Target electromagnetic scattering feature rapid fitting system and method
By building an omnidirectional database and using residual connection modules, the problem of deterioration of prediction results caused by the degradation of the BP neural network model is solved, and fast and accurate electromagnetic scattering feature library modeling and radar capability verification are achieved.
Patent Information
- Application Number
- CN202411253404.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-12-24
AI Technical Summary
The BP neural network using 9 hidden layers in the prior art may cause the model to degenerate when training the model, thereby making the prediction result worse.
Using an omnidirectional database based on different typical coating materials, the basic neural network architecture is constructed, including the input layer, the output layer and the hidden layer. The residual connection module is used to combine the activation function and the loss function to build a fitting system neural network implementation architecture. The omnidirectional database completes the training of the fitting system neural network, determines the number of typical RCS values input and completes the data fit.
It effectively shortens the modeling time of electromagnetic scattering feature library, solves the RCS consistency problem in typical azimuth, prevents model degradation, and improves the accuracy of prediction results. It is suitable for quickly and accurately constructing confrontation scenarios and target characteristic indicators, and completes the closed loop of radar capability verification.
Smart Images
Figure CN119180207A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of digital simulation technology, and in particular to a system and method for quickly fitting target electromagnetic scattering characteristics. Background Art
[0002] At present, the target electromagnetic scattering characteristics are one of the important components of the airborne radar detection echo information, and are also one of the key links in radar signal level simulation and sensor detection closed-loop simulation verification. Related technologies, electromagnetic scattering characteristics calculation is usually implemented using standardized commercial software, which basically achieves process fixation. In the simulation verification process, for the same target, the distribution law of scattering characteristics of targets with different material properties at different incident angles is basically the same. The neural network training model can better achieve the fitting of the target characteristic distribution law.
[0003] It is found that there are at least the following problems in the related technology:
[0004] However, the commonly used training scheme currently uses a BP neural network with 9 hidden layers to increase the network depth to learn more features. However, due to the deep hidden layers of the network structure and the large number of neurons in each layer, the model may be degraded during training, resulting in poor prediction results.
[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present application, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention
[0006] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical components or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.
[0007] The disclosed embodiments provide a system and method for quickly fitting target electromagnetic scattering characteristics to solve the problem of poor prediction results caused by model degradation.
[0008] In some embodiments, the method for rapid fitting of target electromagnetic scattering characteristics includes: calculating the omnidirectional database of the target based on different typical coating materials, the omnidirectional database includes the typical RCS value of the target; constructing a basic neural network architecture, the basic neural network architecture includes an input layer, an output layer and a hidden layer. The input layer includes neurons for pitch angle, azimuth angle, and RCS value, respectively, the output layer includes neurons for RCS fitting values of the next adjacent azimuth angle, and the hidden layers all include residual connection modules for solving network degradation; combining activation functions and loss functions to construct a fitting system neural network implementation architecture; completing the training of the fitting system neural network implementation architecture through the omnidirectional database; determining the number of the input typical RCS values and inputting them to complete data fitting and data file output.
[0009] Optionally, the index items of the omnidirectional database include: incident frequency; incident pitch angle, the angle range of the incident pitch angle is 0°-359°; incident azimuth angle, the angle range of the incident azimuth angle is 0°-180°; wherein the omnidirectional database is indexed in sequence according to the index order of the incident frequency, the incident pitch angle and the corresponding incident azimuth angle.
[0010] Optionally, the step of calculating the omnidirectional database of the target includes: selecting air as the medium; arranging the angle values of the incident pitch angle at intervals of 1° within the angle range of the incident pitch angle; arranging the angle values of the incident pitch angle at intervals of 1° within the angle range of the incident azimuth angle; and randomly combining the angle values of the incident pitch angle and the angle values of the incident pitch angle in pairs, respectively, as the incident pitch angle and the incident azimuth angle to calculate the omnidirectional database of the target.
[0011] Optionally, each of the residual connection modules includes a residual block, and the step of solving network degradation includes: using four of the residual connection modules in the hidden layer, and each of the four residual connection modules uses the residual block with two output feature numbers; making the input feature number and the output feature number of the first residual connection module consistent; making each subsequent residual connection module double the output feature number of the previous residual connection module in the first residual block; and using an identical connection method to perform residual connection.
