A radio fuze anti-jamming performance evaluation method and device

Abstract

The application discloses a radio fuze anti-interference efficiency evaluation method and device, relates to the technical field of radar signal processing, and comprises the following steps: collecting a first anti-interference index matrix and a second anti-interference index matrix used for evaluating the performance of a radio fuze; performing capacity expansion processing and feature compression processing on the first anti-interference matrix and the second anti-interference matrix; obtaining a first optimization matrix and a second optimization matrix respectively; obtaining a first score vector and a second score vector corresponding to the first optimization matrix and the second optimization matrix respectively by using the score results of an expert group; and obtaining radio fuze anti-interference efficiency evaluation results by processing newly obtained anti-interference indexes by using a weighted first anti-interference capability evaluation neural network model and / or a second anti-interference capability evaluation neural network model. The method can take into account the independence and completeness characteristics of the evaluation index system, and can realize radio fuze anti-interference efficiency evaluation model training and evaluation under a small sample condition.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing technology, and more specifically to a method and apparatus for evaluating the anti-jamming effectiveness of radio fuses. Background Technology

[0002] When selecting an evaluation index system for a comprehensive assessment of the anti-interference effectiveness of radio fuses, it is often required that the evaluation index system possess both independence and completeness. That is, each evaluation index must be independent of the others, and the evaluation index system must cover all aspects of the assessment. However, this pair of requirements may contradict each other in practical applications. For example, if there is redundancy between index A and index B, then independence requires that only one index can be included in the comprehensive evaluation index system, but completeness requires that both indicators be included. This contradiction is very common in practical applications. Therefore, the comprehensive evaluation index system constructed above often fails to simultaneously guarantee completeness and independence. The index system optimization method based on the maximum uncorrelation theory essentially removes some indicators directly from the redundant index system, which carries the risk of weakening the completeness of the index system. Summary of the Invention

[0003] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, a first aspect of this invention proposes a method for evaluating the anti-interference performance of a radio fuze, comprising:

[0004] Based on multiple radio fuze anti-interference tests, a first anti-interference index matrix and a second anti-interference index matrix were collected to evaluate the performance of radio fuzes.

[0005] Both the first and second anti-interference matrices are expanded and feature-compressed to obtain the first and second optimized matrices, respectively.

[0006] Using the scoring results of the expert group, the first scoring vector and the second scoring vector corresponding to the first optimization matrix and the second optimization matrix are obtained respectively.

[0007] Using a weighted first anti-interference capability evaluation neural network model and / or a second anti-interference capability evaluation neural network model, the newly acquired feature compression processing anti-interference index is processed to obtain the radio fuze anti-interference effectiveness evaluation result. The first anti-interference capability evaluation neural network model is obtained by training based on the first optimization matrix and the first scoring vector, and the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector.

[0008] Furthermore, based on multiple radio fuze anti-interference tests, a first anti-interference index matrix and a second anti-interference index matrix were collected to evaluate the performance of the radio fuze, including:

[0009] The first anti-interference index and the second anti-interference index, which contain multiple redundant information, are obtained. The first anti-interference index includes the anti-suppression interference index, and the second anti-interference index includes the anti-deception interference index.

[0010] Multiple radio fuze anti-interference tests were conducted to collect test data on the first and second anti-interference indicators, thereby obtaining the corresponding datasets for the first and second anti-interference indicators.

[0011] The first anti-interference index matrix and the second anti-interference index matrix are obtained based on the first anti-interference index dataset and the second anti-interference index dataset.

[0012] Furthermore, both the first and second anti-interference matrices undergo expansion and feature compression processing to obtain the first and second optimized matrices, respectively, including:

[0013] The first sub-anti-interference matrix and the second sub-anti-interference matrix are obtained respectively based on the first anti-interference matrix and the second anti-interference matrix;

[0014] Singular value decomposition is performed on the first and second sub-anti-interference matrices respectively to obtain their respective left and right singular spaces and diagonal matrices arranged in descending order of singular values;

[0015] Using the right singular vectors of the right singular space corresponding to the maximum singular values ​​of the first and second sub-anti-interference matrices, the first and second right singular vectors are obtained respectively.

[0016] By using the dot product of the two first right singular vectors and the dot product of the two second right singular vectors, the corresponding first weighted vector and second weighted vector are obtained respectively.

[0017] By weighting the first anti-interference matrix and the second anti-interference matrix using the first weighting vector and the second weighting vector respectively, multiple first optimization index vectors and multiple second optimization index vectors are obtained;

[0018] The first optimization index vector and multiple second optimization index vectors are respectively added to the first anti-interference matrix and the second anti-interference matrix to obtain the first expansion matrix and the second expansion matrix;

[0019] Feature compression is performed on both the first and second expanded matrices to obtain the first optimized matrix and the second optimized matrix, respectively.

[0020] Furthermore, feature compression processing is performed on both the first and second expanded matrices to obtain the first optimized matrix and the second optimized matrix, respectively, including:

[0021] Based on the eigenvalue decomposition results of the covariance matrices of the first expansion matrix and the second expansion matrix, the first eigenspace and the first triangular matrix arranged in descending order of the corresponding eigenvalues, and the second eigenspace and the second triangular matrix arranged in descending order of the corresponding eigenvalues ​​are obtained respectively.

