CN112098733A - Electromagnetic scattering characteristic data interpolation generation method and device - Google Patents
Electromagnetic scattering characteristic data interpolation generation method and device Download PDFInfo
- Publication number
- CN112098733A CN112098733A CN202011002152.7A CN202011002152A CN112098733A CN 112098733 A CN112098733 A CN 112098733A CN 202011002152 A CN202011002152 A CN 202011002152A CN 112098733 A CN112098733 A CN 112098733A
- Authority
- CN
- China
- Prior art keywords
- electromagnetic scattering
- data
- decision tree
- interpolation
- smoothing
- 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
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000003066 decision tree Methods 0.000 claims abstract description 70
- 238000009499 grossing Methods 0.000 claims abstract description 37
- 238000004138 cluster model Methods 0.000 claims abstract description 29
- 230000007246 mechanism Effects 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims description 32
- 238000012545 processing Methods 0.000 claims description 16
- 238000005457 optimization Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000010355 oscillation Effects 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 230000001174 ascending effect Effects 0.000 claims description 5
- 230000006870 function Effects 0.000 description 70
- 238000013461 design Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010845 search algorithm Methods 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000003534 oscillatory effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
Images
Classifications
-
- 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
- G01R29/0864—Measuring electromagnetic field characteristics characterised by constructional or functional features
- G01R29/0892—Details related to signal analysis or treatment; presenting results, e.g. displays; measuring specific signal features other than field strength, e.g. polarisation, field modes, phase, envelope, maximum value
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Linguistics (AREA)
- Electromagnetism (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides an electromagnetic scattering characteristic data interpolation generation method and device. The method comprises the following steps: acquiring target electromagnetic scattering characteristic data, and performing compilation on the electromagnetic scattering characteristic data according to electromagnetic scattering data space description specifications; smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism; and performing electromagnetic scattering data interpolation generation by using a preset decision tree cluster model. The invention discloses an electromagnetic scattering data interpolation generation method which has high robustness and can generate high-confidence-degree data, and aims to solve the problems that the confidence degree of interpolation data is low and the robustness of the interpolation method is weak in the interpolation generation of target electromagnetic scattering characteristic data.
Description
Technical Field
The invention relates to the technical field of electromagnetic scattering characteristic data generation, in particular to an electromagnetic scattering characteristic data interpolation generation method and device.
Background
The electromagnetic scattering data is the basis for researching the electromagnetic scattering characteristics, and the granularity of the electromagnetic scattering data determines the accuracy and the fineness of the research on the electromagnetic scattering characteristics. At present, the methods for acquiring electromagnetic scattering data mainly include modeling methods and measuring methods. For the modeling method, unknown model parameters, geometric parameters and material parameters are difficult to directly model to obtain detailed electromagnetic scattering characteristic data of a target; for the measurement method, the available electromagnetic scattering property data is limited and coarse in most cases due to the high cost and the limitations of the measurement conditions, and the premise that the target can be fitted.
In the field of electromagnetic scattering property data generation, electromagnetic scattering property data for a target roughness is generally subjected to data granularity refinement by an average value method. For complex targets, a conventional numerical processing method such as a least square method is also used in a targeted manner to perform data granularity refinement. However, the two methods have obvious shortages and restrictions on data interpolation generation to realize data granularity refinement. The average value method is a common method for electromagnetic scattering characteristic data interpolation, has a good interpolation effect on electromagnetic data which change steadily, but is not applicable to the situation that target electromagnetic scattering data is too oscillatory. For the least square method, specific applicable conditions exist, universality is not achieved, and the quality of generated data is difficult to guarantee.
Disclosure of Invention
The invention aims to solve at least part of the problems, and provides an electromagnetic scattering characteristic data interpolation generation method and device, which have high robustness and can generate high-confidence data, so as to solve the problems of low confidence of interpolation data and weak robustness of an interpolation method in the interpolation generation of target electromagnetic scattering characteristic data.
The invention discloses an electromagnetic scattering characteristic data interpolation generation method, which comprises the following steps:
acquiring target electromagnetic scattering characteristic data, and performing compilation on the electromagnetic scattering characteristic data according to electromagnetic scattering data space description specifications;
smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism;
and performing electromagnetic scattering data interpolation generation by using a preset decision tree cluster model.
