A radar countermeasure reconnaissance signal identification method based on dynamic parameter registration

Abstract

A radar emitter signal recognition method based on dynamic parameter registration is provided, comprising the following steps: parameter input; mode selection; dynamic tolerance calculation; dynamic weight calculation; confidence calculation; result output. The method uses mathematical modeling and statistical analysis tools to intelligently analyze the change law of the reconnaissance signal data, and combines the sensor device and the regional characteristics to develop a radar emitter target adaptive recognition technology. Compared with the prior art, on the one hand, the present application can accurately realize intelligent recognition of the radar emitter signal by getting rid of the influence of reconnaissance data from different regions, different sensor devices and different acquisition parameters; on the other hand, since the present application adopts a dynamic weight and dynamic tolerance idea for radar emitter signal recognition, the heavy work of repeatedly adjusting the weight, tolerance and parameters by the operator is avoided, so the work load is greatly reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar reconnaissance and electronic information processing application, in particular to the optimization and improvement of a signal recognition method under complex electromagnetic environment, and more particularly to a radar countermeasure reconnaissance signal recognition method based on dynamic parameter registration. BACKGROUND

[0002] On the basis of completing signal sorting, effective recognition of radar emitter signals is one of the key technologies to improve electronic warfare effectiveness. Signal recognition is the process of comparing and matching the signal characteristic parameters obtained after sorting with the pre-accumulated emitter parameters to confirm the emitter and its related weapon system attributes. The recognition effect is closely related to factors such as feature parameter extraction and classification model construction. Reliable radar emitter recognition results can not only obtain information such as the system, purpose and working state of the counterparty radar, but also analyze its tactical characteristics, working rules and combat effectiveness, form valuable electronic reconnaissance intelligence, and lay a solid foundation for mastering electromagnetic power.

[0003] With the continuous development of electronic information technology, the conventional parameters of new system radars represented by phased arrays have the characteristics of variability and rapid change. The traditional template matching method does not consider this characteristic of modern radar emitter signals, and it is difficult to achieve ideal recognition accuracy by relying only on the measurement of conventional parameters. In recent years, the rapidly developing recognition method based on neural networks has attracted widespread attention, but the non-cooperation of radar countermeasures makes it difficult to obtain the pre-required training samples, and it is difficult to adapt to the requirements of the current complex electromagnetic environment. Therefore, how to realize the rapid and effective recognition of radar emitters in a complex electromagnetic environment has become a problem to be solved. SUMMARY

[0004] To overcome the deficiencies of the prior art, the present application provides a radar emitter signal recognition method based on dynamic parameter registration, the specific steps of which are as follows:

[0005] Step 1. Parameter input

[0006] The input parameters include the radar ID in the library and the radar style parameters in the library: carrier frequency RF, pulse repetition interval PRI, pulse width PW, intra-pulse modulation type MOP, the to-be-recognized emitter number and the style parameters of the to-be-recognized emitter: RF, PRI, PW, MOP, and default parameter values three parts; wherein

[0007] (1) The input of the recognition parameters RF, PRI and PW includes parameter variation type, parameter typical value, sorted style typical sample, parameter maximum value and parameter minimum value; the input of the recognition parameter MOP is the intra-pulse modulation type;

[0008] (2) The default parameters of the identification algorithm include: the maximum number of identified radars, the maximum number of identified patterns, the selection of the identification algorithm, the confidence calculation formula, the fixed tolerance of the identification parameters, the measurement error, and the fixed weight of the identification parameters;

[0009] Step 2. Mode selection

[0010] Two identification modes are adopted according to the complexity of the signal pattern, wherein:

[0011] (1) Mode 1: RF, PRI, and PW parameters are selected to participate in the calculation of the confidence level;

[0012] (2) Mode 2: RF, PRI, PW, and MOP are selected to participate in the calculation of the confidence level;

[0013] Step 3. Dynamic tolerance calculation

[0014] From the forward working principle of the radar, the noise is a random variable subject to uniform distribution, normal distribution, Rayleigh distribution, Weibull distribution, Rice distribution, or K distribution. The probability density function f(x) is calculated for different noise distribution cases;

[0015] For different noise distribution types, the parameter matching tolerance δ is calculated using the following formula

[0016]