[0022] The first compression ratio is determined by the ratio of the sum of the first few eigenvalues ​​of the first triangular matrix to the sum of all eigenvalues.

[0023] The second compression ratio is determined by the ratio of the sum of the first few eigenvalues ​​of the second triangular matrix to the sum of all eigenvalues.

[0024] Based on the first compression ratio greater than the first threshold, determine the first load vector of the first feature space corresponding to the first multiple eigenvalues ​​of the first triangular matrix that are sorted first.

[0025] Based on the second compression ratio, which is greater than the preset first threshold, determine the second load vector of the second feature space corresponding to the first few eigenvalues ​​of the second triangular matrix.

[0026] The first optimization matrix and the second optimization matrix are obtained by multiplying the first load vector and the first expansion matrix, and by multiplying the second load vector and the second expansion matrix, respectively.

[0027] Furthermore, using the scoring results of the expert group, the first scoring vector and the second scoring vector corresponding to the first optimization matrix and the second optimization matrix are obtained respectively, including:

[0028] Based on the independent scoring results of each radio fuze anti-interference test by multiple experts, the first anti-interference score vector and the second anti-interference score vector of each expert are obtained.

[0029] The subjective weight of each expert is obtained based on the results of peer review by multiple experts. The first objective weight and the second objective weight of each expert are obtained based on the first anti-interference score vector and the second anti-interference score vector.

[0030] Based on the weighted subjective weight and the first objective weight, the first comprehensive weight of each expert is obtained;

[0031] Based on the weighted subjective weight and the second objective weight, the second comprehensive weight of each expert is obtained;

[0032] The final first score vector is obtained by summing the products of each expert's first comprehensive weight and the first anti-interference score vector.

[0033] The final second scoring vector is obtained by summing the products of the second comprehensive weight of each expert and the second anti-interference scoring vector.

[0034] Furthermore, the first anti-interference capability assessment neural network model is obtained by training based on the first optimization matrix and the first scoring vector, including:

[0035] The first anti-interference capability evaluation neural network model was trained using the one-out-of-row cross-validation method.

[0036] The neural network model is evaluated by inputting the dataset in the first optimization matrix to assess its first anti-interference capability, and multiple first output values ​​are obtained.

[0037] The condition for ending the training of the first anti-interference capability evaluation neural network model is that multiple first error values ​​between multiple first output values ​​and multiple values ​​in the first scoring vector are all less than multiple preset second thresholds.

[0038] Furthermore, the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector, including:

[0039] The second anti-interference capability evaluation neural network model was trained using the out-of-range cross-validation method.

[0040] The neural network model is evaluated by inputting the dataset from the second optimization matrix into the second anti-interference capability, and multiple second output values ​​are obtained.

[0041] The training of the second anti-interference capability evaluation neural network model ends when multiple second error values ​​of multiple second output values ​​and multiple values ​​of multiple values ​​in the second scoring vector are all less than multiple preset third thresholds.

[0042] In another aspect, the present invention provides a radio fuze anti-interference performance evaluation device, comprising:

[0043] The data acquisition module is used to collect a first anti-interference index matrix and a second anti-interference index matrix for evaluating the performance of radio fuses based on multiple radio fuse anti-interference tests.

[0044] The first data processing module is used to perform expansion processing and feature compression processing on both the first anti-interference matrix and the second anti-interference matrix to obtain the first optimized matrix and the second optimized matrix, respectively.

[0045] The second data processing module is used to obtain the first score vector and the second score vector corresponding to the first optimization matrix and the second optimization matrix, respectively, by utilizing the scoring results of the expert group.

[0046] The data output module is used to process the newly acquired anti-interference index after feature compression using a weighted first anti-interference capability evaluation neural network model and / or a second anti-interference capability evaluation neural network model to obtain the anti-interference effectiveness evaluation result of the radio fuse. The first anti-interference capability evaluation neural network model is obtained by training based on the first optimization matrix and the first scoring vector, and the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector.

[0047] In another aspect, the present invention provides an electronic device comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the radio fuze anti-interference performance evaluation method as described in the first aspect.

[0048] In another aspect, the present invention provides a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the radio fuze anti-interference performance evaluation method as described in the first aspect.

