A feature parameter evaluation method for multifunction radar working mode recognition

Abstract

The application discloses a kind of feature parameter evaluation methods for multi-function radar operating mode identification, first constructs the feature parameter selection evaluation hierarchy of multi-function radar operating mode identification, then constructs interval intuitionistic fuzzy language measure table, and confirms expert selection, obtains the evaluation information of expert to feature parameter, and constructs expert preference scale by summarizing information, and according to language measure comparison result, the normalized priority probability is calculated, finally according to the priority probability of obtained criterion layer, index layer, feature layer, the influence degree of feature layer to purpose layer is calculated, and size is sorted, and the feature parameter sorting result is obtained.The method of the application uses interval intuitionistic fuzzy-AHP to evaluate the feature parameter of multi-function radar operating mode identification, provides theoretical basis and rule guidance for parameter selection of multi-function radar operating mode identification under different scenarios, is conducive to accurate parameter selection, and improves the reliability of identification result.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar characteristic parameter evaluation, and in particular relates to a characteristic parameter evaluation method for multi-function radar working mode recognition. Background Art

[0002] Multifunction radar operating mode recognition is a key component of multifunction radar cognitive technology. This technology facilitates precise understanding of target situation. It provides a quantitative understanding of target type and operating status, providing crucial evidence for developing effective response strategies and decision-making. Multifunction radar operating mode recognition and prediction is a cutting-edge area of ​​research.

[0003] Early research on multifunction radar operating mode recognition commonly used a method to correlate and match pulse descriptors acquired by detection equipment with features in a template database. Pulse descriptors include carrier frequency, pulse width, pulse amplitude, arrival time, and azimuth angle. As research deepened, more deep learning methods were introduced. The literature "Li Y, Zhu M, Ma Y, et al. Work modes recognition and boundary identification of MFR pulse sequences with a hierarchical seq2seq LSTM. IETradar, sonar & navigation, 2020, 14(9): 1343-1353" adopted an end-to-end long short-term memory network for supervised learning, and used carrier frequency, pulse width, and pulse repetition interval to complete the recognition of the multi-function radar working mode; the literature "He C, Zhang L, Wei S, et al. Multifunction Radar Working Mode Recognitionwith Unsupervised Hierarchical Modeling and Functional Semantics EmbeddingBased LSTM. IEEE Sensors Journal, 2024" added bandwidth as the fourth feature parameter on the basis of carrier frequency, pulse width, and pulse repetition interval, and completed the unsupervised hierarchical modeling and functional semantic embedding recognition of the multi-function radar working mode based on LSTM.

[0004] With the deepening of research and the increasing complexity of the electromagnetic environment, the emergence of distributed cooperative reverse detection systems has significantly improved the generalization and recognition capabilities of multi-function radar working mode recognition, and more parameters have been introduced into the recognition of multi-function radar working modes. The literature "Yu Wang, Shi Yan, Song Jiye, et al. Multi-station cooperative multi-function radar working mode recognition method based on CNN and DS evidence theory. Electronic Information Countermeasures Technology, 2024, 39(2): 33-39" proposed a multi-station cooperative multi-function radar working mode recognition method based on convolutional neural network and DS evidence theory, proposed simulating multi-station detection sites, and using waveform data and pulse amplitude data under different working modes of multi-function radar to construct data for the recognition method.

[0005] As research deepens and electromagnetic environments become increasingly complex, many new features have been proposed to complement traditional parameters, such as the number of antenna scan cycles, signal data rate, and signal bandwidth. However, most current research only explains which feature parameters should be used, but rarely focuses on the effectiveness of so many feature parameters in operating mode recognition or how to choose among multiple available feature parameters.

[0006] In summary, it is of great value to study a method for evaluating characteristic parameters of multi-function radar working mode recognition. The results of this study are helpful for the selection of characteristic parameters under multi-function radar working mode recognition, provide rule guidance for the selection of characteristic parameters of multi-function radar working mode, refine the quantitative analysis process, and provide theoretical support for subsequent data. Summary of the Invention

[0007] To solve the above technical problems, the present invention provides a characteristic parameter evaluation method for multi-function radar working mode recognition, which can provide a theoretical basis and rule guidance for parameter selection for multi-function radar working mode recognition in different scenarios.

[0008] The technical solution adopted by the present invention is: a characteristic parameter evaluation method for multi-function radar working mode identification, the specific steps are as follows:

[0009] S1. Constructing a hierarchical structure for selecting and evaluating feature parameters for multi-function radar working mode recognition;

[0010] S2. Construct an interval-valued intuitionistic fuzzy linguistic measurement scale for multi-function radar working mode recognition;

[0011] S3. Identify experts in relevant fields, obtain their evaluation information on the characteristic parameters of multi-function radar working mode recognition, and construct an expert preference scale by summarizing the information;

[0012] Experts in the field of multifunctional radar are identified and asked to evaluate the impact of multiple characteristic parameters on the recognition of multifunctional radar working modes through a questionnaire. The evaluation level is shown in the language measurement table in step S2. Finally, language measurement comparison results of multiple characteristic parameters for multifunctional radar working mode recognition at the criterion level, indicator level, and feature level are obtained, and the information is summarized to obtain an expert preference scale.

[0013] S4. Calculate the interval-valued intuitionistic fuzzy weighted average operator based on the language measurement comparison results obtained in step S3, construct a summary decision information matrix, and complete the consistency test of the decision information matrix;

[0014] The language measurement comparison results include: language measurement comparison results at the criterion layer, indicator layer, and feature layer.

[0015] S5. Calculate the interval score matrix and interval index matrix of the decision information, construct the priority weight vector, construct the decision possibility matrix, and calculate the normalized priority possibility;

[0016] The normalized priority possibility includes: the normalized priority possibility of the criterion layer, the indicator layer, and the feature layer.

[0017] S6. Based on the priority possibilities of the criterion layer, indicator layer, and feature layer obtained in step S5, the influence of the feature layer on the target layer is calculated, and the feature parameter sorting results are obtained by sorting the layers by size, thereby completing the feature parameter evaluation of the multi-function radar working mode recognition.