Preferably, the preset decision tree cluster model obtaining manner includes:
and training the strategy tree cluster model, performing optimization feedback of a target function through a training result, and setting the hyperparameter based on the Bayesian theory to obtain a preset decision tree cluster model.
Preferably, the smoothing the electromagnetic scattering property data according to a preset smoothing mechanism comprises:
and setting a smooth sliding window specification and a sliding window step length according to the oscillation characteristics of the electromagnetic scattering data, and performing smooth processing on the electromagnetic scattering characteristic data according to a data assignment mechanism.
Preferably, the objective function uses a mean square error function as a loss function, L2Regularization is used as an objective function regularization term.
Preferably, the performing of the optimized feedback of the objective function through the training result comprises:
the decision tree objective function Obj (θ) includes a loss function L (θ) and a regularization term Ω (θ), and the expression is:
Obj(θ)=L(θ)+Ω(θ)
wherein theta is a model parameter, L (theta) is a loss function, and omega (theta) is a regularization term;
selecting L2Regularization, the expression is:
Ω(fk)=λ‖fk‖2
wherein f iskIs the kth tree model, lambda is the parameter coefficient to be adjusted, | fkII denotes the pair fkCalculating a norm;
the loss function is a mean square error loss function, and the expression is as follows:
wherein, ytIs the t sample data label value, t isThe number of the samples is the same as the number of the samples,and N is the total number of samples.
Preferably, the setting of the hyper-parameters based on the bayesian theory through the training results comprises:
minimizing a loss function in the regression calculation objective function of the decision tree cluster;
setting a parameter value range and a search step length for each decision tree hyper-parameter;
and adopting a TPE algorithm to search the hyper-parameters of each decision tree one by one according to the set search step length.
Preferably, the target electromagnetic scattering properties data comprises one or more of: electromagnetic scattering pitch angle, electromagnetic scattering azimuth angle, electromagnetic scattering frequency;
the reorganizing of the electromagnetic scattering property data according to the electromagnetic scattering data space description specification comprises:
and arranging the target electromagnetic scattering data according to any non-interpolation object in an ascending or descending order.
On the other hand, the invention also discloses an electromagnetic scattering characteristic data interpolation generation device, which comprises:
the system comprises an organizing module, a data processing module and a data processing module, wherein the organizing module is used for acquiring target electromagnetic scattering characteristic data and organizing the electromagnetic scattering characteristic data according to electromagnetic scattering data space description specifications;
the smoothing module is arranged for smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism;
and the interpolation module is used for performing electromagnetic scattering data interpolation generation by utilizing a preset decision tree cluster model.
Preferably, the obtaining manner of the preset decision tree cluster model in the interpolation module includes:
and training the strategy tree cluster model, performing optimization feedback of a target function through a training result, and setting the hyperparameter based on the Bayesian theory to obtain a preset decision tree cluster model.
Preferably, the smoothing module performs smoothing processing on the electromagnetic scattering characteristic data according to a preset smoothing processing mechanism, and the smoothing processing includes:
and setting a smooth sliding window specification and a sliding window step length according to the oscillation characteristics of the electromagnetic scattering data, and performing smooth processing on the electromagnetic scattering characteristic data according to a data assignment mechanism.
Compared with the prior art, the invention has the following advantages:
according to the electromagnetic scattering characteristic data interpolation generation method, target electromagnetic scattering data which are incomplete or have coarse data granularity are subjected to data compilation, smoothing processing and decision tree cluster design and training in sequence, and electromagnetic scattering data interpolation data are finally obtained, so that the target electromagnetic scattering data interpolation generation is realized, the data granularity is perfected and refined, and the target electromagnetic scattering characteristic research is facilitated.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the example serve to explain the principles of the invention and not to limit the invention.
FIG. 1 is a flow chart of an interpolation generation method of electromagnetic scattering property data according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of an electromagnetic scattering property data interpolation generation apparatus according to an embodiment of the present invention;
FIG. 3 is a flowchart of an electromagnetic scattering property data interpolation method based on decision tree cluster calculation according to an embodiment of the present invention;
FIG. 4 is a graph of electromagnetic scattering property data compilation effect according to an embodiment of the present invention;
FIG. 5 is a visualization of a portion of a decision tree design in a decision tree cluster in accordance with an embodiment of the present invention;
FIG. 6 is a graphical illustration of an overall wave visualization of electromagnetic data in accordance with an embodiment of the present invention;
fig. 7 is a graph for visualizing interpolation effect of electromagnetic scattering data according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments and features of the embodiments in the present application may be arbitrarily combined with each other without conflict.