[0012] Optionally, after performing residual connection using the identical connection method, the method further includes: setting the maximum output feature to 360.
[0013] Optionally, when the calculated target is large, the method further includes: reading the file through VS2015 and adding random differences item by item; filtering large difference values and recalculating random differences after adding; repeatedly constructing the omnidirectional database many times until twenty sets of omnidirectional databases are available.
[0014] Optionally, after completing the training of the fitting system neural network implementation architecture through the omnidirectional database and before determining the number of typical RCS values to be input and inputting them, it also includes: packaging the fitting system neural network implementation architecture, and developing a UI interface after packaging is completed.
[0015] Optionally, the activation function includes: a Tanh function; and a Leaky ReLu function, configured to prevent neurons from being in a saturated state during training.
[0016] Optionally, the loss function includes: an MSE loss function configured to evaluate the degree of change of the data.
[0017] In some embodiments, the target electromagnetic scattering characteristic rapid fitting system comprises a processor and a memory storing program instructions, wherein the processor is configured to execute the above-mentioned target electromagnetic scattering characteristic rapid fitting method when running the program instructions.
[0018] The embodiment of the present disclosure provides a method for quickly fitting target electromagnetic scattering characteristics, which can achieve the following technical effects:
[0019] The embodiment of the present disclosure provides a method for fast fitting of electromagnetic scattering characteristics of a target, including calculating an omnidirectional database of the target based on different typical coating materials, and the omnidirectional database includes a typical RCS value of the target. A basic neural network architecture is constructed, and the basic neural network architecture includes an input layer, an output layer, and a hidden layer. The input layer includes neurons for pitch angle, azimuth angle, and RCS value, respectively, the output layer includes neurons for RCS fitting values of the next adjacent azimuth angle, and the hidden layers all include residual connection modules for solving network degradation. Combined with the activation function and the loss function, a neural network implementation architecture for the fitting system is constructed. The training of the neural network implementation architecture for the fitting system is completed through the omnidirectional database. The number of typical RCS values input is determined and input, and data fitting and data file output are completed. The method can construct a basic neural network architecture based on the scattering characteristics of typical target materials, and construct a neural network implementation architecture for the fitting system in combination with the activation function and the loss function, which can effectively shorten the modeling time of the electromagnetic scattering feature library, and can solve the problem of RCS consistency under typical azimuths. In addition, a residual connection module is added to prevent the prediction results from deteriorating due to model degradation to solve the problem of network degradation. Therefore, it helps to quickly and accurately construct typical confrontation scenarios and target characteristic indicators, complete the closed loop of radar capability verification, and is suitable for engineering simulation practice.
[0020] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] One or more embodiments are exemplarily described by corresponding drawings, and these exemplary descriptions and drawings do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are considered similar elements, and the drawings do not constitute a scale limitation, and wherein:
[0022] Figure 1 It is a schematic diagram of a basic neural network architecture of a target electromagnetic scattering characteristic fast fitting system provided by an embodiment of the present disclosure;
[0023] Figure 2 This is a working principle diagram of a residual connection module of a target electromagnetic scattering feature fast fitting system provided by an embodiment of the present disclosure;
[0024] Figure 3 is a flow chart of a method for quickly fitting target electromagnetic scattering characteristics provided by an embodiment of the present disclosure;
[0025] Figure 4 is another flow chart of a method for quickly fitting target electromagnetic scattering characteristics provided by an embodiment of the present disclosure;
[0026] Figure 5 is another flow chart of a method for quickly fitting target electromagnetic scattering characteristics provided by an embodiment of the present disclosure;
[0027] Figure 6 is another flow chart of a method for quickly fitting target electromagnetic scattering characteristics provided by an embodiment of the present disclosure;
[0028] Figure 7 is another flow chart of a method for quickly fitting target electromagnetic scattering characteristics provided by an embodiment of the present disclosure;
[0029] Figure 8 is another flow chart of a method for quickly fitting target electromagnetic scattering characteristics provided by an embodiment of the present disclosure;
[0030] Fig. 9 It is a system schematic diagram of a target electromagnetic scattering characteristic rapid fitting system provided by an embodiment of the present disclosure. DETAILED DESCRIPTION
[0031] In order to be able to understand the features and technical contents of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.
[0032] The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.
[0033] Unless otherwise stated, the term "plurality" means two or more.
[0034] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.