[0018] Furthermore, the step S1 is specifically as follows:

[0019] The characteristic parameter selection evaluation hierarchy structure is constructed to achieve a comprehensive evaluation of the characteristic parameters to be evaluated. The characteristic parameter selection evaluation hierarchy structure includes: a target layer, a criterion layer, an indicator layer, and a characteristic layer.

[0020] The target layer is the impact assessment of characteristic parameters for multi-function radar working mode identification;

[0021] The criterion layers include: generality C1, accuracy C2, reliability C3, and complexity C4;

[0022] The indicator layer constructs 12 specific factors as the indicator layer based on the criterion layer, as follows:

[0023] 1) Feature Clear Relevance C11: This is measured by whether the feature parameter has a clear physical meaning or relevance for working mode recognition. If there is a clear physical meaning or relevance, then the feature has clear relevance.

[0024] 2) Result interpretability C12: whether the different working modes described by the feature parameter are easy to interpret and read as a measure, the easier to interpret, the higher the result interpretability;

[0025] 3) Feature platform adaptability C13: whether the feature parameter can be effectively used on different detection platforms as a measure, including: single detection site, distributed detection site, the wider the platform can be used, the higher the feature platform adaptability;

[0026] 4) Feature accuracy C21: the mean square error of the feature parameter extracted by the detection system under a fixed received signal-to-noise ratio as a measure, the fixed received signal-to-noise ratio is 13 dB, the smaller the mean square error, the higher the feature accuracy;

[0027] 5) Feature discrimination C22: whether the feature parameter can successfully obtain the working mode result when describing different working modes as a measure, the more accurate the result, the higher the discrimination;

[0028] 6) Feature difference C23: whether the feature parameter will produce multiple working mode recognition results when describing different working modes as a measure, the more results, the lower the difference;

[0029] 7) Feature stability C31: the error rate of the feature parameter in different typical working environments for working mode recognition result recognition as a measure, the higher the error rate, the lower the stability;

[0030] 8) Feature noise robustness C32: the variance of the mean square error of the feature parameter extracted by the detection system under low signal-to-noise ratio complex conditions as a measure, the higher the variance, the lower the feature noise robustness;

[0031] 9) Feature interference robustness C33: the variance of the mean square error of the feature parameter extracted by the detection system under complex external interference conditions as a measure, the higher the variance, the lower the feature interference robustness;

[0032] 10) Feature degradation robustness C34: the variance of the mean square error of the feature parameter extracted by the detection system under feature degradation complex conditions as a measure, the higher the variance, the lower the feature degradation robustness;

[0033] 11) Feature calculation time occupation C41: the mean value of the time required by the detection system to extract the feature parameter under different received signal-to-noise ratio conditions as a measure, the less the occupation, the higher the evaluation;

[0034] 12) Feature calculation resource occupation C42: the occupied resources required by the detection system to extract the feature parameter under different received signal-to-noise ratio conditions as a measure, the less the occupation, the higher the evaluation.

[0035] The feature layer lists detailed feature parameters that need to be evaluated, including carrier frequency RF, pulse repetition interval PRI, and pulse width PW.

[0036] Among them, the specific parameters required for the feature layer are selected according to actual conditions.

[0037] Furthermore, the step S2 is specifically as follows:

[0038] Suppose there is a characteristic parameter set X=[x1,x2,...,x n ].

[0039] Among them, x i , i=1,2,...,n represents the i-th characteristic parameter, and n represents the total number of characteristic parameters.

[0040] For the characteristic parameter x i There is an interval-valued intuitionistic fuzzy set IVIFS A, which is expressed as follows:

[0041]

[0042] in, Represent the characteristic parameters x i For the membership interval and non-membership interval of the feature parameter set X, They represent the lower and upper limits of the membership degree, represent the lower and upper limits of non-membership degree respectively.

[0043] The ambiguity interval π is calculated by formula (2): A (x i ), the expression is as follows:

[0044]

[0045] in, represent the lower and upper limits of the ambiguity, respectively.

[0046] For IVIFSA in formula (1), when X = [x] has only one element, A is called interval-valued intuitionistic fuzzy number IVIFN. At this time, its score function S(A) and exact function H(A) are calculated as follows:

[0047]

[0048] The interval-valued intuitionistic fuzzy linguistic measurement needs to correspond to three quantitative results: membership μ, non-membership υ, and fuzziness π. The interval-valued intuitionistic fuzzy linguistic measurement table for the multi-function radar working mode recognition feature parameters is constructed, specifically including:

[0049] When the language measure A is extremely low (EL), the membership μ = [0.10, 0.25] and the non-membership υ = [0.65, 0.75];

[0050] When the language measure A is low (L), the membership μ = [0.15, 0.30] and the non-membership υ = [0.60, 0.70];

[0051] When the language measure A is low (ML), the membership μ = [0.20, 0.35] and the non-membership υ = [0.55, 0.65];

[0052] When the language measure A is low (SL), the membership μ = [0.25, 0.40], and the non-membership υ = [0.50, 0.60];

[0053] When the language measure A is general (AE), the membership μ = [0.45, 0.55], and the non-membership υ = [0.30, 0.45];

[0054] When the language measure A is equivalence (EE), the membership μ = [0.50, 0.50], and the non-membership υ = [0.50, 0.50];

[0055] When the language measure A is high (SH), the membership μ = [0.50, 0.60], and the non-membership υ = [0.25, 0.40];

[0056] When the language measure A is high (MH), the membership μ = [0.55, 0.65] and the non-membership υ = [0.20, 0.35];

[0057] When the language measure A is high (H), the membership μ = [0.60, 0.70] and the non-membership υ = [0.15, 0.30];

[0058] When the language measure A is extremely high (EH), the membership μ = [0.65, 0.75] and the non-membership υ = [0.10, 0.25].

[0059] In the language measurement table, the membership and non-membership are set to not have extreme values ​​of 0 or 1, and the fuzziness of the language measurement A is fixedly controlled at π=[0.00, 0.25].

[0060] Furthermore, the step S4 is specifically as follows:

[0061] Set the IVIFS obtained based on the language measurement comparison results The xth C Intuitionistic fuzzy numbers The expression is as follows:

[0062]

[0063] in, Respectively represent the lower limit of membership, upper limit of membership, lower limit of non-membership, and upper limit of non-membership of IVIFN; n C Indicates that the interval intuitionistic fuzzy set contains n C factors.