[0049] This invention provides a method and apparatus for evaluating the anti-interference performance of radio fuses, which have the following advantages compared with the prior art:

[0050] This invention first incorporates all radio fuze anti-interference index data, focusing on the representativeness and completeness of the index data without considering the independence of the indicators, greatly facilitating the establishment of the index system. Then, through feature compression, the index system is dimensionality-reduced to obtain a low-dimensional and mutually independent index system, obtaining input training data for the first and second anti-interference capability evaluation neural network models. Experts are then invited to score the evaluation results, obtaining output training data for the first and second anti-interference capability evaluation neural network models. The first and second anti-interference capability evaluation neural network models are trained using the input and output training data. The anti-interference effectiveness is comprehensively evaluated using the first and / or second anti-interference capability evaluation neural network models. The method provided by this invention uses the feature compression results of multiple radio fuze anti-interference tests as input and expert evaluation results to train the neural network model. In subsequent test evaluations, the model can be used to directly obtain evaluation results without repeated expert participation. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the present invention, the accompanying drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0052] Figure 1 A flowchart of a method for evaluating the anti-interference performance of a radio fuze provided in an embodiment of the present invention;

[0053] Figure 2 A network model diagram of a radio fuze anti-interference effectiveness evaluation method provided in an embodiment of the present invention;

[0054] Figure 3 This is a structural diagram of a radio fuze anti-interference performance evaluation device provided in an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] This specification provides the operational steps for the methods described in the embodiments or flowcharts, but may include more or fewer operational steps based on conventional or non-inventive labor. In actual system or server product execution, the methods shown in the embodiments or drawings may be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0057] Figure 1 A flowchart of a method for evaluating the anti-interference performance of a radio fuze provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method includes:

[0058] S101. Based on multiple radio fuze anti-interference tests, collect the first anti-interference index matrix and the second anti-interference index matrix to evaluate the performance of the radio fuze.

[0059] In S101, firstly, based on multiple radio fuze anti-interference tests, a first anti-interference index matrix and a second anti-interference index matrix are collected to evaluate the performance of the radio fuze. The data in the first anti-interference index matrix and the second anti-interference index matrix are non-independent anti-interference evaluation indicators. The first anti-interference index matrix and the second anti-interference index matrix are used to construct a non-independent evaluation index system, which takes into account the independence and completeness of the evaluation index system.

[0060] S102. Both the first anti-interference matrix and the second anti-interference matrix are expanded and feature compressed to obtain the first optimized matrix and the second optimized matrix, respectively.

[0061] In S102, the non-independent evaluation index system is first optimized based on the feature compression method, and the index system is reduced in dimensionality to obtain a low-dimensional and mutually independent index system, namely the first optimization matrix and the second optimization matrix.

[0062] S103. Using the scoring results of the expert group, obtain the first scoring vector and the second scoring vector corresponding to the first optimization matrix and the second optimization matrix, respectively.

[0063] In S103, experts are then invited to score the evaluation results, and the output training data of the first anti-interference capability evaluation neural network model and the second anti-interference capability evaluation neural network model are obtained, namely the first score vector and the second score vector.

[0064] S104. Using the weighted first anti-interference capability evaluation neural network model and / or the second anti-interference capability evaluation neural network model, process the newly acquired feature compression processing anti-interference index to obtain the radio fuze anti-interference effectiveness evaluation result. The first anti-interference capability evaluation neural network model is obtained by training based on the first optimization matrix and the first scoring vector, and the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector.

[0065] In S104, the first optimization matrix and the second optimization matrix are used as input training data for the first anti-interference capability evaluation neural network model and the second anti-interference capability evaluation neural network model, respectively. The first anti-interference capability evaluation neural network model and the second anti-interference capability evaluation neural network model are trained using the input training data and the output training data. The anti-interference effectiveness of the first anti-interference capability evaluation neural network model and / or the second anti-interference capability evaluation neural network model is comprehensively evaluated. This can be done by obtaining a comprehensive evaluation of the anti-interference effectiveness of the first anti-interference capability evaluation neural network model alone, or by obtaining a comprehensive evaluation result of the anti-interference effectiveness of the first anti-interference capability evaluation neural network model and the second anti-interference capability evaluation neural network model simultaneously.

[0066] In one possible implementation, a first anti-jamming index matrix and a second anti-jamming index matrix are collected for evaluating the performance of the radio fuze based on multiple radio fuze anti-jamming tests, including:

[0067] The first anti-interference index and the second anti-interference index, which contain multiple redundant information, are obtained. The first anti-interference index includes the anti-suppression interference index, and the second anti-interference index includes the anti-deception interference index.

[0068] Multiple radio fuze anti-interference tests were conducted to collect test data on the first and second anti-interference indicators, thereby obtaining the corresponding datasets for the first and second anti-interference indicators.

[0069] The first anti-interference index matrix and the second anti-interference index matrix are obtained based on the first anti-interference index dataset and the second anti-interference index dataset.

[0070] In the embodiments provided by the present invention, p anti-suppression interference indicators with redundant information are first selected as the first anti-interference indicators, and p anti-spoofing interference indicators with redundant information are selected as the second anti-interference indicators. Then, experimental data of the p indicators are collected through n experiments to obtain the first anti-interference indicator matrix and the second anti-interference indicator matrix with the same form, which are expressed as formula (1):

[0071]

[0072] In one possible implementation, both the first anti-interference matrix and the second anti-interference matrix undergo expansion and feature compression to obtain the first optimized matrix and the second optimized matrix, respectively, including:

[0073] The first sub-anti-interference matrix and the second sub-anti-interference matrix are obtained respectively based on the first anti-interference matrix and the second anti-interference matrix;

[0074] Singular value decomposition is performed on the first and second sub-anti-interference matrices respectively to obtain their respective left and right singular spaces and diagonal matrices arranged in descending order of singular values;

[0075] Using the right singular vectors of the right singular space corresponding to the maximum singular values ​​of the first and second sub-anti-interference matrices, the first and second right singular vectors are obtained respectively.