[0017] Where n is the degree of freedom, which takes different values to represent different distributions; t 0.01 (n) represents the 99.9% confidence level determined by the t distribution, and the value is obtained from the table; S is the sample standard deviation, which is calculated by the following formula, (x1, x2, …, x N ) are the N parameter samples

[0018]

[0019] Step 4. Dynamic weight calculation

[0020] First, the subjective weight vector is obtained according to expert experience Where is the subjective weight of the carrier frequency parameter, is the subjective weight of the pulse repetition interval parameter, is the subjective weight of the pulse width parameter, is the subjective weight of the intra-pulse parameter; Then, the objective weight result vector is determined according to the different systems adopted by each radar emitter parameter under the working mode Where is the objective weight of the carrier frequency parameter, is the objective weight of the pulse repetition interval parameter, is the pulse width parameter objective weight, is the pulse parameter objective weight; the objective weight obtained by the objective weighting method is determined according to the specific steps for the four characteristic parameters, and the specific steps are as follows:

[0021] Step 1. Determine the weight of the pulse characteristic parameter

[0022]

[0023] wherein, is the pulse parameter objective weight, and if the pulse parameter in the mode is not modulated, 0 is taken; if the pulse parameter in the mode is modulated, 1 is taken;

[0024] Step 2. Determine the weight of the pulse width characteristic parameter

[0025]

[0026] wherein, is the pulse width parameter objective weight, and if the pulse width in the mode is fixed, 0.5 is taken; if the pulse width in the mode is other system, such as pulse width jitter system, 1 is taken;

[0027] Step 3. Determine the weight of the pulse repetition interval characteristic parameter

[0028]

[0029] wherein, is the pulse repetition interval parameter objective weight, and different values are determined according to different systems of the radar pulse interval of the identification library, if the radar of the identification library is frequency fixed, 1 is taken; if the radar of the identification library is other system, such as frequency difference system, 0.5 is taken;

[0030] Step 4. Determine the weight of the carrier frequency characteristic parameter

[0031]

[0032] wherein, is the carrier frequency parameter objective weight, and different values are determined according to different systems of the carrier frequency of the identification library, if the radar of the identification library is carrier frequency fixed, 1 is taken; if the radar of the identification library is other system, such as carrier frequency agile system, 0.5 is taken;

[0033] The subjective weighting result vector and the objective weighting result vector are combined by multiplication and normalization, that is,

[0034]

[0035] to obtain the dynamic weight vector; wherein, represents the dynamic weight of the pth parameter, Subjective weight value of the pth parameter, Objective weight value of the pth parameter, parameter p takes values 1, 2, …, M, respectively representing the characteristic parameters in the application, M represents the number of parameters: 3 in mode 1, 4 in mode 2;

[0036] Step 5. Confidence calculation

[0037] First, the matching confidence calculation formula corresponding to different types of characteristic parameters is given respectively:

[0038] a) Recognition confidence calculation of scalar type characteristic parameters

[0039] Let the parameter value of the pth characteristic of the i th mode of a certain template radar be f ip , the measurement deviation be Δf ip , the mean square error be σ ip , and the measured value of the pth characteristic of the to-be-measured radar radiation source be x p , then the recognition confidence f(x p ) of the pth characteristic of the radiation source is obtained by using fuzzy set theory:

[0040]

[0041] Among them, the measurement deviation Δf ip is determined by the measurement accuracy of the reconnaissance equipment, and the mean square error σ ip is determined by the measurement noise and system noise.

[0042] b) Recognition confidence calculation of interval type characteristic parameters

[0043] Let the parameter value of the pth characteristic of the i th mode of a certain template radar be , and the lower limit and upper limit of the parameter value be , respectively, then the recognition confidence of the pth characteristic of the radiation source can be obtained by using fuzzy set theory:

[0044]

[0045] c) Recognition confidence calculation of sequence type characteristic parameters

[0046] Let the parameter value of the pth characteristic of the i th mode of a certain template radar be a sequence form , and the possible values of the pth characteristic parameter in the i th mode be , respectively, and the measured value of the pth characteristic of the to-be-measured radar radiation source also be a sequence form , respectively, and the recognition confidence of the pth characteristic of the radiation source is obtained by taking "XOR":

[0047]

[0048]

[0049] where β(·) is a decision function;