[0064] Defining IVIFS The interval-valued intuitionistic fuzzy weighted IVIFW operator expression is as follows:

[0065]

[0066] in, express The weight vector of and

[0067] The IVIFS operation criteria are as follows:

[0068] There are two IVIFS settings. The operation rule expression is as follows:

[0069]

[0070] Here, λ represents a constant coefficient.

[0071] Then formula (6) can be expressed as formula (9), which is as follows:

[0072]

[0073] like All are considered to have the same weight, then The IVIFW operator is the interval-valued intuitionistic fuzzy weighted average IVIFWA operator. In this case, Equation (6) can be expressed as Equation (10), which is as follows:

[0074]

[0075] By comparing the language measurement results obtained in formula (10) and step S3, the summary decision information matrix can be constructed The expression is as follows:

[0076]

[0077] in, Indicates the i C and jth C The IVIFWA operator constructed by IVIFN.

[0078] Then we can get Hesitation The expression is as follows:

[0079]

[0080] Then get The consistency test of hesitation is performed, and the expression is as follows:

[0081]

[0082] in, Represents the result of the consistency test, when It is considered to have passed the consistency test; Indicates that the matrix size is n C The consistency check threshold is as follows:

[0083] When the matrix size is 1-2, When the matrix size is 3,

[0084] When the matrix size is 4, When the matrix size is 5,

[0085] When the matrix size is 6, When the matrix size is 7,

[0086] When the matrix size is 8, When the matrix size is 9,

[0087] Based on step S3, the language measurement comparison results are respectively substituted into the language measurement comparison results of the criterion layer, indicator layer, and feature layer, and formulas (5)-(13) are repeated to calculate the interval intuitionistic fuzzy weighted average operators of the criterion layer, indicator layer, and feature layer, and construct a summarized decision information matrix to complete the consistency test of the decision information matrix.

[0088] Furthermore, the step S5 is specifically as follows:

[0089] When the decision information matrix summarized in step S4 If the consistency test conditions are met, the decision information interval score matrix is ​​calculated The expression is as follows:

[0090]

[0091] Among them, the i C and jth C The interval score of the IVIFWA operator constructed by IVIFN is

[0092] The decision information interval score matrix Exponentiation to obtain interval indexation matrix The expression is as follows:

[0093]

[0094] Among them, the i C and jth C The interval indexation score of the IVIFWA operator constructed by IVIFN is

[0095] Calculate the i-th C Interval priority weights of factors Get interval priority weight vector The expression is as follows:

[0096]

[0097]

[0098] By interval priority weight vector Constructing a decision possibility matrix The expression is as follows:

[0099]

[0100] in, Calculated by formula (19), i C =1,2,...,n C ,j C =1,2,...,n C , the calculation expression is as follows:

[0101]

[0102] Here, max(·) means taking the maximum value. and The size relationship can be calculated by formula (3) and And the exact function is calculated by formula (4) and The specific evaluation rules are as follows:

[0103] when It is believed that when It is believed that when and It is believed that when and It is believed that when and It is believed that

[0104] Then the decision possibility matrix is ​​constructed by the interval priority weight vector Calculate the priority possibility, the expression is as follows:

[0105]

[0106] According to step S2, when the language metric is "equivalent", the membership and non-membership are 0.5, which is the middle value of the language metric conversion, and 0.5 is added as a balance value.

[0107] Then the priority possibility is normalized and the expression is as follows:

[0108]

[0109] Based on step S4, the summarized decision information matrix is ​​substituted into the summarized decision information matrices of the criterion layer, indicator layer, and feature layer respectively, and equations (14)-(21) are repeated to calculate the normalized priority possibility of the criterion layer, indicator layer, and feature layer.

[0110] Further, the step S6 is specifically as follows:

[0111] The priority probability and normalized priority probability of the criterion layer obtained in step S5 are respectively and The priority possibility and normalized priority possibility of the indicator layer are and Then construct the weight information of the indicator layer-destination layer,

[0112] According to the calculated weight information ω of the index layer-target layer target And the normalized priority probability of the feature layer Calculate the influence weight I of each feature of feature-purpose target,feature , the expression is as follows:

[0113]

[0114] Finally, the obtained influence weight is normalized to obtain the normalized influence weight Perform size sorting to obtain the characteristic parameter sorting results and complete the characteristic parameter evaluation of the multi-function radar working mode recognition.

[0115] Beneficial effects of the present invention: The method of the present invention first constructs a hierarchical structure for selecting and evaluating feature parameters for multi-function radar working mode recognition, then constructs an interval intuitive fuzzy language measurement table for multi-function radar working mode recognition, identifies experts in related fields, obtains expert evaluation information on the feature parameters of multi-function radar working mode recognition, and constructs an expert preference scale by summarizing the information. Then, based on the language measurement comparison results, the interval intuitive fuzzy weighted average operator is calculated, a summarized decision information matrix is ​​constructed, and a consistency check of the decision information matrix is ​​completed. The interval score matrix and interval index matrix of the decision information are calculated, a priority weight vector is constructed, a decision possibility matrix is ​​constructed, and a normalized priority possibility is calculated. Finally, based on the obtained priority possibility of the criterion layer, the indicator layer, and the feature layer, the influence of the feature layer on the target layer is calculated, and the size is sorted to obtain the feature parameter ranking result. The method of the present invention uses the interval intuitive fuzzy-hierarchy analysis method to evaluate the feature parameters of multi-function radar working mode recognition, provides a theoretical basis and rule guidance for the parameter selection of multi-function radar working mode recognition in different scenarios, is conducive to accurate parameter selection, improves the reliability of recognition results, and provides data support for subsequent research. BRIEF DESCRIPTION OF THE DRAWINGS

[0116] Figure 1 This is a flow chart of a characteristic parameter evaluation method for multi-function radar working mode identification according to the present invention.