[0076] By using the dot product of the two first right singular vectors and the dot product of the two second right singular vectors, the corresponding first weighted vector and second weighted vector are obtained respectively.

[0077] By weighting the first anti-interference matrix and the second anti-interference matrix using the first weighting vector and the second weighting vector respectively, multiple first optimization index vectors and multiple second optimization index vectors are obtained;

[0078] The first optimization index vector and multiple second optimization index vectors are respectively added to the first anti-interference matrix and the second anti-interference matrix to obtain the first expansion matrix and the second expansion matrix;

[0079] Feature compression is performed on both the first and second expanded matrices to obtain the first optimized matrix and the second optimized matrix, respectively.

[0080] In the embodiments provided by the present invention, X should have sufficient sample capacity. If the actual number of experiments is less than the required sample capacity, for example, if the number of experiments is less than twice the number of indicators, then the sample matrix needs to be expanded.

[0081] The basic idea of ​​sample expansion is to select q samples from n sample data, that is, to extract the first sub-anti-interference matrix and the second sub-anti-interference matrix from the first anti-interference matrix and the second anti-interference matrix respectively. They have the same form, as shown in formula (2):

[0082]

[0083] Then, singular value decomposition is performed on both the extracted first and second anti-interference matrices according to the method shown in formula (3);

[0084] X q =U q Λ q V q (3)

[0085] In the formula, U q and V q These are respectively called the left and right singular spaces, Λ q The diagonal elements are arranged in descending order.

[0086] Finally, using the right singular vectors of the right singular space corresponding to the maximum singular values ​​of the first and second sub-anti-interference matrices, respectively, the MATLAB command e is used to... q =V q (:,1) Obtain the first and second right singular vectors of the same form, e q ;

[0087] Based on the result of formula (4), we obtain the first weighted vector and the second weighted vector with the same form:

[0088]

[0089] In the formula, .* represents the dot product symbol, i.e. Each element in the array is e q The square of the corresponding element.

[0090] By using the first weighted vector and the second weighted vector to respectively weight and sum the first anti-interference matrix and the second anti-interference matrix, multiple first optimization index vectors and multiple second optimization index vectors in the form of formula (5) are obtained;

[0091]

[0092] The first optimization index vector and multiple second optimization index vectors are respectively added to the first anti-interference matrix and the second anti-interference matrix to obtain the first expansion matrix and the second expansion matrix, which can then expand the sample set. In equation (4), the following is used: As a weighting vector, it eliminates the influence of the eigenvector direction on the data sign, thus ensuring that the expanded sample data are positive, and also ensures that the sample data are within the interval [0,1]. Clearly, for an original dataset with p samples, it can obtain at most... The new sample data significantly expands the sample size of the dataset. The expanded sample size is denoted as l, and the dataset is denoted as X. l .

[0093] In one possible implementation, feature compression processing is performed on both the first expanded matrix and the second expanded matrix to obtain the first optimized matrix and the second optimized matrix, respectively, including:

[0094] Based on the eigenvalue decomposition results of the covariance matrices of the first expansion matrix and the second expansion matrix, the first eigenspace and the first triangular matrix arranged in descending order of the corresponding eigenvalues, and the second eigenspace and the second triangular matrix arranged in descending order of the corresponding eigenvalues ​​are obtained respectively.

[0095] The first compression ratio is determined by the ratio of the sum of the first few eigenvalues ​​of the first triangular matrix to the sum of all eigenvalues.

[0096] The second compression ratio is determined by the ratio of the sum of the first few eigenvalues ​​of the second triangular matrix to the sum of all eigenvalues.

[0097] Based on the first compression ratio greater than the first threshold, determine the first load vector of the first feature space corresponding to the first multiple eigenvalues ​​of the first triangular matrix that are sorted first.

[0098] Based on the second compression ratio, which is greater than the preset first threshold, determine the second load vector of the second feature space corresponding to the first few eigenvalues ​​of the second triangular matrix.

[0099] The first optimization matrix and the second optimization matrix are obtained by multiplying the first load vector and the first expansion matrix, and by multiplying the second load vector and the second expansion matrix, respectively.

[0100] In the embodiments provided by the present invention, after the dataset expansion is completed, the first expansion matrix and the second expansion matrix X, which have the same form, are then... l Find the eigenvalue decomposition of its covariance matrix;

[0101] The relevant mathematical methods for eigenvalue decomposition of the covariance matrix are as follows:

[0102] Let X be a data table with n sample points and p variables, i.e.

[0103]

[0104] The covariance matrix of data table X Perform eigenvalue decomposition, i.e.