[0050] For the intrapulse feature parameter type, its confidence is simplified to the operation matching, and "0" represents no intrapulse modulation type, "1" represents linear frequency modulation, "2" represents binary coding, and "4" represents quad coding. The parameter value f ip ∈{0, 1, 2, 4}, and the parameter of the radar emitter to be tested is represented as x p The recognition confidence of the parameter can be obtained by using "XOR" and "NOT":

[0051] f(x p ) = NOT(XOR(f ip , x p )) (18)

[0052] The following four cases can be divided when calculating the confidence:

[0053] (1) Fixed tolerance, fixed weight;

[0054] 1) The confidence f(x p ) of each parameter is calculated respectively. The confidence of RF, PRI, and PW is calculated according to the typical value or interval of the radar pattern to be identified test and the radar pattern ku in the library, the given tolerance, and the device measurement error, and the specific formula is shown in formula (14-16). The confidence of MOP is calculated according to the modulation type of test and ku, and the specific formula is shown in formula 18;

[0055] 2) The weight of the pth feature of the radar emitter to be tested is w p The confidence of each parameter is weighted and averaged by using formula 19 to obtain the confidence F of the matching of test and ku. Here, the fixed weight is considered, so the subjective weight determined by human is used to determine;

[0056]

[0057] where F is the confidence of the matching of test and ku, represents the subjective weight of the pth feature of the radar emitter to be tested;

[0058] (2) Dynamic tolerance, fixed weight;

[0059] 1) The confidence f(x p) ; wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern test and ku in the library, dynamic tolerance and equipment measurement error, see formula (14-16) ; the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18; the dynamic tolerance of parameters RF, PRI, PW is calculated according to the typical sample corresponding to each parameter, see step 3 for specific calculation method, and then compared with the given tolerance to take the smaller value between the two, that is, the calculated dynamic tolerance is compared with the artificially given fixed tolerance, and the smaller value between the two is taken;

[0060] 2) The confidence of each parameter is weighted and averaged by formula 19 to obtain the confidence F of the matching of test and ku; here, the fixed weight is considered, so the subjective weight determined by human being ;

[0061] (3) Fixed tolerance, dynamic weight;

[0062] 1) Calculate the confidence f(x p ) of each parameter respectively; wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern test and ku in the library, given tolerance and equipment measurement error, see formula (14-16) ; the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18;

[0063] 2) First, calculate the objective weight of parameters RF, PRI, PW, MOP , and then combine the given subjective weight , use formula 13 to comprehensively calculate the dynamic weight of each parameter , see step 4 for specific calculation method;

[0064] 3) The confidence of each parameter is weighted and averaged by formula 20 to obtain the confidence F of the matching of test and ku;

[0065]

[0066] In the formula, , which represents the dynamic weight of the pth parameter of the radar emitter to be measured;

[0067] (4) Dynamic tolerance, dynamic weight;

[0068] 1) Calculate the confidence f(x p); wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern ku in the library and the test to be identified, the dynamic tolerance and the equipment measurement error, see formula (14-16); the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18; the dynamic tolerance of the parameters RF, PRI, PW is calculated according to the typical sample corresponding to each parameter, see step 3 for specific calculation method, and then compared with the given tolerance to take the smaller value between the two, that is, the calculated dynamic tolerance is compared with the artificially given fixed tolerance, and the smaller value between the two is taken;

[0069] 2) First, calculate the objective weight of the parameters (RF, PRI, PW, MOP) Then combine the given subjective weight The dynamic weight of each parameter is calculated by formula 13 See step 4 for specific calculation method;

[0070] 3) The confidence of each parameter is weighted and averaged by formula 20 to obtain the confidence F of the matching of test and ku;

[0071] Step 6. Result output

[0072] The output result includes the radiation source identification result: the pattern serial number corresponding to the radiation source, the radar ID in the library identified and the corresponding confidence, the pattern identification result: the serial number of the radar pattern in the library identified and the corresponding confidence.

[0073] In a specific embodiment of the present application, in step 4, a set of subjective weight vectors W a =(0.25, 0.25, 0.2, 0.3) is obtained according to expert experience.