[0117] Figure 2 This is a diagram of the evaluation hierarchy for selecting characteristic parameters for multi-function radar operating mode recognition in an embodiment of the present invention. DETAILED DESCRIPTION

[0118] The method of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0119] like Figure 1 As shown in the flowchart of a characteristic parameter evaluation method for multi-function radar working mode identification of the present invention, the specific steps are as follows:

[0120] S1. Construct a hierarchical structure for selecting and evaluating feature parameters for multi-function radar working mode recognition;

[0121] S2. Construct an interval-valued intuitionistic fuzzy linguistic measurement scale for multi-function radar working mode recognition;

[0122] S3. Identify experts in relevant fields, obtain their evaluation information on the characteristic parameters of multi-function radar working mode recognition, and construct an expert preference scale by summarizing the information;

[0123] Experts in the field of multifunctional radar are identified and asked to evaluate the impact of multiple characteristic parameters on the recognition of multifunctional radar working modes through a questionnaire. The evaluation level is shown in the language measurement table in step S2. Finally, language measurement comparison results of multiple characteristic parameters for multifunctional radar working mode recognition at the criterion level, indicator level, and feature level are obtained, and the information is summarized to obtain an expert preference scale.

[0124] S4. Calculate the interval-valued intuitionistic fuzzy weighted average operator based on the language measurement comparison results obtained in step S3, construct a summary decision information matrix, and complete the consistency test of the decision information matrix;

[0125] The language measurement comparison results include: language measurement comparison results at the criterion layer, indicator layer, and feature layer.

[0126] S5. Calculate the interval score matrix and interval index matrix of the decision information, construct the priority weight vector, construct the decision possibility matrix, and calculate the normalized priority possibility;

[0127] The normalized priority possibility includes: the normalized priority possibility of the criterion layer, the indicator layer, and the feature layer.

[0128] S6. Based on the priority possibilities of the criterion layer, indicator layer, and feature layer obtained in step S5, the influence of the feature layer on the target layer is calculated, and the feature parameter sorting results are obtained by sorting the layers by size, thereby completing the feature parameter evaluation of the multi-function radar working mode recognition.

[0129] In this embodiment, step S1 is specifically as follows:

[0130] The key to establishing an evaluation index system is to build an appropriate evaluation hierarchy to achieve a comprehensive evaluation of the characteristic parameters to be evaluated. The evaluation hierarchy is generally divided into three layers: the target layer, the criterion layer, and the indicator layer.

[0131] like Figure 2 As shown, the characteristic parameter selection evaluation hierarchy structure described in this embodiment includes: a target layer, a criterion layer, an indicator layer, and a characteristic layer.

[0132] The target layer is the impact assessment of characteristic parameters for multi-function radar working mode identification;

[0133] The criterion layers include: generality C1, accuracy C2, reliability C3, and complexity C4;

[0134] The indicator layer constructs 12 specific factors as the indicator layer based on the criterion layer, as follows:

[0135] 1) Feature explicit relevance C11: whether the feature parameter has explicit physical meaning or relevance for work mode recognition, if there is explicit physical meaning or relevance, there is feature explicit relevance;

[0136] 2) Result interpretability C12: whether the different work modes described by the feature parameter are easy to interpret and read as a measure, the easier to interpret, the higher the result interpretability;

[0137] 3) Feature platform adaptability C13: whether the feature parameter can be effectively used on different detection platforms as a measure, including: single detection site, distributed detection site, the more extensive the platform can be used, the higher the feature platform adaptability;

[0138] 4) Feature accuracy C21: the mean square error of the feature parameter extracted by the detection system under a fixed received signal-to-noise ratio as a measure, the fixed received signal-to-noise ratio is 13dB, the smaller the mean square error, the higher the feature accuracy;

[0139] 5) Feature discrimination C22: whether the feature parameter can successfully obtain the work mode result when describing different work modes as a measure, the more accurate the result, the higher the discrimination;

[0140] 6) Feature difference C23: whether the feature parameter will produce multiple work mode recognition results when describing different work modes as a measure, the more results, the lower the difference;

[0141] 7) Feature stability C31: the error rate of the feature parameter in different typical work environments for work mode recognition result recognition as a measure, the higher the error rate, the lower the stability;

[0142] 8) Feature noise robustness C32: the variance of the mean square error of the feature parameter extracted by the detection system under low signal-to-noise ratio complex conditions as a measure, the higher the variance, the lower the feature noise robustness;

[0143] 9) Feature interference robustness C33: the variance of the mean square error of the feature parameter extracted by the detection system under external interference complex conditions as a measure, the higher the variance, the lower the feature interference robustness;

[0144] 10) Feature degradation robustness C34: the variance of the mean square error of the feature parameter extracted by the detection system under feature degradation complex conditions as a measure, the higher the variance, the lower the feature degradation robustness;

[0145] 11) Feature calculation time occupation C41: the mean value of the time required by the detection system to extract the feature parameter under different received signal-to-noise ratio conditions as a measure, the less the occupation, the higher the evaluation;

[0146] 12) Resource occupancy of feature calculation C42: The resource occupancy required by the detection system to extract the feature parameters under different receiving signal-to-noise ratio conditions is used as a measure. The lower the occupancy, the higher the evaluation.

[0147] The feature layer lists detailed feature parameters that need to be evaluated, including carrier frequency RF, pulse repetition interval PRI, and pulse width PW.

[0148] The specific parameters required for the feature layer are selected based on the actual situation. The specific parameters to be selected need to be determined by environmental factors such as the working environment and working background as well as the task objectives.

[0149] In this embodiment, step S2 is specifically as follows:

[0150] Suppose there is a characteristic parameter set X=[x1,x2,...,x n ].

[0151] Among them, x i , i=1,2,...,n represents the i-th characteristic parameter, and n represents the total number of characteristic parameters.

[0152] For the characteristic parameter x i There is an interval-valued intuitionistic fuzzy set IVIFS (Interval-Valued Intuitionistic Fuzzy Set) A, which is expressed as follows:

[0153]

[0154] in, Represent the characteristic parameters x i For the membership interval and non-membership interval of the feature parameter set X, They represent the lower and upper limits of the membership degree, represent the lower and upper limits of non-membership degree respectively.

[0155] The ambiguity interval π is calculated by formula (2): A (x i ), the expression is as follows:

[0156]

[0157] in, represent the lower and upper limits of the ambiguity, respectively.