[0105] R = UΛU T (7)

[0106] Where U is called the feature space of V, its columns are called eigenvectors, and Λ is a diagonal matrix with diagonal elements λ. i Arranged in descending order, representing the eigenvalues ​​of V;

[0107] In feature compression applications, it is generally believed that the eigenvectors corresponding to larger eigenvalues ​​represent most of the original information, while the eigenvectors corresponding to smaller eigenvalues ​​represent redundant information in the original data. Therefore, the eigenvectors P = U(:,1:r) corresponding to the first r largest eigenvalues ​​of U can be used to compress the original data table, with a compression ratio η.

[0108]

[0109] Generally, η is taken as ≥ 85%.

[0110] In this embodiment, based on the eigenvalue decomposition results of the covariance matrices of the first expansion matrix and the second expansion matrix, a first triangular matrix with a first eigenspace and corresponding eigenvalues ​​arranged in descending order, and a second triangular matrix with a second eigenspace and corresponding eigenvalues ​​arranged in descending order are obtained respectively.

[0111] The eigenvalue decomposition formulas for the covariance matrices of the first and second expanded matrices are both as shown in formula (9):

[0112]

[0113] The first compression ratio η1 is determined by the ratio of the sum of the first few eigenvalues ​​of the first triangular matrix to the sum of all eigenvalues.

[0114] The second compression ratio η2 is determined by the ratio of the sum of the first few eigenvalues ​​of the second triangular matrix to the sum of all eigenvalues.

[0115] Based on the first compression ratio greater than the first threshold, determine the first load vector of the first feature space corresponding to the first multiple eigenvalues ​​of the first triangular matrix that are sorted first.

[0116] Based on the second compression ratio, which is greater than the preset first threshold, determine the second load vector of the second feature space corresponding to the first few eigenvalues ​​of the second triangular matrix.

[0117] After selecting the respective dimensions r based on the first and second compression ratios, use the MATLAB formula P = U l (:,1:r) compresses and optimizes the original sample set, that is, it determines the first load vector of the first feature space corresponding to the first multiple eigenvalues ​​of the first triangular matrix and the second load vector of the second feature space corresponding to the first multiple eigenvalues ​​of the second triangular matrix.

[0118] T = X l P (10)

[0119] Finally, by multiplying the first load vector and the first expansion matrix, and by multiplying the second load vector and the second expansion matrix, the corresponding first optimization matrix and second optimization matrix are obtained, and the formulas used are as shown in formula (10).

[0120] The first and second optimization matrices serve as the final score matrices for evaluating fuze performance, i.e., the optimized index system.

[0121] Clearly, each row of data in the first and second optimization matrices represents an optimized experimental sample.

[0122] It should be noted that after obtaining the first and second expansion matrices and before feature compression, the following also includes:

[0123] The first and second expansion matrices are normalized as follows:

[0124] Assuming the input and output ranges of the neural network are both [0,1], then the consistency range should be [0,1]. To meet the consistency requirements of the dataset, data that does not meet the requirements can be preprocessed using normalization. Assume that the maximum value that a certain indicator might obtain in the experiment is x. max The minimum value is x min If the experimental result is x, then when a larger consistency requirement value represents stronger anti-interference capability (positive consistency), the normalization method is as follows:

[0125]

[0126] When a smaller value is required to represent stronger anti-interference capability (reverse consistency), the normalization method is as follows:

[0127]

[0128] Additionally, for values ​​that may reach infinity, normalization can be performed using the Sigmoid function or the negative Sigmoid function.

[0129]

[0130] Then, the original experimental dataset is compressed using the index set optimization method. Assuming the number of indicators after optimization is c, the optimized experimental index dataset can be obtained.

[0131] In one possible implementation, the scoring results of the expert group are used to obtain the first scoring vector and the second scoring vector corresponding to the first optimization matrix and the second optimization matrix, respectively, including:

[0132] Based on the independent scoring results of each radio fuze anti-interference test by multiple experts, the first anti-interference score vector and the second anti-interference score vector of each expert are obtained.

[0133] The subjective weight of each expert is obtained based on the results of peer review by multiple experts. The first objective weight and the second objective weight of each expert are obtained based on the first anti-interference score vector and the second anti-interference score vector.

[0134] Based on the weighted subjective weight and the first objective weight, the first comprehensive weight of each expert is obtained;

[0135] Based on the weighted subjective weight and the second objective weight, the second comprehensive weight of each expert is obtained;

[0136] The final first score vector is obtained by summing the products of each expert's first comprehensive weight and the first anti-interference score vector.

[0137] The final second scoring vector is obtained by summing the products of the second comprehensive weight of each expert and the second anti-interference scoring vector.

[0138] In the embodiments provided by this invention, in order to obtain the network output of the training data, d experts are invited to independently score the results of n experiments. The experts assign scores according to a five-level scale: "Excellent," "Good," "Average," "Acceptable," and "Poor," with scores of 0.95, 0.85, 0.75, 0.65, and 0.3 respectively. Let the scoring vector of the k-th expert be...

[0139]

[0140] It should be noted that, in this invention, the expert's scoring vector is distinguished as a first scoring vector and a second scoring vector, and the first scoring vector and the second scoring vector correspond to the first optimization matrix and the second optimization matrix, respectively.