[0074] The present application provides a new signal identification method based on dynamic parameter registration, which intelligently analyzes the change rule of reconnaissance signal data by using mathematical modeling and statistical analysis tools, and develops radar radiation source target adaptive identification technology combined with sensor devices and regional characteristics. On the one hand, the present application can accurately realize radar radiation source signal identification by getting rid of the influence of reconnaissance data from different regions, different sensor devices and different collection parameters; on the other hand, since the present application adopts dynamic weight and dynamic tolerance for radar radiation source signal identification, the heavy work of repeatedly adjusting weight, tolerance and parameters by operators is avoided, and the workload is greatly reduced.

[0075] The new method provided by the application can dynamically calculate the tolerance and weight of the radar radiation source signal characteristic parameters in the carrier frequency (RF), pulse repetition interval (PRI), pulse width (PW) and intra-pulse modulation type (MOP) information, and then obtain the confidence of the to-be-identified radiation source pattern, so as to provide a basis for making a judgment decision in the next step, while reducing the algorithm complexity, and effectively improving the identification ability of the radar radiation source signal in a complex electromagnetic environment. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 The overall workflow of the application is shown.

[0077] Figure 2 The calculation block diagram of the dynamic tolerance is shown. DETAILED DESCRIPTION

[0078] The application will be further described below in combination with embodiments and drawings.

[0079] The application is a radar countermeasure reconnaissance signal identification method based on dynamic parameter registration, the basic idea of which is to respectively match each pattern of the to-be-identified radiation source with each pattern corresponding to the radar ID in the library, calculate the confidence by introducing dynamic tolerance and weight, and realize effective identification of the radar radiation source signal in a complex electromagnetic environment. The radar radiation source signal identification parameters involved in the application mainly include the carrier frequency (RF), pulse repetition interval (PRI), pulse width (PW) and intra-pulse modulation type (MOP). The specific working process of the radar radiation source signal identification algorithm based on dynamic parameter registration is shown in the accompanying drawings, and the specific steps of implementing the method of the application are as follows. Figure 1

[0080] Step 1. Parameter input

[0081] The input parameters include three parts: the radar ID in the library and the radar pattern parameters (RF, PRI, PW, MOP) in the library, the to-be-identified radiation source number and the pattern parameters (RF, PRI, PW, MOP) of the to-be-identified radiation source, and the default parameter values. Among them:

[0082] (1) The input of the identification parameters RF, PRI and PW includes the parameter variation type, the parameter typical value, the sorted pattern typical sample, the parameter maximum value and the parameter minimum value; the input of the identification parameter MOP is the intra-pulse modulation type.

[0083] (2) The default parameters of the identification algorithm mainly include: the maximum number of the identified radars output; the maximum number of the identified patterns output; the selection of the identification algorithm; the confidence calculation formula; the fixed tolerance of the identification parameters, the measurement error; the fixed weight of the identification parameters, etc.

[0084] Step 2. Mode selection ​

[0085] Two recognition modes are taken according to the complexity of the signal pattern, wherein:

[0086] (1) Mode 1: RF, PRI, and PW are selected to participate in the calculation of the confidence degree;

[0087] (2) Mode 2: RF, PRI, PW, and MOP are selected to participate in the calculation of the confidence degree.

[0088] Specifically, mode 1 means that only RF, PRI, and PW are processed in the subsequent algorithm, and is mainly used for the recognition of a simple signal pattern; mode 2 means that RF, PRI, PW, and MOP are all processed in the subsequent algorithm, and is mainly used for the recognition of a complex signal pattern. The difference between mode 1 and mode 2 is only in the types of the parameters to be processed, and the subsequent processing methods are the same.

[0089] Step 3. Dynamic tolerance calculation

[0090] For different parameters, a corresponding parameter variation type function expression relationship is established, the standard deviation corresponding to different parameters is obtained by using a statistical analysis tool, and then the tolerance required for matching different parameters is obtained by using different noise distribution types. If the device measurement error is provided, the minimum value between the tolerance and the measurement error is taken as the final tolerance setting value. The dynamic tolerance calculation workflow is shown in FIG. 8. Figure 2

[0091] According to the principle of radar forward work, the noise is a random variable subject to uniform distribution, normal distribution, Rayleigh distribution, Weibull distribution, Rice distribution, or K distribution. Taking the PRI jitter variation type as an example, the tolerance calculation formula under different noise distribution conditions is briefly described below (other parameters such as RF, PW, etc. can be calculated to match the tolerance in a similar manner, and x is a random variable, μ is the mean, and σ is the standard deviation in the following expression)

[0092] (1) Uniform distribution model

[0093] The probability density function is:

[0094]

[0095] In the formula, a and b are constant parameters.

[0096] (2) Normal distribution model

[0097] The probability density function is:

[0098]

[0099] (3) Rayleigh distribution model ​

[0100] The probability density function is:

[0101]

[0102] (4) Weibull distribution model

[0103] The probability density function is:

[0104]

[0105] where λ>0 is the scale parameter and k>0 is the shape parameter.

[0106] (5) Rice distribution model

[0107] The probability density function is:

[0108]

[0109] where R is the sinusoidal (cosine) signal plus narrowband Gaussian random signal envelope, parameter A is the peak value of the main signal amplitude, and I0(·) is the modified 0th order Bessel function.

[0110] (6) K distribution model

[0111] The probability density function is:

[0112]

[0113] where λ>0 is the scale parameter, Γ(·) is the gamma function, and I k (·) is the modified kth order Bessel function.

[0114] For the above different noise distribution types, the parameter matching tolerance δ is calculated using the following formula

[0115]

[0116] where n is the degree of freedom, and different values represent different distributions (for example, n=∞ represents a normal distribution); t 0.01 (n) represents the 99.9% confidence level determined by the t distribution, which can be obtained from a table; S is the sample standard deviation, which can be calculated by the following formula, (x1, x2, …, x N ) are the N parameter samples of the sampling

[0117]

[0118] Step 4. Dynamic weight calculation

[0119] First, the subjective weight vector is obtained according to expert experience where is the subjective weight of the carrier frequency parameter is the subjective weight of the pulse repetition interval parameter, is the subjective weight of the pulse width parameter, is the subjective weight of the pulse-in-pulse parameter. The objective weight vector is determined according to different systems of each parameter in the working mode of each radar radiation source wherein is the objective weight of the carrier frequency parameter, is the objective weight of the pulse repetition interval parameter, is the objective weight of the pulse width parameter, is the objective weight of the pulse-in-pulse parameter. In the present application, a set of subjective weight vectors W a =(0.25, 0.25, 0.2, 0.3) is obtained according to the experience of experts. The objective weight obtained by the objective weight method is determined according to the specific steps for the four characteristic parameters, and the specific steps are as follows:

[0120] Step 1. Determine the weight of the pulse-in-pulse characteristic parameter

[0121]

[0122] wherein, is the objective weight of the pulse-in-pulse parameter, which is 0 if the pulse-in-pulse parameter in the mode is not modulated, and is 1 if the pulse-in-pulse parameter in the mode is modulated.

[0123] Step 2. Determine the weight of the pulse width characteristic parameter

[0124]

[0125] wherein, is the objective weight of the pulse width parameter, which is 0.5 if the pulse width in the mode is fixed, and is 1 if the pulse width in the mode is other systems such as pulse width jitter system.

[0126] Step 3. Determine the weight of the pulse repetition interval characteristic parameter

[0127]

[0128] wherein, is the objective weight of the pulse repetition interval parameter, which is determined according to different values of different systems of the radar pulse interval in the recognition library, and is 1 if the radar in the recognition library is a fixed frequency, and is 0.5 if the radar in the recognition library is other systems such as frequency difference system.

[0129] Step 4. Determine the weight of the carrier frequency characteristic parameter

[0130]

[0131] wherein, is the carrier frequency parameter objective weight value, different values are determined according to different systems of the identification library radar carrier frequency, if the identification library radar is carrier frequency fixed, 1 is taken; if the identification library radar carrier frequency is other system, such as carrier frequency agile system, 0.5 is taken.

[0132] The subjective weighting result vector and the objective weighting result vector are subjected to a multiplication synthesis normalization method, that is,

[0133]

[0134] The dynamic weight vector is obtained. In the formula, represents the dynamic weight of the pth parameter, represents the subjective weight value of the pth parameter, represents the objective weight value of the pth parameter, the parameter p takes values 1, 2, …, M, respectively representing the characteristic parameters (carrier frequency (RF), pulse repetition interval (PRI), pulse width (PW), intra-pulse parameter (MOP)) in the application, and M represents the number of parameters (3 in mode 1 and 4 in mode 2).