[0035] The term "and / or" is a description of the association relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.
[0036] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.
[0037] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other.
[0038] Combination Figures 1 to 3 As shown, the embodiment of the present disclosure provides a method for quickly fitting target electromagnetic scattering characteristics, including:
[0039] S301, based on different typical coating materials, calculating an omnidirectional database of the target, the omnidirectional database including a typical RCS value of the target.
[0040] S302, constructing a basic neural network architecture, which includes an input layer, an output layer and a hidden layer. The input layer includes neurons for the pitch angle, azimuth angle and RCS value, respectively, the output layer includes neurons for the RCS fitting value of the next adjacent azimuth angle, and the hidden layers all include residual connection modules for solving network degradation.
[0041] S303, combining the activation function and the loss function to construct a neural network implementation architecture for the fitting system.
[0042] S304, completing the training of the neural network implementation architecture of the fitting system through the omnidirectional database.
[0043] S305, determine the number of typical RCS values to be input and input them, and complete data fitting and data file output.
[0044] The disclosed embodiment provides a method for fast fitting of target electromagnetic scattering characteristics. The basic framework of a neural network is constructed based on the scattering characteristics of typical target materials, and the neural network implementation framework of the fitting system is constructed in combination with the activation function and the loss function. This method can effectively shorten the modeling time of the electromagnetic scattering feature library and solve the problem of RCS consistency under typical orientations. In addition, a residual connection module is added to prevent the prediction results from deteriorating due to model degradation, so as to solve the problem of network degradation. Therefore, it is helpful to quickly and accurately construct typical confrontation scenarios and target characteristic indicators, complete the closed loop of radar capability verification, and is suitable for engineering simulation practice.
[0045] Optionally, the number of typical RCS value settings includes 1, 4, 8, and 12.
[0046] In the embodiment of the present disclosure, the number of typical RCS values that can be set can be 1, 4, 8, or 12 fitting input points, so as to adapt to different typical value data acquisition situations.
[0047] Optionally, the index items of the omnidirectional database include incident frequency, incident pitch angle, and incident azimuth. The angle range of the incident pitch angle is 0°-359°. The angle range of the incident azimuth angle is 0°-180°. The omnidirectional database is indexed in the index order of the incident frequency, the incident pitch angle, and the corresponding incident azimuth angle.
[0048] In the disclosed embodiment, the radar wave incident frequency can be selected as 8 GHz, the radar wave incident elevation angle range is 0°-359°, and the radar wave incident azimuth angle range is 0°-180°, which can meet the actual simulation environment.
[0049] Combination Figure 1 and Figure 2 As shown, in some embodiments, Figure 4 As shown, a method for quickly fitting target electromagnetic scattering characteristics is provided, comprising:
[0050] S401, based on different typical coating materials, select air as the medium, arrange the angle values of the incident elevation angles at intervals of 1° within the angle range of the incident elevation angle, arrange the angle values of the incident elevation angles at intervals of 1° within the angle range of the incident azimuth angle, randomly combine the angle values of the incident elevation angles and the angle values of the incident elevation angles in pairs, respectively, as the incident pitch angles and the incident azimuth angles to calculate the omnidirectional database of the target, the omnidirectional database including the typical RCS value of the target.
[0051] S402, constructing a basic neural network architecture, which includes an input layer, an output layer and a hidden layer. The input layer includes neurons for the pitch angle, azimuth angle and RCS value, respectively, the output layer includes neurons for the RCS fitting value of the next adjacent azimuth angle, and the hidden layers all include a residual connection module for solving network degradation.
[0052] S403, combining the activation function and the loss function to construct a neural network implementation architecture for the fitting system.
[0053] S404, completing the training of the neural network implementation architecture of the fitting system through the omnidirectional database.
[0054] S405, determine the number of typical RCS values to be input and input them, and complete data fitting and data file output.
[0055] In the disclosed embodiment, the step of calculating the omnidirectional database of the target includes first selecting air as the medium. Then, within the angle range of the incident pitch angle, the angle values of the incident pitch angle are arranged at intervals of 1°. Then, within the angle range of the incident azimuth angle, the angle values of the incident pitch angle are arranged at intervals of 1°. Finally, the angle values of the incident pitch angle and the angle values of the incident pitch angle are randomly combined in pairs, and the omnidirectional database of the target is calculated as the incident pitch angle and the incident azimuth angle, respectively. By adopting the method of randomly combining the angle values of the incident pitch angle and the angle values of the incident pitch angle in pairs, a large number of input data combinations can be obtained, thereby fully simulating, and after being calculated as the incident pitch angle and the incident azimuth angle, respectively, the data volume of the omnidirectional database can be guaranteed.