[0141] Subjective weights are obtained based on expert peer review, and objective weights are obtained based on each expert's rating vector.

[0142]

[0143] In the formula, μ∈(0,1), For experts who obtained their qualifications through peer review k Subjective weight, For the rating vector f (k) The computational expert e k Objective weighting.

[0144] It should be noted that in this invention, the objective weight includes a first objective weight and a second objective weight, while the subjective weight is obtained based on the peer review results of multiple experts.

[0145] Finally, the scores are weighted and summed based on the combined weights of the expert group to obtain the final score vector.

[0146]

[0147] It should be noted that the first and second scoring vectors are obtained by weighting and summing the subjective weights of the experts with the first and second objective weights, respectively.

[0148] In one possible implementation, the first anti-interference capability assessment neural network model is obtained by training based on a first optimization matrix and a first scoring vector, including:

[0149] The first anti-interference capability evaluation neural network model was trained using the one-out-of-row cross-validation method.

[0150] The neural network model is evaluated by inputting the dataset in the first optimization matrix to assess its first anti-interference capability, and multiple first output values ​​are obtained.

[0151] The condition for ending the training of the first anti-interference capability evaluation neural network model is that multiple first error values ​​between multiple first output values ​​and multiple values ​​in the first scoring vector are all less than multiple preset second thresholds.

[0152] In the embodiments provided by the present invention, during the neural network model training phase, the input data for each training of the model is one row of T (one optimized test sample), and the expected output is the score value corresponding to the sample in the score vector f. The training ends when the error between the model output value and the expected output value is less than a set threshold.

[0153] One-out-of-pairs cross-validation is employed to improve model training accuracy. The basic idea of ​​one-out-of-pairs cross-validation is as follows: First, one set of training data is excluded, and the remaining data are used as input for model training. Second, after training, the model is used to predict the excluded data, and the error between the prediction result and the expert evaluation result is calculated. Finally, the prediction error is calculated using the above method for all data, and the model with the smallest prediction error is selected as the final model. The interference suppression index system T is used. s For example, the model training process based on the 1-out-of-place cross-validation method is as follows:

[0154] (1) Let i = 1;

[0155] (2) Let in Not including T s The i-th row in Not containing f s The i-th element in;

[0156] (3) Randomly initialize the network model and the model error threshold e, and the number of neurons in the model input layer is a. s The output layer has 1 neuron, and the hidden layer has 1 neuron. (Round up);

[0157] (4) Random selection Let f be a line in the array (let's say it's the j-th line). j The expected output of the neural network;

[0158] (5) Use the data from step (4) to train the network model once, and calculate the actual output of the trained model, assuming it is...

[0159] (6) Calculate the output deviation If Δf > e, then return to step (4); otherwise, denote the training model result as M. i , and let i=i+1;

[0160] (7) If i≤n, return to step (2); otherwise, go to step (8);

[0161] (8) For i = 1, 2, ..., n, calculate model M i For t s,i The prediction results and f s,i deviation And select The smallest neural network model is used as the final neural network model.

[0162] The final neural network structure model of the algorithm is as follows: Figure 2 As shown in the figure, a sa d U represents the optimized anti-suppression interference and anti-spoofing interference indexes, respectively. s and U d The outputs of the suppression interference and the deception interference neural networks are respectively, and the two are weighted to obtain the final output U.

[0163] In one possible implementation, the second anti-interference capability evaluation neural network model is obtained based on a second optimization matrix and a second scoring vector, including:

[0164] The second anti-interference capability evaluation neural network model was trained using the out-of-range cross-validation method.

[0165] The neural network model is evaluated by inputting the dataset from the second optimization matrix into the second anti-interference capability, and multiple second output values ​​are obtained.

[0166] The training of the second anti-interference capability evaluation neural network model ends when multiple second error values ​​of multiple second output values ​​and multiple values ​​of multiple values ​​in the second scoring vector are all less than multiple preset third thresholds.

[0167] In the embodiments provided by the present invention, the construction process of the network model is the same as that of the first anti-interference capability evaluation neural network model described above.

[0168] Finally, for the newly acquired data, the data is compressed using a load matrix. Then, the outputs of the suppression interference and deception interference models are calculated by the first anti-interference capability evaluation neural network model and the second anti-interference capability evaluation neural network model, respectively. Finally, the performance evaluation is achieved by calculating the final result based on the comprehensive weight.

[0169] In practical applications, if only the radar fuze's ability to resist suppression (or deception) jamming is evaluated, it can be achieved by setting ω. s =1, ω d =0 (or ω) s =0, ω d =1) to achieve, ω s and ω d These represent the sum of the weights of the first and second anti-interference capability evaluation neural network models, respectively.

[0170] In another aspect, the present invention also provides a radio fuze anti-interference performance evaluation device 200, such as... Figure 3 As shown, the device includes:

[0171] Data acquisition module 201 is used to collect a first anti-interference index matrix and a second anti-interference index matrix for evaluating the performance of radio fuses based on multiple radio fuse anti-interference tests.

