
RF model 0.9532
SVM model 0.9247
ANN model 0.9393
XGboost model 0.9599
The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.

Claims (4)

1. A weather radar ground object echo identification method based on XGboost is characterized by comprising the following steps:
s1, selecting radar parameters input into the XGboost model, wherein the radar parameters comprise reflectivity factors, reflectivity factor textures, differential reflectivity factors, variation degree of echo intensity along the radial direction, regional average value of radial speed and correlation coefficients;
step S2, initializing the XGboost model, and determining the parameters of the initial XGboost model;
step S3, inputting a meteorological radar measured data sample set [ x ]i,yi]Wherein x isiIs the ith meteorological radar parameter, and the magnitude is a vector in m dimensions, yiDividing a meteorological radar actual measurement data sample set into a training set and a testing set for the tag of the ith meteorological data sample;
step S4, carrying out Bayesian optimization on the initial XGboost model by using a training set to obtain a final XGboost optimization model;
and step S5, inputting the test set into the XGboost optimization model, performing iteration, and stopping iteration when the maximum iteration times is reached or the loss function of the XGboost optimization model reaches the optimal solution of the target function, so as to finish the final identification classification.
2. The method for identifying the weather radar ground object echo based on the XGboost as claimed in claim 1, wherein the Bayesian optimization in the step S4 is as follows:
step S41, initializing a Bayes optimization model, inputting parameters to be optimized in the XGboost model into the Bayes model, initializing a population P (0), and enabling an evolution algebra t to be 0;
step S42, selecting a candidate solution S (t) of the t iteration from the population P (t) of the t iteration;
step S43, establishing a bayesian network B (μ, σ) from the solution candidates S (t), the formula being:
Figure FDA0003431565170000011
wherein x istSampling position for next XGboost model
Figure FDA0003431565170000012
As a weight parameter, mut-1(x) In order to predict the mean value of the mean,
Figure FDA0003431565170000013
the variance is obtained, x is a super parameter of a meteorological radar ground object echo identification XGboost model, and D is a super parameter data set of the meteorological radar ground object echo identification XGboost model;
step S44, generating a new solution O (t) according to the joint distribution function of the Bayesian network B (mu, sigma);
step S45, replacing partial solutions in p (t) with o (t), forming a new population, making t ═ t +1, returning to step S42 until x is foundtAnd obtaining a final XGboost optimization model.
3. The XGboost-based weather radar terrain echo identification method of claim 1, wherein in step S2, the parameters are set as: the maximum tree depth Max _ depth is 4, the child node minimum sample weight min _ child _ weight is 1, the minimum loss reduction amount gamma is 0, the learning rate eta is 0.1, the regularization parameter lamda is 0, and aplha is 0.
4. The method for identifying the weather radar ground object echo based on the XGboost according to the claim 1, wherein in the step S4, a Bayesian optimization method is used for optimizing an XGboost model; the Bayesian optimization method comprises the steps of firstly, randomly selecting an initial population and a candidate solution thereof by using an evolutionary algorithm, constructing a new Bayesian network model, updating the candidate solution, substituting the updated candidate solution into the original population, and repeating the process until a function maximum value is found.
CN202111596988.9A 2021-12-24 2021-12-24 XGboost-based weather radar ground object echo identification method Pending CN114384487A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111596988.9A CN114384487A (en) 2021-12-24 2021-12-24 XGboost-based weather radar ground object echo identification method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111596988.9A CN114384487A (en) 2021-12-24 2021-12-24 XGboost-based weather radar ground object echo identification method

Publications (1)

Publication Number Publication Date
CN114384487A true CN114384487A (en) 2022-04-22

Family

ID=81196978

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111596988.9A Pending CN114384487A (en) 2021-12-24 2021-12-24 XGboost-based weather radar ground object echo identification method

Country Status (1)

Country Link
CN (1) CN114384487A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118033548A (en) * 2024-04-12 2024-05-14 成都远望科技有限责任公司 Dual-transmitting dual-receiving top-sweeping cloud radar same-frequency interference identification method and device

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118033548A (en) * 2024-04-12 2024-05-14 成都远望科技有限责任公司 Dual-transmitting dual-receiving top-sweeping cloud radar same-frequency interference identification method and device

Similar Documents

Publication Publication Date Title
Sharifzadeh et al. Ship classification in SAR images using a new hybrid CNN–MLP classifier
CN112001270B (en) Ground radar automatic target classification and identification method based on one-dimensional convolutional neural network
CN110210463A (en) Radar target image detecting method based on Precise ROI-Faster R-CNN
CN109871902B (en) SAR small sample identification method based on super-resolution countermeasure generation cascade network
Copsey et al. Bayesian gamma mixture model approach to radar target recognition
CN113866742B (en) Method for point cloud processing and target classification of 4D millimeter wave radar
CN112434643A (en) Classification and identification method for low-slow small targets
Lakshmanan et al. Which polarimetric variables are important for weather/no-weather discrimination?
CN113392931B (en) Hyperspectral open set classification method based on self-supervision learning and multitask learning
Singh et al. A deep learning based approach with adversarial regularization for Doppler weather radar ECHO prediction
CN113609895A (en) Road traffic information acquisition method based on improved Yolov3
CN114384487A (en) XGboost-based weather radar ground object echo identification method
CN115393631A (en) Hyperspectral image classification method based on Bayesian layer graph convolution neural network
CN109784142B (en) Hyperspectral target detection method based on conditional random projection
CN111368653A (en) Low-altitude small target detection method based on R-D (R-D) graph and deep neural network
Han et al. Automatic binary and ternary change detection in SAR images based on evolutionary multiobjective optimization
CN115310559A (en) Target identification method based on Bayes-Transformer neural network model
CN112949524B (en) Engine fault detection method based on empirical mode decomposition and multi-core learning
CN115542279A (en) Weather radar clutter classification and identification method and device
Liu et al. Fuzzy cluster analysis of spatio-temporal data
Xie et al. Moving Object Detection Algorithm Based on Adaptive Clustering
Hamidi et al. Texture‐based classification of high‐resolution precipitation forecasts with machine‐learning methods
Weekley et al. Aerosol plume detection algorithm based on image segmentation of scanning atmospheric lidar data
CN116908806B (en) Multi-dimensional feature target detection method based on self-supervised learning
CN107942313A (en) Sea clutter optimal soft survey instrument and method based on TSP question drosophila optimization algorithm optimization wavelet neural network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination