Sample selection method and device for seeker anti-jamming test based on response model

Abstract

The application relates to a seeker anti-interference test sample selection method and device based on a response model. The method comprises the following steps: through electromagnetic effect mechanism analysis on performance index data and test factors in acquired initial test data, a seeker anti-interference response model is constructed, a test sample selection optimization model is constructed according to the seeker anti-interference response model and a test sample selection target, and an optimal test scheme construction algorithm is used to optimize the initial test sample set constructed in advance according to the test sample selection optimization model, so that a final test sample set is obtained. By using the method, the seeker anti-interference electromagnetic effect mechanism is considered, the test sample is optimized and selected by constructing the seeker anti-interference response model, the method is closer to the seeker anti-interference engineering background, has better interpretability, a more effective test sample set can be obtained, and a solid foundation is provided for evaluation of a seeker anti-interference performance boundary.

Description

Technical Field

[0001] This application relates to the field of seeker anti-interference testing technology, and in particular to a method and apparatus for selecting seeker anti-interference test samples based on a response model. Background Technology

[0002] Missile weapons are core equipment for modern warfare offense and defense, a cornerstone of national security, a pillar for maintaining strategic balance, and a "trump card" for the Chinese military to counter powerful adversaries. Whether a missile weapon's guidance head can meet operational requirements in complex battlefield environments and possess good anti-jamming capabilities is crucial to its combat effectiveness. However, evaluating the anti-jamming performance of a guidance head requires considering the comprehensive influence of interference, background, and targets, exhibiting characteristics such as numerous test factors, high test space dimensions, and complex constraints. How to select a representative test sample set to thoroughly investigate the anti-jamming performance of guidance heads in conjunction with the requirements of realistic combat assessments is a difficult problem of great concern to both military and civilian parties, urgently requiring the support of scientific experimental design methods. Summary of the Invention

[0003] Therefore, it is necessary to provide a response model-based method and apparatus for selecting anti-interference test samples for seekers, which can provide a new technical approach for the scientific selection of anti-interference test samples for seekers, in order to address the above-mentioned technical problems.

[0004] A method for selecting anti-interference test samples for a seeker based on a response model, the method comprising:

[0005] Historical data of the initial anti-jamming test scheme for the seeker are obtained, and the initial anti-jamming test scheme for the seeker is simulated using a digital simulation system or a hardware-in-the-loop simulation system in combination with the historical data to obtain initial test data; wherein, the initial test data includes the performance index data of the seeker's anti-jamming and the test factors affecting the seeker's anti-jamming;

[0006] The electromagnetic effect mechanism relationship between performance index data and test factors is calculated. When the electromagnetic effect mechanism relationship exceeds a set threshold, the relationship between performance index data and test factors is calculated based on the electromagnetic effect mechanism, and the relationship is used as a meta-model to construct the seeker's anti-interference response model. When the electromagnetic effect mechanism relationship is lower than the set threshold, a surrogate model is constructed based on the fidelity of the initial test data, and the surrogate model is used as the seeker's anti-interference response model.

[0007] Based on the seeker anti-interference response model and the target of test sample selection, an optimization model for test sample selection is constructed. Then, based on the optimization model for test sample selection, an optimal test scheme construction algorithm is used to optimize the pre-constructed initial test samples to obtain the final test sample set.

[0008] In one embodiment, the test factors affecting the seeker's anti-jamming capabilities include continuous numerical factors, discrete numerical factors, and disordered qualitative factors. Continuous numerical factors include jamming power and jamming bandwidth, discrete numerical factors include sea state level and number of false targets, and disordered qualitative factors include wind direction and polarization mode of jamming antennas.

[0009] In one embodiment, a surrogate model is constructed based on the fidelity of the initial test data, and the surrogate model is used as the seeker's anti-interference response model, including:

[0010] When all initial experimental data belong to a single fidelity, a single fidelity surrogate model is constructed and used as the seeker's anti-interference response model;

[0011] When all initial experimental data belong to different fidelity levels, different fidelity proxy models are constructed based on the fidelity information. The different fidelity proxy models are then fused using a correction function to obtain a multi-fidelity proxy model, which is then used as the seeker's anti-interference response model.

[0012] In one embodiment, the single-fidelity surrogate model includes a single surrogate model and a combined surrogate model. The single surrogate model includes a multinomial regression model, a Gaussian process model, a radial basis function model, and a support vector regression model. The combined surrogate model is obtained by weighting and superimposing multiple single surrogate models. The weighting coefficients of the single surrogate model include average weights and adaptive weights.

[0013] In one embodiment, different fidelity proxy models are fused using a correction function to obtain a multi-fidelity proxy model, including:

[0014] Different fidelity proxy models are fused using correction functions. During fusion, the low-fidelity information of the low-fidelity proxy model is corrected based on the high-fidelity information of the high-fidelity proxy model, resulting in a multi-fidelity proxy model. The correction functions include multiplication correction functions, addition correction functions, synthesis correction functions, and spatial mapping correction functions.

[0015] In one embodiment, an optimization model for selecting test samples is constructed based on the seeker's anti-interference response model and the target of the test sample selection, including:

[0016] Based on the seeker's anti-interference response model, a test sample selection model is constructed, and the test sample selection model is optimized by the test sample selection objectives to obtain an optimized test sample selection model. The test sample selection objectives include accuracy indicators and / or space filling indicators. The accuracy indicators include the optimality criterion, and the space filling indicators include the maximum projection criterion and the spatial point distance criterion.

[0017] In one embodiment, an optimal test scheme construction algorithm is used to optimize the pre-constructed initial test samples based on the test sample selection optimization model, resulting in a final test sample set, including:

[0018] The optimal experimental scheme construction algorithm is used to update the initial experimental samples based on the selected experimental samples to obtain the final experimental sample set. The update operation includes column swapping, row swapping and coordinate transformation of the initial experimental samples.

[0019] A sample selection device for anti-interference test of a seeker head based on a response model, the device comprising:

[0020] The data preparation module is used to acquire historical data of the initial anti-jamming test scheme of the seeker, and to conduct simulation tests on the initial anti-jamming test scheme of the seeker using a digital simulation system or a hardware-in-the-loop simulation system in combination with the historical data to obtain initial test data; wherein, the initial test data includes the performance index data of the seeker's anti-jamming and the test factors affecting the seeker's anti-jamming;

[0021] The response model construction module is used to calculate the electromagnetic effect mechanism relationship between performance index data and test factors. When the electromagnetic effect mechanism relationship exceeds a set threshold, the relationship between performance index data and test factors is calculated based on the electromagnetic effect mechanism, and the relationship is used as the meta-model to construct the seeker's anti-interference response model. When the electromagnetic effect mechanism relationship is lower than the set threshold, a surrogate model is constructed based on the fidelity of the initial test data, and the surrogate model is used as the seeker's anti-interference response model.

[0022] The test sample selection module is used to construct an optimized test sample selection model based on the seeker's anti-interference response model and the test sample selection target. Based on the optimized test sample selection model, the optimal test scheme construction algorithm is used to optimize the pre-constructed initial test samples to obtain the final test sample set.

[0023] The aforementioned method and apparatus for selecting test samples for seeker anti-interference based on a response model addresses the challenges of complex test spaces and the difficulty in scientifically and effectively selecting representative test samples in seeker anti-interference. It constructs a seeker anti-interference response model by analyzing the electromagnetic effect mechanism of performance index data and test factors in the acquired initial test data. Based on this model and the test sample selection objectives, an optimization model for test sample selection is built. This optimization model then employs an optimal test scheme construction algorithm to optimize the pre-constructed initial test samples, resulting in the final test sample set. Compared to traditional design methods, this approach, by considering the electromagnetic effect mechanism of seeker anti-interference and constructing a seeker anti-interference response model for optimized test sample selection, is closer to the engineering background of seeker anti-interference, has better interpretability, and can obtain a more effective test sample set, providing a solid foundation for evaluating the boundary of seeker anti-interference performance. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a method for selecting test samples for a seeker based on a response model in one embodiment.

[0025] Figure 2 This is a schematic diagram illustrating the construction process of an optimization model for selecting experimental samples based on a regression model in one embodiment. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0027] In one embodiment, such as Figure 1 As shown, a method for selecting test samples for a seeker based on a response model is provided, including the following steps:

[0028] First, historical data of the initial anti-jamming test scheme for the seeker is acquired, and the initial anti-jamming test scheme is simulated using a digital simulation system or a hardware-in-the-loop simulation system in conjunction with the historical data to obtain initial test data. The initial test data includes the performance index data of the seeker's anti-jamming and the test factors affecting the seeker's anti-jamming. The test factors include continuous numerical factors, discrete numerical factors, and disordered qualitative factors. Continuous numerical factors include interference power and interference bandwidth, discrete numerical factors include sea state level and number of false targets, and disordered qualitative factors include wind direction and polarization mode of the jamming antenna.

[0029] It is understandable that the test factors affecting the seeker's anti-interference ability are essentially the factors influencing the seeker's anti-interference test.

[0030] Then, the electromagnetic effect mechanism relationship between performance index data and test factors is calculated. When the electromagnetic effect mechanism relationship exceeds the set threshold, the relationship between performance index data and test factors is calculated based on the electromagnetic effect mechanism, and the relationship is used as the meta-model to construct the seeker anti-interference response model. When the electromagnetic effect mechanism relationship is lower than the set threshold, a surrogate model is constructed based on the fidelity of the initial test data, and the surrogate model is used as the seeker anti-interference response model.

[0031] It is understandable that constructing a response model based on electromagnetic effect mechanisms treats the seeker as a "gray box" model. Based on a partial understanding of the electromagnetic environment effect mechanism of the seeker, mechanistic analysis is used to clarify the influence relationship between various performance indicators and factors of the seeker, guiding the construction of the response model. This type of response model has a certain degree of physical interpretability. Constructing a response model based on initial experimental data occurs when the effect mechanism between some interference performance indicators and influencing factors of the seeker is unclear. Based on the input and output data of the seeker's anti-interference test, a statistical model of its performance indicators and influencing factors is established. This method treats the seeker as a "black box" model, constructing a response model from experimental data without any understanding of the system's internal structure and mechanism.

[0032] Finally, based on the seeker's anti-interference response model and the experimental sample selection objectives, an optimized experimental sample selection model is constructed. Then, based on this model, an optimal experimental scheme construction algorithm is used to optimize the pre-constructed initial experimental samples, resulting in the final experimental sample set. The experimental sample selection objectives include accuracy indicators and / or space filling indicators. Accuracy indicators include optimality criteria (commonly such as D-criteria, A-criteria, E-criteria, G-criteria, etc.), and space filling indicators include maximum projection criteria and spatial point distance criteria.

[0033] In a specific embodiment, due to the complexity of the electromagnetic effect mechanism of the seeker's anti-interference capabilities, its response model often lacks a clear analytical expression. That is, it is difficult to theoretically establish an analytical electromagnetic effect mechanism model between the seeker's anti-interference performance indicators and numerous experimental factors. However, when the effect mechanism between some of the seeker's interference performance indicators and influencing factors is sufficiently clear—that is, when the electromagnetic effect mechanism relationship between performance indicator data and experimental factors exceeds a set threshold—the relationship between performance indicator data and experimental factors can be calculated based on the electromagnetic effect mechanism. This relationship can then be used as a meta-model to construct the seeker's anti-interference response model.

[0034] Taking tracking error in performance data as an example, tracking error is a crucial performance indicator for the seeker during the tracking phase, including range tracking error, angular tracking error (azimuth and pitch angles), and velocity tracking error. Under shipborne active jamming, the sources of tracking error mainly include three aspects: first, random tracking error caused by internal noise and jamming; second, system tracking error (intentional maneuvering) or random tracking error (unintentional maneuvering) caused by target maneuvering; and third, random tracking error caused by target flashing. After selecting the seeker model, based on the electromagnetic effect mechanism analysis of the relationship between angular error and various test factors, it can be seen that the range tracking error σ caused by shipborne active jamming... zs Distance R from the target t Proportional to the jammer's transmit power P j Transmit antenna gain G j The loss coefficient v of polarization mismatch j It is proportional to the 1 / 2 power of the value, and is related to the target's RCS (Radar Cross-section) σ, the jammer antenna feed loss, and the radio wave propagation attenuation loss L. j Interference bandwidth B j It is inversely proportional to the 1 / 2 power. Therefore, the distance tracking error σ can be calculated. zs The relationship between various experimental factors is as follows:

[0035]

[0036] As can be seen from the above relationship, the relationship between distance tracking error and various experimental factors can be characterized by a first-order polynomial model. Therefore, a first-order polynomial can be used as a meta-model to construct the seeker's anti-interference response model.

[0037] In a specific embodiment, if the effect mechanism between the interference performance index of the seeker and the influencing factors is not clear, that is, when the electromagnetic effect mechanism relationship between the performance index data and the test factors is lower than a set threshold, assuming that the interference resistance of the seeker involves m factors in a certain test sample space, and the initial test data consists of n test points x = (x1, ..., x...) n ) T and corresponding responses constitute.

[0038] When n initial experimental data belong to a single fidelity level, a single fidelity surrogate model is constructed and used as the seeker's anti-interference response model. The single fidelity surrogate model includes a single surrogate model and a combined surrogate model. The single surrogate models include multinomial regression, Gaussian process, radial basis function, and support vector regression models. The combined surrogate model is obtained by weighted superposition of multiple single surrogate models to combine the advantages of different single surrogate models. The weighting coefficients of the single surrogate models include average weights and adaptive weights.

[0039] When all initial experimental data belong to different fidelity levels, different fidelity surrogate models are constructed based on the fidelity information. These different fidelity surrogate models are then fused using correction functions to obtain a multi-fidelity surrogate model, which is used as the seeker's anti-interference response model. Specifically, high- and low-fidelity information from different fidelity surrogate models is fused using correction functions to further reduce experimental costs. During fusion, the low-fidelity information of the low-fidelity surrogate model is corrected based on the high-fidelity information of the high-fidelity surrogate model, resulting in a multi-fidelity surrogate model. The correction functions include multiplicative correction functions, additive correction functions, comprehensive correction functions, and spatial mapping correction functions.

[0040] In a specific embodiment, the response models corresponding to the various anti-interference performance indicators of the seeker are not the same, including the differences in their influencing factors and meta-model forms, but the selection process for their test samples is consistent.

[0041] The range tracking error σ under shipborne suppression jamming zs Taking the corresponding seeker anti-interference response model as an example, such as Figure 2 As shown, firstly, the experimental sample selection model is constructed in the form of a regression model based on the seeker's anti-interference response model, which is expressed as follows:

[0042] y = β T f(X)+ε=lnσ zs , ε i ~N(0, σ 2 )

[0043] f(X)=(1,f1(X,...,f p (X)) T

[0044] = (1, ln R) t , ln P j ln G j , ln v j ,lnσ,ln L j , ln B j ) T

[0045] Where, ε i They are independent and follow the same normal distribution N(0, σ). 2 The random variable is f(X), where β represents the model parameters, and f(X) = (1, f1(X), ..., f2). p (X)) T Let P represent p linearly independent continuous functions, and T represent the matrix transpose.

[0046] Then, the experimental sample selection model is optimized using the D criterion in the experimental sample selection objective, resulting in an optimized experimental sample selection model, denoted as:

[0047] D * =argmin det(F T F)

[0048]

[0049] Where F represents the generalized design matrix, F T denotes the transpose of the generalized design matrix, det(F T F) represents F T The determinant of F, f p (x n () represents the p-th linearly independent continuous function at the test point x. n The value at;

[0050] Finally, based on the experimental sample selection optimization model, the optimal experimental scheme construction algorithm is used to update and optimize the pre-constructed initial experimental samples, resulting in the final experimental sample set. The update operations include column swapping, row swapping, and coordinate transformation of the initial experimental samples. It can be understood that the optimal experimental scheme construction algorithm starts with an initial design scheme generated randomly (or otherwise), constructs a new design by implementing some update operation on the current design, calculates the criterion value for the new design, and decides whether to replace the current design with the new design. Specifically, row swapping of the initial experimental samples means selecting one or several experimental points in the design matrix each time, and randomly selecting experimental points in the entire design space or candidate point set to swap with that experimental point. Column swapping is widely used in standard LHD (Latin hypercube design) because changing the arrangement of elements in a column can maintain the structural balance property of LHD relative to the columns. Based on this characteristic, a method for swapping coordinates from one or several columns of the design matrix has been further developed. The above row and column swapping designs also require the use of some algorithm to implement the experimental design update operation. Commonly used algorithms include simulated annealing, threshold acceptance, stochastic evolution, and genetic algorithms.

[0051] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0052] In one embodiment, a seeker anti-interference test sample selection device based on a response model is provided, comprising: a data preparation module, a response model construction module, and a test sample selection module, wherein:

[0053] The data preparation module is used to acquire historical data of the initial anti-jamming test scheme of the seeker, and to conduct simulation tests on the initial anti-jamming test scheme of the seeker using a digital simulation system or a hardware-in-the-loop simulation system in combination with the historical data to obtain initial test data; wherein, the initial test data includes the performance index data of the seeker's anti-jamming and the test factors affecting the seeker's anti-jamming;

[0054] The response model construction module is used to calculate the electromagnetic effect mechanism relationship between performance index data and test factors. When the electromagnetic effect mechanism relationship exceeds a set threshold, the relationship between performance index data and test factors is calculated based on the electromagnetic effect mechanism, and the relationship is used as the meta-model to construct the seeker's anti-interference response model. When the electromagnetic effect mechanism relationship is lower than the set threshold, a surrogate model is constructed based on the fidelity of the initial test data, and the surrogate model is used as the seeker's anti-interference response model.

[0055] The test sample selection module is used to construct an optimized test sample selection model based on the seeker's anti-interference response model and the test sample selection target. Based on the optimized test sample selection model, the optimal test scheme construction algorithm is used to optimize the pre-constructed initial test samples to obtain the final test sample set.

[0056] Specific limitations regarding the seeker anti-interference test sample selection device based on the response model can be found in the limitations of the seeker anti-interference test sample selection method based on the response model described above, and will not be repeated here. Each module in the aforementioned seeker anti-interference test sample selection device based on the response model can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0057] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0058] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for selecting test samples for an anti-interference test of a seeker based on a response model, characterized in that, The method includes: Historical data of the initial anti-interference test scheme for the seeker is obtained, and the initial anti-interference test scheme for the seeker is simulated using a digital simulation system or a hardware-in-the-loop simulation system in combination with the historical data to obtain initial test data; wherein, the initial test data includes performance index data of the seeker's anti-interference and test factors affecting the seeker's anti-interference. The electromagnetic effect mechanism relationship between the performance index data and the test factor is calculated. When the electromagnetic effect mechanism relationship exceeds a set threshold, the relationship between the performance index data and the test factor is calculated based on the electromagnetic effect mechanism, and the relationship is used as a meta-model to construct the seeker's anti-interference response model. When the electromagnetic effect mechanism relationship is lower than the set threshold, a surrogate model is constructed based on the fidelity of the initial test data, and the surrogate model is used as the seeker's anti-interference response model. Based on the seeker anti-interference response model and the test sample selection target, a test sample selection optimization model is constructed, and based on the test sample selection optimization model, an optimal test scheme construction algorithm is used to optimize the pre-constructed initial test samples to obtain the final test sample set; A proxy model is constructed based on the fidelity of the initial test data, and the proxy model is used as the seeker's anti-interference response model, including: When all initial experimental data belong to a single fidelity, a single fidelity proxy model is constructed and the single fidelity proxy model is used as the seeker's anti-interference response model; When all initial experimental data belong to different fidelity levels, different fidelity proxy models are constructed based on the fidelity information. The different fidelity proxy models are then fused using a correction function to obtain a multi-fidelity proxy model. This multi-fidelity proxy model is then used as the seeker's anti-interference response model. The single-fidelity surrogate model includes a single surrogate model and a combined surrogate model. The single surrogate model includes a multinomial regression model, a Gaussian process model, a radial basis function model, and a support vector regression model. The combined surrogate model is obtained by weighted superposition of multiple single surrogate models. The weighting coefficients of the single surrogate model include average weights and adaptive weights.