[0172] The first data processing module 202 is used to perform expansion processing and feature compression processing on both the first anti-interference matrix and the second anti-interference matrix to obtain the first optimized matrix and the second optimized matrix respectively.

[0173] The second data processing module 203 is used to obtain the first score vector and the second score vector corresponding to the first optimization matrix and the second optimization matrix respectively by using the scoring results of the expert group;

[0174] The data output module 204 is used to process the newly acquired anti-interference index after feature compression using a weighted first anti-interference capability evaluation neural network model and / or a second anti-interference capability evaluation neural network model to obtain the anti-interference effectiveness evaluation result of the radio fuse. The first anti-interference capability evaluation neural network model is obtained by training based on the first optimization matrix and the first scoring vector, and the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector.

[0175] In another embodiment of the present invention, a device is also provided, the device including a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, the at least one instruction, the at least one program, the code set or instruction set being loaded and executed by the processor to implement the radio fuze anti-interference performance evaluation method described in the embodiment of the present invention.

[0176] In another embodiment of the present invention, a computer-readable storage medium is also provided, wherein at least one instruction, at least one program, code set or instruction set is stored in the storage medium, wherein the at least one instruction, the at least one program, the code set or instruction set is loaded and executed by a processor to implement the radio fuze anti-interference performance evaluation method described in the embodiment of the present invention.

[0177] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes multiple computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates multiple available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0178] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0179] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0180] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A method for evaluating the anti-interference effectiveness of a radio fuze, characterized in that, include: Based on multiple radio fuze anti-interference tests, a first anti-interference index matrix and a second anti-interference index matrix were collected to evaluate the performance of radio fuzes. Both the first and second anti-interference matrices are expanded and feature-compressed to obtain the first and second optimized matrices, respectively. Using the scoring results of the expert group, the first scoring vector and the second scoring vector corresponding to the first optimization matrix and the second optimization matrix are obtained respectively. Using a weighted first anti-interference capability evaluation neural network model and / or a second anti-interference capability evaluation neural network model, the newly acquired feature compression processing anti-interference index is processed to obtain the radio fuze anti-interference effectiveness evaluation result. The first anti-interference capability evaluation neural network model is trained based on the first optimization matrix and the first scoring vector, and the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector. The process of expanding and compressing both the first and second anti-interference matrices to obtain the first and second optimized matrices, respectively, includes: The first sub-anti-interference matrix and the second sub-anti-interference matrix are obtained respectively based on the first anti-interference matrix and the second anti-interference matrix; Singular value decomposition is performed on the first and second sub-anti-interference matrices respectively to obtain their respective left and right singular spaces and diagonal matrices arranged in descending order of singular values; Using the right singular vectors of the right singular space corresponding to the maximum singular values ​​of the first and second sub-anti-interference matrices, the first and second right singular vectors are obtained respectively. By using the dot product of the two first right singular vectors and the dot product of the two second right singular vectors, the corresponding first weighted vector and second weighted vector are obtained respectively. By weighting the first anti-interference matrix and the second anti-interference matrix using the first weighting vector and the second weighting vector respectively, multiple first optimization index vectors and multiple second optimization index vectors are obtained; The first optimization index vector and multiple second optimization index vectors are respectively added to the first anti-interference matrix and the second anti-interference matrix to obtain the first expansion matrix and the second expansion matrix; Feature compression is performed on both the first and second expanded matrices to obtain the first optimized matrix and the second optimized matrix, respectively.

2. The method for evaluating the anti-interference effectiveness of a radio fuze as described in claim 1, characterized in that, The process involves collecting a first anti-interference index matrix and a second anti-interference index matrix to evaluate the performance of radio fuses based on multiple radio fuse anti-interference tests, including: The first anti-interference index and the second anti-interference index, which contain multiple redundant information, are obtained. The first anti-interference index includes the anti-suppression interference index, and the second anti-interference index includes the anti-deception interference index. Multiple radio fuze anti-interference tests were conducted to collect test data on the first and second anti-interference indicators, thereby obtaining the corresponding datasets for the first and second anti-interference indicators. The first anti-interference index matrix and the second anti-interference index matrix are obtained based on the first anti-interference index dataset and the second anti-interference index dataset.

3. The method for evaluating the anti-interference effectiveness of a radio fuze as described in claim 1, characterized in that, The step of performing feature compression processing on both the first and second expanded matrices to obtain the first optimized matrix and the second optimized matrix, respectively, includes: Based on the eigenvalue decomposition results of the covariance matrices of the first expansion matrix and the second expansion matrix, the first eigenspace and the first triangular matrix arranged in descending order of the corresponding eigenvalues, and the second eigenspace and the second triangular matrix arranged in descending order of the corresponding eigenvalues ​​are obtained respectively. The first compression ratio is determined by the ratio of the sum of the first few eigenvalues ​​of the first triangular matrix to the sum of all eigenvalues. The second compression ratio is determined by the ratio of the sum of the first few eigenvalues ​​of the second triangular matrix to the sum of all eigenvalues. Based on a first compression ratio greater than a first threshold, determine the first load vector of the first feature space corresponding to the first few sorted eigenvalues ​​of the first triangular matrix; Based on the second compression ratio, which is greater than the preset first threshold, determine the second load vector of the second feature space corresponding to the first few sorted feature values ​​of the second triangular matrix; The first optimization matrix and the second optimization matrix are obtained by multiplying the first load vector and the first expansion matrix, and by multiplying the second load vector and the second expansion matrix, respectively.