The steps illustrated in the flow charts of the figures may be performed in a computer system such as a set of computer-executable instructions. Also, while a logical order is shown in the flow diagrams, in some cases, the steps shown or described may be performed in an order different than here.
Example one
Fig. 1 is a flowchart of an electromagnetic scattering property data interpolation generation method according to an embodiment of the present invention, where the electromagnetic scattering property data interpolation generation method according to the embodiment of the present invention includes:
s101, obtaining target electromagnetic scattering characteristic data, and performing compilation on the electromagnetic scattering characteristic data according to electromagnetic scattering data space description specifications;
s102, smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism;
s103, performing electromagnetic scattering data interpolation generation by using a preset decision tree cluster model.
In this embodiment of the present invention, the obtaining manner of the preset decision tree cluster model in step S103 includes:
and training the strategy tree cluster model, performing optimization feedback of a target function through a training result, and setting the hyperparameter based on the Bayesian theory to obtain a preset decision tree cluster model.
In this embodiment of the present invention, the step S102 of smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism includes:
and setting a smooth sliding window specification and a sliding window step length according to the oscillation characteristics of the electromagnetic scattering data, and performing smooth processing on the electromagnetic scattering characteristic data according to a data assignment mechanism.
In the embodiment of the present invention, in step S103, the objective function selects a mean square error function as a loss function, L2Regularization is used as an objective function regularization term.
In the embodiment of the present invention, the performing optimization feedback of the objective function through the training result in step S103 includes:
the decision tree objective function Obj (θ) includes a loss function L (θ) and a regularization term Ω (θ), and the expression is:
Obj(θ)=L(θ)+Ω(θ)
wherein theta is a model parameter, L (theta) is a loss function, and omega (theta) is a regularization term;
selecting L2Regularization, the expression is:
Ω(fk)=λ‖fk‖2
wherein f iskIs the kth tree model, lambda is the parameter coefficient to be adjusted, | fkII denotes the pair fkCalculating a norm;
the loss function is a mean square error loss function, and the expression is as follows:
wherein, ytIs the t-th sample data tag value, t is the number of samples,and N is the total number of samples.
In the embodiment of the invention, the loss function and the target function are determined after the regularization selection, and the parameter setting and the model training of the target function to obtain the optimal solution are carried out, so that the target function value is minimum, and the optimal model is obtained. In the embodiment of the invention, the target function comprises a loss function and regularization, the minimum loss function is an optimal target function, and the regularization plays a role in preventing overfitting.
In the embodiment of the present invention, the setting of the hyper-parameters based on the bayesian theory through the training results in step S103 includes:
minimizing a loss function in the regression calculation objective function of the decision tree cluster;
setting a parameter value range and a search step length for each decision tree hyper-parameter;
and adopting a TPE algorithm to search the hyper-parameters of each decision tree one by one according to the set search step length.
In this embodiment of the present invention, the target electromagnetic scattering characteristic data in step S101 includes one or more of the following items: electromagnetic scattering pitch angle, electromagnetic scattering azimuth angle, electromagnetic scattering frequency;
the reorganizing of the electromagnetic scattering property data according to the electromagnetic scattering data space description specification comprises:
and arranging the target electromagnetic scattering data according to any non-interpolation object in an ascending or descending order.
Example two
As shown in fig. 2, an electromagnetic scattering property data interpolation generation apparatus includes:
the organizing module 100 is configured to acquire target electromagnetic scattering characteristic data and organize the electromagnetic scattering characteristic data according to electromagnetic scattering data spatial description specifications;
the smoothing module 200 is configured to smooth the electromagnetic scattering characteristic data according to a preset smoothing mechanism;
the interpolation module 300 is configured to perform interpolation generation of electromagnetic scattering data by using a preset decision tree cluster model.
In the interpolation in the embodiment of the present invention, the obtaining manner of the preset decision tree cluster model in the interpolation module 300 includes:
and training the strategy tree cluster model, performing optimization feedback of a target function through a training result, and setting the hyperparameter based on the Bayesian theory to obtain a preset decision tree cluster model.