[0135] Step 5. Confidence calculation

[0136] First, the matching confidence calculation formula corresponding to different types of characteristic parameters is given:

[0137] a) Identification confidence calculation of scalar type characteristic parameter

[0138] Let the parameter value of the pth feature of the i th mode of a template radar be f ip , the measurement deviation be Δf ip , the mean square error be σ ip , and the measured value of the pth feature of the measured radar emitter be x p , then the identification confidence f(x p ) of the pth feature of the emitter can be obtained by using fuzzy set theory (Yang Sunmin, Cui Jing, Zhao Deyong, Liu Qiming. Research on fast identification algorithm of radar emitter signal based on rough set theory and SVM [J]. Electronic technology and software engineering, 2022 (13): 134-138.):

[0139]

[0140] Wherein, the measurement deviation Δf ip is determined by the measurement accuracy of the reconnaissance equipment, the mean square error σ i p is determined by the measurement noise and system noise.

[0141] b) Identification confidence calculation of interval type characteristic parameter

[0142] Let the parameter value of the pth feature of the i th mode of a template radar be f ip , the measurement deviation be Δf ip , the mean square error be σ ip , and the measured value of the pth feature of the measured radar emitter be x p , then the identification confidence f(x p ) of the pth feature of the emitter can be obtained by using fuzzy set theory (Yang Sunmin, Cui Jing, Zhao Deyong, Liu Qiming. Research on fast identification algorithm of radar emitter signal based on rough set theory and SVM [J]. Electronic technology and software engineering, 2022 (13): 134-138.): The lower limit and upper limit of the parameter value, respectively (here, the characteristic parameter is expressed in interval form, i.e., the parameter value range is an interval), and the identification confidence of the pth characteristic of the radiation source can be obtained by using fuzzy set theory:

[0143]

[0144] c) Identification confidence calculation of sequence type characteristic parameter

[0145] Let the parameter value of the pth characteristic of the ith mode of a template radar be in sequence form The possible values of the pth characteristic parameter of the ith mode are The parameter measurement value of the pth characteristic of the to-be-detected radar radiation source is also in sequence form The r measurement values of the parameter p are The identification confidence of the pth characteristic of the radiation source can be obtained by using "XOR":

[0146]

[0147]

[0148] In the formula, β(·) is a judgment function.

[0149] For the intra-pulse characteristic parameter type, its confidence calculation can be simplified to operation matching. Set "0" to represent no intra-pulse modulation type, "1" to represent linear frequency modulation, "2" to represent binary coding, and "4" to represent quadrature coding. Let the parameter value f ip ∈{0,1,2,4} of the template radar, and the parameter x p of the to-be-detected radar radiation source, then the identification confidence of the parameter can be obtained by using "XOR" and "NOT":

[0150] f(x p )=NOT(XOR(f ip ,x p )) (18)

[0151] The following four cases can be divided when calculating the confidence:

[0152] (1) Fixed tolerance and fixed weight.

[0153] 1) Calculate the confidence f(x p ) of each parameter, respectively. The confidences of RF, PRI, and PW are calculated according to the typical values or intervals of the to-be-identified radiation source style (test) and the radar style (ku) in the library, the given tolerance, and the device measurement error, and the specific formula is shown in formula (14-16). The confidence of MOP is calculated according to the modulation type of test and ku, and the specific formula is shown in formula 18;

[0154] 2) Set the weight of the pth feature of the radar emitter under test as w p , the confidence of test and ku matching is obtained by weighted average of each parameter confidence using formula 19. Here, the fixed weight is considered, so the subjective weight determined by human being is used.

[0155]

[0156] In the formula, F is the confidence of test and ku matching, represents the subjective weight of the pth feature of the radar emitter under test.

[0157] (2) Dynamic tolerance, fixed weight.

[0158] 1) Calculate the confidence of each parameter f(x p ) respectively. The confidence of RF, PRI, PW is calculated according to the typical value or interval of test and ku, dynamic tolerance and equipment measurement error, see formula (14-16). The confidence of MOP is calculated according to the modulation type of test and ku, see formula 18. The dynamic tolerance of parameter (RF, PRI, PW) is calculated according to the typical sample corresponding to each parameter, see step 3 for specific calculation method. Then, the calculated dynamic tolerance is compared with the given tolerance to take the smaller value between them, that is, the calculated dynamic tolerance is compared with the given fixed tolerance, and the smaller value between them is taken;

[0159] 2) The confidence of each parameter is weighted and averaged using formula 19 to obtain the confidence F of test and ku matching. Here, the fixed weight is considered, so the subjective weight determined by human being is used.