4. The method for evaluating the anti-interference effectiveness of a radio fuze as described in claim 1, characterized in that, The step of using the scoring results of the expert group to obtain the first scoring vector and the second scoring vector corresponding to the first optimization matrix and the second optimization matrix, respectively, includes: Based on the independent scoring results of each radio fuze anti-interference test by multiple experts, the first anti-interference score vector and the second anti-interference score vector of each expert are obtained. The subjective weight of each expert is obtained based on the results of peer review by multiple experts. The first objective weight and the second objective weight of each expert are obtained based on the first anti-interference score vector and the second anti-interference score vector. Based on the weighted subjective weight and the first objective weight, the first comprehensive weight of each expert is obtained; Based on the weighted subjective weight and the second objective weight, the second comprehensive weight of each expert is obtained; The final first score vector is obtained by summing the products of each expert's first comprehensive weight and the first anti-interference score vector. The final second scoring vector is obtained by summing the products of the second comprehensive weight of each expert and the second anti-interference scoring vector.

5. The method for evaluating the anti-interference effectiveness of a radio fuze as described in claim 4, characterized in that, The first anti-interference capability assessment neural network model is obtained by training based on the first optimization matrix and the first scoring vector, including: The first anti-interference capability evaluation neural network model was trained using the one-out-of-row cross-validation method. The neural network model is evaluated by inputting the dataset in the first optimization matrix to assess its first anti-interference capability, and multiple first output values ​​are obtained. The condition for ending the training of the first anti-interference capability evaluation neural network model is that multiple first error values ​​between multiple first output values ​​and multiple values ​​in the first scoring vector are all less than multiple preset second thresholds.

6. The method for evaluating the anti-interference effectiveness of a radio fuze as described in claim 1, characterized in that, The second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector, including: The second anti-interference capability evaluation neural network model was trained using the out-of-range cross-validation method. The neural network model is evaluated by inputting the dataset from the second optimization matrix into the second anti-interference capability, and multiple second output values ​​are obtained. The training of the second anti-interference capability evaluation neural network model ends when multiple second error values ​​of multiple second output values ​​and multiple values ​​of multiple values ​​in the second scoring vector are all less than multiple preset third thresholds.

7. A device for evaluating the anti-interference performance of a radio fuze, characterized in that, include: The data acquisition module is used to collect a first anti-interference index matrix and a second anti-interference index matrix for evaluating the performance of radio fuses based on multiple radio fuse anti-interference tests. The first data processing module is used to perform expansion processing and feature compression processing on both the first anti-interference matrix and the second anti-interference matrix to obtain the first optimized matrix and the second optimized matrix, respectively. The second data processing module is used to obtain the first score vector and the second score vector corresponding to the first optimization matrix and the second optimization matrix, respectively, by utilizing the scoring results of the expert group. The data output module is used to process the newly acquired anti-interference index after feature compression using the weighted first anti-interference capability evaluation neural network model and / or the second anti-interference capability evaluation neural network model to obtain the anti-interference effectiveness evaluation result of the radio fuse. The first anti-interference capability evaluation neural network model is trained based on the first optimization matrix and the first scoring vector, and the second anti-interference capability evaluation neural network model is obtained based on the second optimization matrix and the second scoring vector. The process of expanding and compressing both the first and second anti-interference matrices to obtain the first and second optimized matrices, respectively, includes: The first sub-anti-interference matrix and the second sub-anti-interference matrix are obtained respectively based on the first anti-interference matrix and the second anti-interference matrix; Singular value decomposition is performed on the first and second sub-anti-interference matrices respectively to obtain their respective left and right singular spaces and diagonal matrices arranged in descending order of singular values; Using the right singular vectors of the right singular space corresponding to the maximum singular values ​​of the first and second sub-anti-interference matrices, the first and second right singular vectors are obtained respectively. By using the dot product of the two first right singular vectors and the dot product of the two second right singular vectors, the corresponding first weighted vector and second weighted vector are obtained respectively. By weighting the first anti-interference matrix and the second anti-interference matrix using the first weighting vector and the second weighting vector respectively, multiple first optimization index vectors and multiple second optimization index vectors are obtained; The first optimization index vector and multiple second optimization index vectors are respectively added to the first anti-interference matrix and the second anti-interference matrix to obtain the first expansion matrix and the second expansion matrix; Feature compression is performed on both the first and second expanded matrices to obtain the first optimized matrix and the second optimized matrix, respectively.

8. An electronic device, characterized in that, The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the radio fuze anti-interference effectiveness evaluation method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or instruction set is loaded and executed by a processor to implement the radio fuze anti-interference performance evaluation method as described in any one of claims 1-6.

Images