In this embodiment of the present invention, the interpolating the smoothing module 200 to smooth the electromagnetic scattering characteristic data according to a preset smoothing mechanism includes:
and setting a smooth sliding window specification and a sliding window step length according to the oscillation characteristics of the electromagnetic scattering data, and performing smooth processing on the electromagnetic scattering characteristic data according to a data assignment mechanism.
In the embodiment of the present invention, the objective function in the interpolation module 300 selects a mean square error function as a loss function, L2Regularization is used as an objective function regularization term.
In the interpolation of the embodiment of the present invention, the performing, by the interpolation module 300, the optimization feedback of the objective function according to the training result includes:
the decision tree objective function Obj (θ) includes a loss function L (θ) and a regularization term Ω (θ), and the expression is:
Obj(θ)=L(θ)+Ω(θ)
wherein theta is a model parameter, L (theta) is a loss function, and omega (theta) is a regularization term;
selecting L2Regularization, the expression is:
Ω(fk)=λ‖fk‖2
wherein f iskIs the kth tree model, lambda is the parameter coefficient to be adjusted, | fkII denotes the pair fkCalculating a norm;
the loss function is a mean square error loss function, and the expression is as follows:
wherein, ytIs the t-th sample data tag value, t is the number of samples,and N is the total number of samples.
In the embodiment of the present invention, the setting of the hyper-parameter by the interpolation module 300 through the training result based on the bayesian theory includes:
minimizing a loss function in the regression calculation objective function of the decision tree cluster;
setting a parameter value range and a search step length for each decision tree hyper-parameter;
and adopting a TPE algorithm to search the hyper-parameters of each decision tree one by one according to the set search step length.
In an embodiment of the present invention, the target electromagnetic scattering characteristic data of the compiling module 100 includes one or more of the following items: electromagnetic scattering pitch angle, electromagnetic scattering azimuth angle, electromagnetic scattering frequency;
the reorganizing module 100 reorganizes the electromagnetic scattering property data according to the electromagnetic scattering data spatial description specification, including:
and arranging the target electromagnetic scattering data according to any non-interpolation object in an ascending or descending order.
EXAMPLE III
As shown in fig. 3, the embodiment of the present invention describes a process for generating interpolation of target electromagnetic scattering property data based on decision tree cluster calculation:
1. determining an interpolation object and editing data: firstly, determining an interpolation object (a pitch angle, an azimuth angle and a frequency are selected as required) of the data interpolation, and editing the acquired target electromagnetic scattering characteristic data according to an electromagnetic scattering characteristic data space description method;
2. and (3) data smoothing treatment: according to the oscillation characteristics of the electromagnetic scattering data, setting the specification and the step length of a smooth sliding window and carrying out data smoothing processing according to a data assignment mechanism;
3. designing a decision tree cluster: selecting design of an objective function and optimizing a decision tree hyperparameter, wherein the objective function selects a mean square error function as a loss function, L2Regularization is used as a regularization term of the target function;
4. training a decision tree model: continuously iterating the versions of the decision tree models by training and internally feeding back the decision tree clusters until the decision tree cluster interpolation model with the highest precision is obtained;
5. and (3) electromagnetic scattering data interpolation generation: and generating a refined data file according to actual data granularity refinement requirements, performing RCS value interpolation generation through a decision tree cluster model, and combining the generated data file and an original training file to obtain target electromagnetic scattering characteristic data after data granularity refinement.
Example four
The embodiment of the invention explains the process of electromagnetic scattering data interpolation of decision tree cluster calculation:
(1) interpolation object determination and data marshalling
The determination of the interpolation object is selected by a user according to actual research and application requirements, and the selectable objects comprise a pitch angle, an azimuth angle and a frequency. After the interpolation object is determined, the acquired target electromagnetic scattering characteristic data is compiled according to an electromagnetic scattering characteristic data space description method, specifically, all the electromagnetic scattering characteristic data are collected in a numpy format file in a two-dimensional array format, the target electromagnetic scattering data are arranged in an ascending order or a descending order according to any non-interpolation object, and finally a standard data interpolation file is formed, which can refer to fig. 4, and fig. 4 shows the result after data compilation.