[0056] Optionally, combined Figure 1 As shown, the number of hidden layers is nine, and the numbers of neurons in the nine hidden layers are 70, 140, 210, 280, 361, 280, 190, 100 and 50 respectively.
[0057] In the disclosed embodiment, by increasing the number of hidden layers and reasonably allocating the number of neurons in each hidden layer, the network depth is increased to learn more features.
[0058] Optionally, each residual connection module includes a residual block, such as Figure 6 As shown, the steps to address network degradation include:
[0059] S501, four residual connection modules are used in the hidden layer, and each of the four residual connection modules uses a residual block with two output feature numbers.
[0060] S502, making the input feature number and output feature number of the first residual connection module consistent.
[0061] S503, making each subsequent residual connection module double the number of output features of the previous residual connection module in the first residual block.
[0062] S504, performing residual connection using an identical connection method.
[0063] In the disclosed embodiment, the input feature number of the first residual connection module is made consistent with the output feature number, and then each subsequent residual connection module doubles the output feature number of the previous residual connection module in the first residual block, and finally an identical connection method is used for residual connection to solve the network degradation problem, so as to avoid model degradation during training and thus lead to poor prediction results.
[0064] Alternatively, if Figure 6 As shown, the steps to address network degradation include:
[0065] S601, four residual connection modules are used in the hidden layer, and each of the four residual connection modules uses a residual block with two output feature numbers.
[0066] S602, making the input feature number and output feature number of the first residual connection module consistent.
[0067] S603, making each subsequent residual connection module double the number of output features of the previous residual connection module in the first residual block.
[0068] S604, performing residual connection using an identical connection method.
[0069] S605, setting the maximum output feature to 360.
[0070] In the disclosed embodiment, after the residual connection is performed using the identical connection mode, the maximum output feature is set to 360. By setting the maximum output feature value, the maximum value and correlation of the RCS circumferential direction are limited to improve the network performance.
[0071] Combination Figure 1 and 2 As shown, in some embodiments, Figure 7 As shown, a method for quickly fitting target electromagnetic scattering characteristics is provided, comprising:
[0072] S701, based on different typical coating materials, select air as the medium, arrange the angle values of the incident elevation angle at intervals of 1° within the angle range of the incident elevation angle, arrange the angle values of the incident elevation angle at intervals of 1° within the angle range of the incident azimuth angle, randomly combine the angle values of the incident elevation angle and the angle values of the incident elevation angle in pairs, and use them as the incident elevation angle and the incident azimuth angle to calculate the omnidirectional database of the target respectively. When the calculated target is large, read the file through VS2015, and add random difference values item by item. After adding, filter the large difference values and recalculate the random difference. Repeat the construction of the omnidirectional database many times until there are twenty sets of omnidirectional databases, and the omnidirectional database includes the typical RCS value of the target.
[0073] S702, constructing a basic neural network architecture, which includes an input layer, an output layer, and a hidden layer. The input layer includes neurons for the pitch angle, azimuth angle, and RCS value, respectively, the output layer includes neurons for the RCS fitting value of the next adjacent azimuth angle, and the hidden layers all include a residual connection module for solving network degradation.
[0074] S703, combining the activation function and the loss function to construct a neural network implementation architecture for the fitting system.
[0075] S704, completing the training of the neural network implementation architecture of the fitting system through the omnidirectional database.
[0076] S705, determine the number of typical RCS values to be input and input them, and complete data fitting and data file output.
[0077] In the disclosed embodiment, when the calculated target is large, the file is first read through VS2015 and random differences are added item by item (normal distribution, mean 3db, implemented through the random library), and then after the addition is completed, large difference values are filtered and random differences are recalculated (the filtering method is abs (true-fit) <5, unit db). Finally, the omnidirectional database is repeatedly constructed many times until twenty sets of omnidirectional databases are available, so as to solve the problem of long repeated calculation cycle of large-size targets.