[0158] For IVIFS A in formula (1), when X = [x] has only one element, A is called an interval-valued intuitionistic fuzzy number (IVIFN), and the score function S(A) and the accuracy function H(A) are calculated as follows:

[0159]

[0160] Therefore, the interval-valued intuitionistic fuzzy language measurement needs to correspond to the membership degree μ, the non-membership degree υ, and the fuzzy degree π. The interval-valued intuitionistic fuzzy language measurement table for the multi-functional radar working mode recognition characteristic parameters is constructed, which specifically includes:

[0161] When the language measurement A is extremely low (EL), the membership degree μ = [0.10, 0.25] and the non-membership degree υ = [0.65, 0.75];

[0162] When the language measurement A is low (L), the membership degree μ = [0.15, 0.30] and the non-membership degree υ = [0.60, 0.70];

[0163] When the language measurement A is moderately low (ML), the membership degree μ = [0.20, 0.35] and the non-membership degree υ = [0.55, 0.65];

[0164] When the language measurement A is slightly low (SL), the membership degree μ = [0.25, 0.40] and the non-membership degree υ = [0.50, 0.60];

[0165] When the language measurement A is average (AE), the membership degree μ = [0.45, 0.55] and the non-membership degree υ = [0.30, 0.45];

[0166] When the language measurement A is equal (EE), the membership degree μ = [0.50, 0.50] and the non-membership degree υ = [0.50, 0.50];

[0167] When the language measurement A is slightly high (SH), the membership degree μ = [0.50, 0.60] and the non-membership degree υ = [0.25, 0.40];

[0168] When the language measurement A is moderately high (MH), the membership degree μ = [0.55, 0.65] and the non-membership degree υ = [0.20, 0.35];

[0169] When the language measurement A is high (H), the membership degree μ = [0.60, 0.70] and the non-membership degree υ = [0.15, 0.30];

[0170] When the language measure A is extremely high (EH), the membership μ = [0.65, 0.75] and the non-membership υ = [0.10, 0.25].

[0171] In the language metric table, to prevent information loss and other issues that could lead to unreasonable results, the membership and non-membership values ​​are set to avoid extreme values ​​of 0 or 1, and the ambiguity of the language metric A is fixed at π = [0.00, 0.25]. By setting appropriate language metrics, the expert assessment of the impact of characteristic parameters on multifunction radar operating mode recognition can be refined, resulting in more accurate assessment results.

[0172] In this embodiment, step S3 is specifically as follows:

[0173] This example uses a Likert scale to conduct a questionnaire survey with two experts (Expert 1 and Expert 2) to obtain their evaluation information on the characteristic parameters for multifunction radar operating mode recognition under a single detection platform. The characteristic parameters selected in the feature layer include seven characteristic parameters: carrier frequency (RF), pulse width (PW), pulse repetition interval (PRI), angle of arrival (AOA), pulse amplitude (PA), bandwidth (BW), and data rate (DR). These parameters are generally known to industry professionals. If other signal characteristic parameters need to be added, they can be selected and evaluated based on the actual detected data characteristics.

[0174] The preliminary results of the questionnaire in this embodiment are shown in Tables 1-6. Table 1 shows the language measurement results of expert 1's evaluation criteria layer, Table 2 shows the language measurement results of expert 1's evaluation indicator layer, Table 3 shows the language measurement results of expert 1's evaluation feature-indicator layer C11, Table 4 shows the language measurement results of expert 2's evaluation criteria layer, Table 5 shows the language measurement results of expert 2's evaluation indicator layer, and Table 6 shows the language measurement results of expert 2's evaluation feature-indicator layer C11.

[0175] Table 1

[0176]

[0177] Table 2

[0178]

[0179] Table 3

[0180]

[0181] Table 4

[0182]

[0183] Table 5

[0184]

[0185] Table 6

[0186]

[0187] Among them, the results of the language measurement in Tables 1, 2, 4, and 5 represent the influence of the criteria / indicators on the radar working mode recognition results relative to the comparison criteria / indicators; Tables 3 and 6 represent the performance of each feature parameter in the indicator layer C11. There are as many such matrices here as there are indicator layers, that is, 12, which represent the performance of each feature parameter in each indicator and are not repeated here.

[0188] Furthermore, the preliminary results of the questionnaire represent only two experts' assessments of the characteristic parameters for multi-function radar operating mode recognition. If further requirements are met, a new multi-expert evaluation survey and a multi-detection platform survey will be necessary based on actual circumstances. During the survey, to simulate situations where information is difficult to obtain and uncertainty is high, only a portion of the simulated working environment and external conditions were provided to the two experts. This resulted in a high number of intermediate values ​​in the survey, making it extremely difficult to distinguish between these values ​​using the existing analytic hierarchy process.

[0189] In this embodiment, step S4 is specifically as follows:

[0190] Set the IVIFS obtained based on the language measurement comparison results The xth C Intuitionistic fuzzy numbers The expression is as follows:

[0191]

[0192] in, Respectively represent the lower limit of membership, upper limit of membership, lower limit of non-membership, and upper limit of non-membership of IVIFN; n C Indicates that the interval intuitionistic fuzzy set contains n C factors.

[0193] Defining IVIFS The interval valued intuitionistic fuzzy weighted IVIFW (Interval Valued Intuitionistic Fuzzy Weighted) operator expression is as follows:

[0194]

[0195] in, express The weight vector of and

[0196] The IVIFS operation criteria are as follows:

[0197] There are two IVIFS settings. The operation rule expression is as follows:

[0198]

[0199] Here, λ represents a constant coefficient.

[0200] Then formula (6) can be expressed as formula (9), which is as follows:

[0201]

[0202] like All are considered to have the same weight, then The IVIFW operator is the Interval Valued Intuitionistic Fuzzy Weighted Averaging (IVIFWA) operator. In this case, Equation (6) can be expressed as Equation (10), which is as follows:

[0203]

[0204] By comparing the language measurement results obtained in formula (10) and step S3, the summary decision information matrix can be constructed The expression is as follows:

[0205]

[0206] in, Indicates the i C and jth C The IVIFWA operator constructed by IVIFN.