(2) Data smoothing
Based on the generated standard data file, smoothing processing of the RCS value corresponding to the interpolation object is performed. The details of the data smoothing process are shown in table 1, and mainly include the sliding window specification, the setting of the sliding window step length, and the description of the weighting coefficient and the weighting assignment rule.
TABLE 1
(3) Decision tree cluster design
The design of the decision tree cluster mainly relates to the splitting and growing of the decision tree, and the splitting and growing of the decision tree are mainly determined by the objective function and the hyper-parameter of the decision tree. The invention mainly selects and designs decision tree objective functions and optimally sets relevant hyper-parameters of the decision trees based on the Bayesian theory so as to obtain the optimal decision tree cluster.
1) Target function selection and design
The composition of the decision tree objective function includes a loss function and a regularization term, as shown in equation 1.
Obj(θ)=L(θ)+Ω(θ) (1)
Where θ is the model parameter, L (θ) is the loss function, and Ω (θ) is the regularization term. The loss function is used for better fitting the prediction model, and the regularization term is used for simplifying the model and improving the generalization capability of the model.
The regularization term part of the decision tree is mainly L1And L2Regularization, in the invention, L is selected in consideration of the implicit association relationship among different characteristics of electromagnetic scattering data2Regularization, as shown in equation 2. For the loss function, the requirement of second-order derivation needs to be met, and the mean square error loss function is selected and used in the invention, as shown in formula 3.
Ω(fk)=λ‖fk‖2 (2)
2) Hyper-parametric optimization
In the invention, the optimization of the hyperparameters of the decision tree is mainly completed by a distributed asynchronous algorithm configuration/hyperparameter optimization (Hyperopt) class library based on the Bayesian theory. Aiming at parameters such as the learning rate of the decision tree, the maximum tree depth, the cotyledon weight and the like, the complicated hyper-parameter optimization process can be eliminated by using Hyperopt, and the optimal hyper-parameter collocation can be automatically obtained according to the characteristics of training data.
The implementation flow and related content and parameter setting for parameter tuning by Hyperopt are summarized as follows:
the fn function is set to be minimized. And setting the fn function as a loss function in a regression calculation target function of the decision tree cluster, and minimizing the fn function by the function fmin in the Hyperopt so as to carry out parameter adjustment work.
And setting a search space. And listing the hyper-parameters needing to be adjusted, and defining a parameter space (value range) and a search step length aiming at each hyper-parameter.
And setting a search algorithm. The search algorithm may simply call the default tpe (tree of park estimators) algorithm.
(4) Decision tree cluster model training
And completing the construction of a decision tree model environment and dependence, training a decision tree cluster model, performing model interpolation precision verification by extracting part of training data as verification data, and using a verification result as the adjustment of the reverse feedback auxiliary model parameters and training. Part of the structure of the decision tree in the result of training the decision tree model can be seen in fig. 5, where fig. 5 is a schematic diagram of a certain decision tree in the trained decision tree model (about 60 more than ten thousand trees). Wherein "Yes, missing" in the decision tree of fig. 5 means "Yes, missing"; "No" means "No".
(5) Electromagnetic scattering data interpolation generation
Firstly, preparing an interpolated file based on a data granularity refining requirement, carrying out data interpolation based on the decision tree cluster model obtained in the step (4), and carrying out co-compilation and sorting on an interpolated data result and an original training data set so as to realize fine-grained complete target electromagnetic scattering data. According to the embodiment of the invention, unmanned aerial vehicle electromagnetic scattering data is selected for testing, the data fluctuation is visualized and can be seen in the attached figure 6, the figure 6 is an electromagnetic scattering data fluctuation schematic diagram, the fluctuation of the visible electromagnetic scattering data is unstable and complex, the difference effect of the method of the embodiment of the invention can be seen in the attached figure 7, and the generated value has effectiveness and robustness.
It will be understood by those of ordinary skill in the art that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components in cooperation. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.
Claims (10)
1. An electromagnetic scattering property data interpolation generation method is characterized by comprising the following steps:
acquiring target electromagnetic scattering characteristic data, and performing compilation on the electromagnetic scattering characteristic data according to electromagnetic scattering data space description specifications;
smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism;
and performing electromagnetic scattering data interpolation generation by using a preset decision tree cluster model.