[0078] Combination Figure 1 and Figure 2 As shown, in some embodiments, Figure 8 As shown, a method for quickly fitting target electromagnetic scattering characteristics is provided, comprising:
[0079] S801, based on different typical coating materials, select air as the medium, arrange the angle values of the incident elevation angle at intervals of 1° within the angle range of the incident elevation angle, arrange the angle values of the incident elevation angle at intervals of 1° within the angle range of the incident azimuth angle, randomly combine the angle values of the incident elevation angle and the angle values of the incident elevation angle in pairs, and use them as the incident elevation angle and the incident azimuth angle to calculate the omnidirectional database of the target. When the calculated target is large, read the file through VS2015, and add random differences item by item. After adding, filter the large difference values and recalculate the random differences. Repeat the construction of the omnidirectional database many times until there are twenty sets of omnidirectional databases, and the omnidirectional database includes the typical RCS value of the target.
[0080] S802, constructing a basic neural network architecture, which includes an input layer, an output layer, and a hidden layer. The input layer includes neurons for the pitch angle, azimuth angle, and RCS value, respectively, the output layer includes neurons for the RCS fitting value of the next adjacent azimuth angle, and the hidden layers all include residual connection modules for solving network degradation.
[0081] S803, combining the activation function and the loss function to construct a neural network implementation architecture for the fitting system.
[0082] S804, completing the training of the neural network implementation architecture of the fitting system through the omnidirectional database.
[0083] S805, encapsulating the neural network implementation architecture of the fitting system, and developing a UI interface after encapsulation is completed.
[0084] S806, determine the number of typical RCS values to be input and input them, completing data fitting and data file output.
[0085] In the disclosed embodiment, after completing the training of the fitting system neural network implementation architecture through the omnidirectional database and before determining the number of typical RCS values to be input and inputting, it also includes encapsulating the fitting system neural network implementation architecture, and developing a UI interface after encapsulation. The UI interface supports training database import and training status feedback (with the number of databases as a progress reference), thereby providing a convenient interactive interface for users.
[0086] Optionally, the activation function includes a Tanh function and a Leaky ReLu function. The Tanh function and the Leaky ReLu function are configured to prevent neurons from being in a saturated state during training.
[0087] In the disclosed embodiment, the Tanh function and the Leaky ReLu function are used to prevent neurons from being in a saturated state during training, thereby avoiding the phenomenon of gradient disappearance caused by almost no update of their weights during back propagation.
[0088] Optionally, the loss function includes an MSE loss function. The MSE loss function is configured to evaluate the degree of variation of the data.
[0089] In the disclosed embodiment, the MSE loss function is used to calculate the accuracy of the prediction model in describing the circumferential data, thereby evaluating the degree of change in the data.
[0090] Combination Fig. 9 As shown, an embodiment of the present disclosure provides a target electromagnetic scattering feature fast fitting system 900, including a processor (Processor) 901 and a memory (Memory) 902. Optionally, the device may also include a bus 903 and a communication interface (Communication Interface) 904. Among them, the processor 901, the memory 902, and the communication interface 904 can communicate with each other through the bus 903. The communication interface 904 can be used for information transmission. The processor 901 can call the logic instructions in the memory 902 to execute the above-mentioned target electromagnetic scattering feature fast fitting method.
[0091] In addition, the logic instructions in the memory 902 described above may be implemented in the form of software functional units and when sold or used as independent products, may be stored in a computer-readable storage medium.
[0092] The memory 902 is a storage medium that can be used to store software programs and computer executable programs, such as program instructions / modules corresponding to the method in the embodiment of the present disclosure. The processor 901 executes the function application and data processing by running the program instructions / modules stored in the memory 902, that is, the control method for the air conditioner in the above embodiment is implemented.
[0093] The memory 902 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory 902 may include a high-speed random access memory and may also include a non-volatile memory.
[0094] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned method for quickly fitting target electromagnetic scattering characteristics.
[0095] The computer-readable storage medium mentioned above may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.
[0096] The technical solution of the embodiment of the present disclosure can be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in the embodiment of the present disclosure. The aforementioned storage medium may be a non-transient storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes, or a transient storage medium.
[0097] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible changes. Unless explicitly required, separate components and functions are optional, and the order of operation may vary. The parts and features of some embodiments may be included in or replace the parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of listings containing one or more associated ones. In addition, when used in the present application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the presence of other identical elements in the process, method or device comprising the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.