[0207] Then we can get Hesitation The expression is as follows:

[0208]

[0209] Then get The consistency test of hesitation is performed, and the expression is as follows:

[0210]

[0211] in, Represents the result of the consistency test, when It is considered to have passed the consistency test; Indicates that the matrix size is n CThe consistency check threshold is set as follows:

[0212] When the matrix size is 1-2, When the matrix size is 3,

[0213] When the matrix size is 4, When the matrix size is 5,

[0214] When the matrix size is 6, When the matrix size is 7,

[0215] When the matrix size is 8, When the matrix size is 9,

[0216] Based on step S3, the language measure comparison results are respectively substituted into the language measure comparison results of the criterion layer, the index layer and the feature layer, and formulas (5)-(13) are repeatedly calculated to obtain the interval intuitionistic fuzzy weighted average operator of the criterion layer, the index layer and the feature layer, and a summarized decision information matrix is constructed to complete the consistency check of the decision information matrix.

[0217] In the embodiment, step S5 is specifically as follows:

[0218] When the decision information matrix summarized in step S4 satisfies the consistency check condition, the decision information interval score matrix is calculated.

[0219]

[0220] Wherein, the interval score of the IVIFWA operator constructed by the ith C and the jth C IVIFN is

[0221] The decision information interval score matrix is exponentiated to obtain the interval exponentiation matrix The expression is as follows:

[0222]

[0223] Wherein, the interval exponentiation score of the IVIFWA operator constructed by the ith C and the jth C IVIFN is

[0224] The interval priority weight of the ith C factor is calculated by the interval exponentiation score Get interval priority weight vector The expression is as follows:

[0225]

[0226] By interval priority weight vector Constructing a decision possibility matrix The expression is as follows:

[0227]

[0228] in, Calculated by formula (19), i C =1,2,...,n C ,j C =1,2,...,n C , the calculation expression is as follows:

[0229]

[0230] Here, max(·) means taking the maximum value. and The size relationship can be calculated by formula (3) and And the exact function is calculated by formula (4) and The specific evaluation rules are as follows:

[0231] when It is believed that when It is believed that when and It is believed that when and It is believed that when and It is believed that

[0232] Then the decision possibility matrix is ​​constructed by the interval priority weight vector Calculate the priority possibility, the expression is as follows:

[0233]

[0234] According to step S2, when the language metric is "equivalent", the membership and non-membership are 0.5, which is the intermediate value of the language metric conversion. For subsequent normalization processing, it is necessary to prevent the calculated probability from being less than 0, and 0.5 is added as a balance value.

[0235] Then the priority probability is normalized, the expression is as follows:

[0236]

[0237] Based on step S4, the summarized decision information matrix is substituted into the summarized decision information matrix of the criterion layer, the index layer and the feature layer respectively, and the normalized priority probability of the criterion layer, the index layer and the feature layer is calculated by repeating formula (14)-(21).

[0238] In this embodiment, according to the language measure comparison result described in Table 1, the language measure-number conversion table of expert 1 between the criterion layers can be obtained as shown in Table 7, and the calculation results are all rounded to two decimal places after the decimal point for display.

[0239] Table 7

[0240]

[0241] Similarly, according to Table 4, the language measure-number conversion table of expert 2 between the criterion layers can be obtained as shown in Table 8, and the calculation results are all rounded to two decimal places after the decimal point for display.

[0242] Table 8

[0243]

[0244] According to the language measure-number conversion table between the criterion layers of expert 1 and expert 2, the summarized interval decision information table between the criterion layers can be obtained by formula (11), as shown in Table 9, and the calculation results are all rounded to two decimal places after the decimal point for display.

[0245] Table 9

[0246]

[0247] The hesitation degree of the summarized interval decision information table between the criterion layers can be obtained by formula (12), as shown in Table 10, and the calculation results are all rounded to two decimal places after the decimal point for display.

[0248] Table 10

[0249]

[0250] According to the obtained hesitation degree result and formula (13), the consistency test result of the criterion layer decision information can be calculated as CR criterion = 0.0416 < 0.1, that is, passing the consistency test.

[0251] In this embodiment, according to the results in Table 9, and formula (14) and formula (15), the criterion layer interval index matrix can be obtained, as shown in Table 11. The calculation results are rounded to two decimal places for display.

[0252] Table 11

[0253]

[0254] According to the interval indexation matrix of the criterion layer in Table 11 and equations (16)-(17), the interval priority weight vector of the criterion layer can be obtained: As shown in Table 12, the calculation results are rounded to two decimal places.

[0255] Table 12

[0256]

[0257] According to the criterion layer priority weight vector in Table 12 and formula (18) and formula (19), the decision possibility matrix is ​​constructed, and the priority possibility of the criterion layer is calculated according to formula (20) and formula (21): and normalized priority probability As shown in Table 13, the calculation results are rounded to two decimal places.

[0258] Table 13

[0259]

[0260] It can be seen from Table 13 that for the criterion layers C1-C4, their importance is ranked as follows: C2>C1>C3>>C4.

[0261] In summary, this embodiment also uses the comparison results of the criterion layer language metrics corresponding to each indicator layer in Table 2 and Table 5 to directly give the normalized priority possibility of the criterion layer corresponding to each indicator layer. The normalized priority probability of the indicator layer corresponding to each feature As shown in Tables 14 and 15, the calculation results are rounded to two decimal places for display. (The specific corresponding steps are consistent with steps S4-S5, so the detailed calculation steps are not listed here).

[0262] Table 14

[0263]

[0264] Table 15

[0265]

[0266] In this embodiment, step S6 is specifically as follows:

[0267] In this embodiment, the priority probability and normalized priority probability of the criterion layer obtained in step S5 are respectively and The priority possibility and normalized priority possibility of the indicator layer are and Then construct the weight information of the indicator layer-destination layer, As shown in Table 16, the calculation results are rounded to two decimal places.

[0268] Table 16

[0269]

[0270] According to the calculated weight information ω of the index layer-target layer target And the normalized priority probability of the feature layer Calculate the influence weight I of each feature of feature-purpose target,feature , the expression is as follows:

[0271]

[0272] Finally, the obtained influence weight is normalized to obtain the normalized influence weight The feature parameter ranking results are obtained by sorting by size, completing the feature parameter evaluation for multi-function radar working mode recognition. The feature layer priority probability is shown in Table 17. The calculation results are rounded to two decimal places for display.