2. The method of claim 1, wherein, The test factors affecting the seeker's anti-jamming capabilities include continuous numerical factors, discrete numerical factors, and disordered qualitative factors. The continuous numerical factors include interference power and interference bandwidth. The discrete numerical factors include sea state level and number of false targets. The disordered qualitative factors include wind direction and polarization mode of the jamming antenna.

3. The method of claim 1, wherein, Different fidelity proxy models are fused using a correction function to obtain a multi-fidelity proxy model, including: Different fidelity proxy models are fused using correction functions. During fusion, the low-fidelity information of the low-fidelity proxy model is corrected based on the high-fidelity information of the high-fidelity proxy model, resulting in a multi-fidelity proxy model. The correction functions include multiplication correction functions, addition correction functions, synthesis correction functions, and spatial mapping correction functions.

4. The method of claim 1, wherein, Based on the seeker's anti-interference response model and the target selection of test samples, an optimization model for test sample selection is constructed, including: A test sample selection model is constructed based on the seeker anti-interference response model, and the test sample selection model is optimized by the test sample selection target to obtain the test sample selection optimization model; wherein, the test sample selection target includes accuracy index and / or space filling index, the accuracy index includes optimality criterion, and the space filling index includes maximum projection criterion and spatial point distance criterion.

5. The method according to claim 1, characterized in that, Based on the experimental sample selection optimization model, the optimal experimental scheme construction algorithm is used to optimize the pre-constructed initial experimental samples to obtain the final experimental sample set, including: The initial test samples are updated using the optimal test scheme construction algorithm based on the test sample selection optimization model to obtain the final test sample set; wherein, the update operation includes column swapping, row swapping and coordinate transformation of the initial test samples.

6. A response model-based seeker anti-jamming test sample selection device, characterized in that, The device includes: The data preparation module is used to acquire historical data of the initial anti-interference test scheme of the seeker, and to conduct simulation tests on the initial anti-interference test scheme of the seeker using a digital simulation system or a hardware-in-the-loop simulation system in combination with the historical data to obtain initial test data; wherein, the initial test data includes performance index data of the seeker's anti-interference and test factors affecting the seeker's anti-interference. The response model construction module is used to calculate the electromagnetic effect mechanism relationship between the performance index data and the test factor. When the electromagnetic effect mechanism relationship exceeds a set threshold, the module calculates the relationship between the performance index data and the test factor based on the electromagnetic effect mechanism, and uses the relationship as a meta-model to construct the seeker's anti-interference response model. When the electromagnetic effect mechanism relationship is lower than the set threshold, the module constructs a surrogate model based on the fidelity of the initial test data, and uses the surrogate model as the seeker's anti-interference response model. The test sample selection module is used to construct a test sample selection optimization model based on the seeker anti-interference response model and the test sample selection target, and to optimize the pre-constructed initial test samples using the optimal test scheme construction algorithm based on the test sample selection optimization model to obtain the final test sample set. A proxy model is constructed based on the fidelity of the initial test data, and the proxy model is used as the seeker's anti-interference response model, including: When all initial experimental data belong to a single fidelity, a single fidelity proxy model is constructed and the single fidelity proxy model is used as the seeker's anti-interference response model; When all initial experimental data belong to different fidelity levels, different fidelity proxy models are constructed based on the fidelity information. The different fidelity proxy models are then fused using a correction function to obtain a multi-fidelity proxy model. This multi-fidelity proxy model is then used as the seeker's anti-interference response model. The single-fidelity surrogate model includes a single surrogate model and a combined surrogate model. The single surrogate model includes a multinomial regression model, a Gaussian process model, a radial basis function model, and a support vector regression model. The combined surrogate model is obtained by weighted superposition of multiple single surrogate models. The weighting coefficients of the single surrogate model include average weights and adaptive weights.

Images