2. The method for interpolating and generating electromagnetic scattering property data according to claim 1, wherein the obtaining manner of the preset decision tree cluster model comprises:
and training the strategy tree cluster model, performing optimization feedback of a target function through a training result, and setting the hyperparameter based on the Bayesian theory to obtain a preset decision tree cluster model.
3. The method for interpolating and generating electromagnetic scattering property data according to claim 1, wherein smoothing the electromagnetic scattering property data according to a preset smoothing mechanism comprises:
and setting a smooth sliding window specification and a sliding window step length according to the oscillation characteristics of the electromagnetic scattering data, and performing smooth processing on the electromagnetic scattering characteristic data according to a data assignment mechanism.
4. The method of claim 2, wherein the objective function is a mean square error function, L, as a loss function2Regularization is used as an objective function regularization term.
5. The method of interpolating generation of electromagnetic scattering property data of claim 4, wherein the performing the optimized feedback of the objective function through the training result comprises:
the decision tree objective function Obj (θ) includes a loss function L (θ) and a regularization term Ω (θ), and the expression is:
Obj(θ)=L(θ)+Ω(θ)
wherein theta is a model parameter, L (theta) is a loss function, and omega (theta) is a regularization term;
selecting L2Regularization, the expression is:
Ω(fk)=λ‖fk‖2
wherein f iskIs the kth tree model, lambda is the parameter coefficient to be adjusted, | fkII denotes the pair fkCalculating a norm;
the loss function is a mean square error loss function, and the expression is as follows:
6. The method for generating the electromagnetic scattering property data through interpolation according to claim 2 or 4, wherein the setting of the hyper-parameters based on Bayesian theory through the training results comprises:
minimizing a loss function in the regression calculation objective function of the decision tree cluster;
setting a parameter value range and a search step length for each decision tree hyper-parameter;
and adopting a TPE algorithm to search the hyper-parameters of each decision tree one by one according to the set search step length.
7. The method of claim 1, wherein the target electromagnetic scattering property data comprises one or more of the following: electromagnetic scattering pitch angle, electromagnetic scattering azimuth angle, electromagnetic scattering frequency;
the reorganizing of the electromagnetic scattering property data according to the electromagnetic scattering data space description specification comprises:
and arranging the target electromagnetic scattering data according to any non-interpolation object in an ascending or descending order.
8. An electromagnetic scattering property data interpolation generation apparatus, comprising:
the system comprises an organizing module, a data processing module and a data processing module, wherein the organizing module is used for acquiring target electromagnetic scattering characteristic data and organizing the electromagnetic scattering characteristic data according to electromagnetic scattering data space description specifications;
the smoothing module is arranged for smoothing the electromagnetic scattering characteristic data according to a preset smoothing mechanism;
and the interpolation module is used for performing electromagnetic scattering data interpolation generation by utilizing a preset decision tree cluster model.
9. The apparatus according to claim 8, wherein the obtaining manner of the preset decision tree cluster model in the interpolation module includes:
and training the strategy tree cluster model, performing optimization feedback of a target function through a training result, and setting the hyperparameter based on the Bayesian theory to obtain a preset decision tree cluster model.