[0273] Table 17

[0274]

[0275] As can be seen from Table 17, in this embodiment, the ranking result of the importance of the feature parameters evaluated by experts is: RF>PW>PRI>BW>PA>AOA≈DR.

[0276] In summary, the method of the present invention uses the interval-valued intuitionistic fuzzy-hierarchy analysis method to evaluate the characteristic parameters of the multi-function radar working mode recognition, which provides a theoretical basis and rule guidance for the parameter selection of the multi-function radar working mode recognition in different scenarios, is conducive to accurate parameter selection, improves the reliability of the recognition results, and provides data support for subsequent research.

[0277] Those skilled in the art will appreciate that the embodiments described herein are presented for purposes of illustration and that the inventive principles are not limited to these particular embodiments. Other variations and modifications can be made to the embodiments without departing from the spirit and scope of the inventive principles.

Claims

1. A characteristic parameter evaluation method for multi-function radar working mode recognition, the specific steps are as follows: S1. Constructing a hierarchical structure for selecting and evaluating feature parameters for multi-function radar working mode recognition; S2. Construct an interval-valued intuitionistic fuzzy linguistic measurement scale for multi-function radar working mode recognition; S3. Identify experts in relevant fields, obtain their evaluation information on the characteristic parameters of multi-function radar working mode recognition, and construct an expert preference scale by summarizing the information; Identify experts in the field of multi-function radar and ask them to evaluate the impact of multiple feature parameters on multi-function radar working mode recognition through a questionnaire. The evaluation level is shown in the language measurement table in step S2. Finally, language measurement comparison results of the multiple feature parameters for multi-function radar working mode recognition at the criterion level, indicator level, and feature level are obtained, and the information is summarized to obtain an expert preference scale. S4. Calculate the interval-valued intuitionistic fuzzy weighted average operator based on the language measurement comparison results obtained in step S3, construct a summary decision information matrix, and complete the consistency test of the decision information matrix; The language measurement comparison results include: language measurement comparison results at the criterion level, indicator level, and feature level; S5. Calculate the interval score matrix and interval index matrix of the decision information, construct the priority weight vector, construct the decision possibility matrix, and calculate the normalized priority possibility; The normalized priority possibility includes: the normalized priority possibility of the criterion layer, the indicator layer, and the feature layer; S6. Based on the priority possibility of the criterion layer, indicator layer, and feature layer obtained in step S5, the influence of the feature layer on the target layer is calculated, and the feature parameter sorting result is obtained by sorting the size, thereby completing the feature parameter evaluation of the multi-function radar working mode recognition; The step S6 is specifically as follows: The priority probability and normalized priority probability of the criterion layer obtained in step S5 are respectively and The priority possibility and normalized priority possibility of the indicator layer are and Then construct the weight information of the indicator layer-target layer, According to the calculated weight information ω of the index layer-target layer target And the normalized priority probability of the feature layer Calculate the influence weight I of each feature of feature-purpose target,feature , the expression is as follows: Finally, the obtained influence weight is normalized to obtain the normalized influence weight Perform size sorting to obtain the characteristic parameter sorting results and complete the characteristic parameter evaluation of the multi-function radar working mode recognition.

2. The characteristic parameter evaluation method for multi-function radar working mode recognition according to claim 1 is characterized in that: The step S1 is specifically as follows: Constructing the characteristic parameter selection evaluation hierarchy structure to achieve a comprehensive evaluation of the characteristic parameters to be evaluated, the characteristic parameter selection evaluation hierarchy structure comprising: a target layer, a criterion layer, an indicator layer, and a characteristic layer; The target layer is the impact assessment of characteristic parameters for multi-function radar working mode identification; The criterion layers include: generality C1, accuracy C2, reliability C3, and complexity C4; The indicator layer constructs 12 specific factors as the indicator layer based on the criterion layer, as follows: 1) Feature Clear Relevance C11: This is measured by whether the feature parameter has a clear physical meaning or relevance for working mode recognition. If there is a clear physical meaning or relevance, then the feature has clear relevance. 2) Result interpretability C12: This is measured by whether the different working modes described by the characteristic parameters are easy to explain and interpret. The easier it is to explain, the higher the result interpretability. 3) Feature Platform Adaptability C13: This is measured by whether the feature parameter can be effectively used on different detection platforms, including single detection sites and distributed detection sites. The wider the range of platforms that can be used, the higher the feature platform adaptability. 4) Feature Accuracy C21: The mean square error of the feature parameter extracted by the detection system under a fixed received signal-to-noise ratio is used as a metric. The fixed received signal-to-noise ratio is 13dB. The smaller the mean square error, the higher the feature accuracy. 5) Feature discrimination C22: This is measured by whether the feature parameter can successfully obtain the working mode results when describing different working modes. The more accurate the result, the higher the discrimination. 6) Feature difference C23: This is measured by whether the feature parameter can generate multiple working mode recognition results when describing different working modes. The more results it generates, the lower the difference. 7) Feature stability C31: The error rate of the working mode recognition result of the feature parameter under different typical working environments is used as a measure. The higher the error rate, the lower the stability; 8) Feature noise robustness C32: The variance of the mean square error of the feature parameter extracted by the detection system under complex conditions with low signal-to-noise ratio is used as a metric. The higher the variance, the lower the feature noise robustness; 9) Feature interference robustness C33: The variance of the mean square error of the feature parameter extracted by the detection system under complex external interference conditions is used as a metric. The higher the variance, the lower the feature interference robustness; 10) Feature degradation robustness C34: The variance of the mean square error of the feature parameter extracted by the detection system under complex feature degradation conditions is used as a metric. The higher the variance, the lower the feature degradation robustness. 11) Time occupation of feature calculation C41: The average time required by the detection system to extract the feature parameter under different receiving signal-to-noise ratio conditions is used as a metric. The lower the occupation, the higher the evaluation; 12) Resource occupancy of feature calculation C42: This is measured by the resources occupied by the detection system to extract the feature parameters under different receiving signal-to-noise ratio conditions. The lower the occupancy, the higher the evaluation; The feature layer lists detailed characteristic parameters that need to be evaluated, including carrier frequency RF, pulse repetition interval PRI, and pulse width PW; Among them, the specific parameters required for the feature layer are selected according to actual conditions.