[0098] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods for each specific application to implement the described functions, but such implementations should not be considered to exceed the scope of the embodiments of the present disclosure. The technicians may clearly understand that, for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above may refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.
[0099] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units can be only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to implement this embodiment. In addition, each functional unit in the embodiment of the present disclosure may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit.
[0100] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.
Claims
1. A method for rapid fitting of target electromagnetic scattering characteristics, characterized in that: include: Based on different typical coating materials, an omnidirectional database of the target is calculated, wherein the omnidirectional database includes typical RCS values of the target; Constructing a basic neural network architecture, the basic neural network architecture includes an input layer, an output layer and a hidden layer. The input layer includes neurons for the pitch angle, azimuth angle and RCS value respectively, the output layer includes neurons for the RCS fitting value of the next adjacent azimuth angle, and the hidden layers all include a residual connection module for solving network degradation; Combine activation functions and loss functions to build a neural network implementation architecture for the fitting system; Complete the training of the neural network implementation architecture of the fitting system through the omnidirectional database; The number of typical RCS values to be input is determined and input, and data fitting and data file output are completed.
2. A method for rapid fitting of target electromagnetic scattering characteristics according to claim 1, characterized in that: The index items of the omnidirectional database include: Incident frequency; An incident pitch angle, wherein the incident pitch angle has an angle range of 0°-359°; An incident azimuth angle, wherein the incident azimuth angle has an angle range of 0° to 180°; The omnidirectional database is indexed in sequence according to the index order of the incident frequency, the incident pitch angle and the corresponding incident azimuth angle.
3. A method for rapid fitting of target electromagnetic scattering characteristics according to claim 2, characterized in that: The step of calculating the omnidirectional database of the target comprises: Choose air as the medium; Arrange the angle values of the incident elevation angle at intervals of 1° within the angle range of the incident elevation angle; Arrange the angle values of the incident elevation angle at intervals of 1° within the angle range of the incident azimuth angle; The angle values of the incident elevation and depression angles and the angle values of the incident elevation and depression angles are randomly combined in pairs, and used as the incident elevation angle and the incident azimuth angle to calculate the omnidirectional database of the target respectively.
4. A method for rapid fitting of target electromagnetic scattering characteristics according to claim 4, characterized in that: Each of the residual connection modules includes a residual block, and the step of solving network degradation includes: Using four residual connection modules in the hidden layer, each of the four residual connection modules uses the residual block with two output feature numbers; Making the input feature number and output feature number of the first residual connection module consistent; Each of the subsequent residual connection modules doubles the number of output features of the previous residual connection module in the first residual block; Use the identity connection method for residual connection.
5. A method for rapid fitting of target electromagnetic scattering characteristics according to claim 5, characterized in that: After the residual connection is performed using the identical connection method, the method further includes: Set the maximum number of output features to 360.
6. A method for rapid fitting of target electromagnetic scattering characteristics according to any one of claims 1 to 5, characterized in that: When the calculated target is large, it also includes: Read the file through VS2015 and add random differences item by item; After adding, filter the large difference value and recalculate the random difference; The omnidirectional database is constructed repeatedly many times until twenty sets of omnidirectional databases are available.
7. A method for rapid fitting of target electromagnetic scattering characteristics according to any one of claims 1 to 5, characterized in that: After completing the training of the neural network implementation architecture of the fitting system through the omnidirectional database and before determining and inputting the number of typical RCS values input, it also includes: The neural network implementation architecture of the fitting system is encapsulated, and a UI interface is developed after the encapsulation is completed.
8. A method for rapid fitting of target electromagnetic scattering characteristics according to any one of claims 1 to 5, characterized in that: The activation function includes: Tanh function; and The Leaky ReLu function is configured to prevent neurons from being in a saturated state during training.
9. A method for rapid fitting of target electromagnetic scattering characteristics according to any one of claims 1 to 5, characterized in that: The loss function includes: The MSE loss function is configured to evaluate the degree of variation in the data.
10. A rapid fitting system for target electromagnetic scattering characteristics, characterized in that: The invention comprises a processor and a memory storing program instructions, wherein the processor is configured to execute a method for quickly fitting target electromagnetic scattering characteristics as claimed in any one of claims 1 to 9 when running the program instructions.