10. The interpolation generation apparatus of claim 8, wherein the smoothing module performs smoothing on the electromagnetic scattering property data according to a preset smoothing mechanism, and comprises:
and setting a smooth sliding window specification and a sliding window step length according to the oscillation characteristics of the electromagnetic scattering data, and performing smooth processing on the electromagnetic scattering characteristic data according to a data assignment mechanism.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011002152.7A CN112098733A (en) | 2020-09-22 | 2020-09-22 | Electromagnetic scattering characteristic data interpolation generation method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011002152.7A CN112098733A (en) | 2020-09-22 | 2020-09-22 | Electromagnetic scattering characteristic data interpolation generation method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112098733A true CN112098733A (en) | 2020-12-18 |
Family
ID=73755812
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011002152.7A Pending CN112098733A (en) | 2020-09-22 | 2020-09-22 | Electromagnetic scattering characteristic data interpolation generation method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112098733A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103198227A (en) * | 2013-04-18 | 2013-07-10 | 南京理工大学 | Electromagnetic scattering analyzing method for superspeed flight targets |
CN104732298A (en) * | 2015-04-02 | 2015-06-24 | 南京天溯自动化控制***有限公司 | Method for achieving EMS load prediction based on decision tree and linear regression |
CN106872972A (en) * | 2017-03-31 | 2017-06-20 | 北京环境特性研究所 | Near space Electromagnetic Scattering of Target data capture method based on sextuple interpolation |
CN109711446A (en) * | 2018-12-18 | 2019-05-03 | 中国科学院深圳先进技术研究院 | A kind of terrain classification method and device based on multispectral image and SAR image |
CN110362877A (en) * | 2019-06-24 | 2019-10-22 | 南京理工大学 | The environment electromagnetics analysis of scattering method of uncertain factor |
CN111382897A (en) * | 2019-10-25 | 2020-07-07 | 广州供电局有限公司 | Transformer area low-voltage trip prediction method and device, computer equipment and storage medium |
-
2020
- 2020-09-22 CN CN202011002152.7A patent/CN112098733A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103198227A (en) * | 2013-04-18 | 2013-07-10 | 南京理工大学 | Electromagnetic scattering analyzing method for superspeed flight targets |
CN104732298A (en) * | 2015-04-02 | 2015-06-24 | 南京天溯自动化控制***有限公司 | Method for achieving EMS load prediction based on decision tree and linear regression |
CN106872972A (en) * | 2017-03-31 | 2017-06-20 | 北京环境特性研究所 | Near space Electromagnetic Scattering of Target data capture method based on sextuple interpolation |
CN109711446A (en) * | 2018-12-18 | 2019-05-03 | 中国科学院深圳先进技术研究院 | A kind of terrain classification method and device based on multispectral image and SAR image |
CN110362877A (en) * | 2019-06-24 | 2019-10-22 | 南京理工大学 | The environment electromagnetics analysis of scattering method of uncertain factor |
CN111382897A (en) * | 2019-10-25 | 2020-07-07 | 广州供电局有限公司 | Transformer area low-voltage trip prediction method and device, computer equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
FENG CHEN ETC.: "An EM Scattering Data Interpolation Method Based on Decision and Regression Tree", 《2019 PHOTONICS & ELECTROMAGNETICS RESEARCH SYMPOSIUM - FALL (PIERS - FALL)》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Shen et al. | Pcw-net: Pyramid combination and warping cost volume for stereo matching | |
Rau et al. | Accurate photometric redshift probability density estimation–method comparison and application | |
CN102693266B (en) | Search for method, the navigation equipment and method of generation index structure of database | |
US20130343619A1 (en) | Density estimation and/or manifold learning | |
CN110858062B (en) | Target optimization parameter obtaining method and model training method and device | |
Anukrishna et al. | A review on feature selection for high dimensional data | |
JPWO2013051619A1 (en) | Similarity detection apparatus and directivity vicinity detection method | |
CN109409496A (en) | One kind being based on the improved LDTW sequence similarity amount method of ant group algorithm | |
CN111754546A (en) | Target tracking method, system and storage medium based on multi-feature map fusion | |
CN111191555B (en) | Target tracking method, medium and system combining high-low spatial frequency characteristics | |
CN111967616B (en) | Automatic time series regression method and device | |
Zhang et al. | Efficient history matching with dimensionality reduction methods for reservoir simulations | |
CN113808681A (en) | ABO (abnormal noise) rapid prediction based on SHAP-Catboost3Method and system for specific surface area of perovskite material | |
Gaudrie et al. | Budgeted Multi-Objective Optimization with a Focus on the Central Part of the Pareto Front--Extended Version | |
CN112098733A (en) | Electromagnetic scattering characteristic data interpolation generation method and device | |
CN117371511A (en) | Training method, device, equipment and storage medium for image classification model | |
US9760933B1 (en) | Interactive shopping advisor for refinancing product queries | |
CN115953430B (en) | Video single-target tracking method and system based on improved batch sample loss function | |
CN111639712A (en) | Positioning method and system based on density peak clustering and gradient lifting algorithm | |
CN112613574B (en) | Training method of image classification model, image classification method and device | |
Liu et al. | A survey of the baseline correction algorithms for real-time spectroscopy processing | |
Majarian et al. | Cellorganizer: Learning and using cell geometries for spatial cell simulations | |
CN110378390B (en) | Multitask graph classification method | |
CN110660079A (en) | Single target tracking method based on space-time context | |
Jin et al. | Target tracking based on hierarchical feature fusion of residual 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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201218 |