3. The characteristic parameter evaluation method for multi-function radar working mode recognition according to claim 1 is characterized in that: The step S2 is specifically as follows: Suppose there is a characteristic parameter set X=[x1,x2,...,x n ]; Among them, x i , i=1,2,...,n represents the i-th characteristic parameter, and n represents the total number of characteristic parameters; For the characteristic parameter x i There is an interval-valued intuitionistic fuzzy set IVIFS A, which is expressed as follows: in, Represent the characteristic parameters x i For the membership interval and non-membership interval of the feature parameter set X, They represent the lower and upper limits of the membership degree, They represent the lower and upper limits of non-membership respectively; The ambiguity interval π is calculated by formula (3): A (x i ), the expression is as follows: in, denote the lower and upper limits of the fuzziness, respectively; For IVIFSA in formula (2), when X = [x] has only one element, A is called interval-valued intuitionistic fuzzy number IVIFN. At this time, its score function S(A) and exact function H(A) are calculated as follows: The interval-valued intuitionistic fuzzy linguistic measurement needs to correspond to three quantitative results: membership μ, non-membership υ, and fuzziness π. The interval-valued intuitionistic fuzzy linguistic measurement table for the multi-function radar working mode recognition feature parameters is constructed, specifically including: When the language measure A is extremely low (EL), the membership μ = [0.10, 0.25] and the non-membership υ = [0.65, 0.75]; When the language measure A is low (L), the membership μ = [0.15, 0.30] and the non-membership υ = [0.60, 0.70]; When the language measure A is low (ML), the membership μ = [0.20, 0.35] and the non-membership υ = [0.55, 0.65]; When the language measure A is low (SL), the membership μ = [0.25, 0.40], and the non-membership υ = [0.50, 0.60]; When the language measure A is general (AE), the membership μ = [0.45, 0.55], and the non-membership υ = [0.30, 0.45]; When the language measure A is equivalence (EE), the membership μ = [0.50, 0.50], and the non-membership υ = [0.50, 0.50]; When the language measure A is high (SH), the membership μ = [0.50, 0.60], and the non-membership υ = [0.25, 0.40]; When the language measure A is high (MH), the membership μ = [0.55, 0.65] and the non-membership υ = [0.20, 0.35]; When the language measure A is high (H), the membership μ = [0.60, 0.70] and the non-membership υ = [0.15, 0.30]; When the language measure A is extremely high (EH), the membership μ = [0.65, 0.75] and the non-membership υ = [0.10, 0.25]; In the language measurement table, the membership and non-membership are set to not have extreme values ​​of 0 or 1, and the fuzziness of the language measurement A is fixedly controlled at π=[0.00, 0.25].

4. The characteristic parameter evaluation method for multi-function radar working mode recognition according to claim 3 is characterized in that: The step S4 is specifically as follows: Set the IVIFS obtained based on the language measurement comparison results The xth C Intuitionistic fuzzy numbers The expression is as follows: in, Respectively represent the lower limit of membership, upper limit of membership, lower limit of non-membership, and upper limit of non-membership of the IVIFN; n C Indicates that the interval intuitionistic fuzzy set contains n C factors; Defining IVIFS The interval-valued intuitionistic fuzzy weighted IVIFW operator expression is as follows: in, express The weight vector of and The IVIFS operation criteria are as follows: There are two IVIFS settings. The operation rule expression is as follows: Where λ represents the constant coefficient; Then formula (7) can be expressed as formula (10), which is as follows: like All are considered to have the same weight, then The IVIFW operator is the interval-valued intuitionistic fuzzy weighted average IVIFWA operator. In this case, Equation (7) can be expressed as Equation (11), which is as follows: By comparing the language measurement results obtained in formula (11) and step S3, the summary decision information matrix can be constructed The expression is as follows: in, Indicates the i C and jth C IVIFWA operator constructed by IVIFN; Then we can get Hesitation The expression is as follows: Then get The consistency test of hesitation is performed, and the expression is as follows: in, Represents the result of the consistency test, when It is considered to have passed the consistency test; Indicates that the matrix size is n C The consistency check threshold is as follows: When the matrix size is 1-2, When the matrix size is 3, When the matrix size is 4, When the matrix size is 5, When the matrix size is 6, When the matrix size is 7, When the matrix size is 8, When the matrix size is 9, Based on step S3, the language measurement comparison results are respectively substituted into the language measurement comparison results of the criterion layer, indicator layer, and feature layer, and equations (6)-(14) are repeated to calculate the interval intuitionistic fuzzy weighted average operators of the criterion layer, indicator layer, and feature layer, and construct a summarized decision information matrix to complete the consistency test of the decision information matrix.

5. The characteristic parameter evaluation method for multi-function radar working mode recognition according to claim 3 is characterized in that: The step S5 is specifically as follows: When the decision information matrix summarized in step S4 If the consistency test conditions are met, the decision information interval score matrix is ​​calculated The expression is as follows: Among them, the i C and jth C The interval score of the IVIFWA operator constructed by IVIFN is The decision information interval score matrix Exponentiation to obtain interval indexation matrix The expression is as follows: Among them, the i C and jth C The interval indexation score of the IVIFWA operator constructed by IVIFN is Calculate the i-th C Interval priority weights of factors Get interval priority weight vector The expression is as follows: By interval priority weight vector Constructing a decision possibility matrix The expression is as follows: in, Calculated by formula (20), i C =1,2,...,n C ,j C =1,2,...,n C , the calculation expression is as follows: Among them, max(·) means taking the maximum value; and The size relationship can be calculated by formula (4) and And the exact function is calculated by formula (5) and The specific evaluation rules are as follows: when It is believed that when It is believed that when and It is believed that when and It is believed that when and It is believed that Then the decision possibility matrix is ​​constructed by the interval priority weight vector Calculate the priority possibility, the expression is as follows: According to step S2, when the language metric is "equivalent", the membership and non-membership are 0.5, which is the middle value of the language metric conversion, and 0.5 is added as a balance value; Then the priority possibility is normalized and the expression is as follows: Based on step S4, the summarized decision information matrix is ​​substituted into the summarized decision information matrices of the criterion layer, indicator layer, and feature layer respectively, and equations (15)-(22) are repeated to calculate the normalized priority possibility of the criterion layer, indicator layer, and feature layer.

Images