[0160] (3) Fixed tolerance, dynamic weight.

[0161] 1) Calculate the confidence of each parameter f(x p ) respectively. The confidence of RF, PRI, PW is calculated according to the typical value or interval of test and ku, given tolerance and equipment measurement error, see formula (14-16). The confidence of MOP is calculated according to the modulation type of test and ku, see formula 18;

[0162] 2) First, calculate the objective weight of parameter (RF, PRI, PW, MOP) Then, combine the given subjective weight Use formula 13 to comprehensively calculate the dynamic weight of each parameter See step 4 for specific calculation method;

[0163] ​3) The confidence of each parameter is weighted and averaged using formula 20 to obtain the confidence F of the test and ku matching.

[0164]

[0165] In the formula, represents the dynamic weight of the pth parameter of the radar emitter to be measured.

[0166] (4) Dynamic tolerance, dynamic weight.

[0167] 1) Calculate the confidence f(x p ) of each parameter respectively. The confidence of RF, PRI, and PW is calculated according to the typical value or interval of the radar pattern to be identified (test) and the radar pattern in the library (ku), dynamic tolerance, and device measurement error, as shown in formulas (14-16). The confidence of MOP is calculated according to the modulation type of test and ku, as shown in formula 18. The dynamic tolerance of the parameters (RF, PRI, PW) is first calculated according to the typical sample corresponding to each parameter, and the specific calculation method is shown in step 3. Then, the smaller value between the given tolerance and the calculated dynamic tolerance is taken, that is, the calculated dynamic tolerance is compared with the artificially given fixed tolerance, and the smaller value between the two is taken;

[0168] 2) First, calculate the objective weight of the parameters (RF, PRI, PW, MOP) Then, combine the given subjective weight Use formula 13 to calculate the dynamic weight of each parameter The specific calculation method is shown in step 4.

[0169] 3) The confidence of each parameter is weighted and averaged using formula 20 to obtain the confidence F of the test and ku matching.

[0170] Step 6. Result output

[0171] The output results of the algorithm include the emitter identification results (the pattern serial number corresponding to the emitter, the identified radar ID in the library, and the corresponding confidence), and the pattern identification results (the serial number of the identified radar pattern in the library and the corresponding confidence).

Claims

1. A method of radar emitter signal identification based on dynamic parameter registration, characterized by, The specific steps are as follows: Step 1. Parameter input The input parameters include radar ID in the library and radar style parameters in the library: carrier frequency RF, pulse repetition interval PRI, pulse width PW, intra-pulse modulation type MOP, to-be-identified radiation source number and to-be-identified radiation source style parameters: RF, PRI, PW, MOP, and default parameter values; Wherein (1) The input of identification parameters RF, PRI and PW includes parameter variation type, parameter typical value, sorting style typical sample, parameter maximum value and parameter minimum value; The input of identification parameter MOP is intra-pulse modulation type; (2) The default parameters of the identification algorithm include: the maximum number of identified radars, the maximum number of identified styles, the selection of the identification algorithm, the confidence calculation formula, the fixed tolerance of the identification parameter, the measurement error, and the fixed weight of the identification parameter; Step 2. Mode selection Two identification modes are adopted according to the complexity of the signal style, wherein: (1) Mode 1: RF, PRI and PW are selected to participate in the confidence calculation; (2) Mode 2: RF, PRI, PW and MOP are selected to participate in the confidence calculation; Step 3. Dynamic tolerance calculation From the radar forward working principle, the noise is a random variable subject to uniform distribution, normal distribution, Rayleigh distribution, Weibull distribution, Rice distribution or K distribution. The probability density function f(x) is calculated for different noise distribution cases; For different noise distribution types, the parameter matching tolerance δ is calculated using the following formula where n is the degree of freedom, and takes different values to represent different distributions; t 0.01 (n) represents the 99.9% confidence level determined by the t distribution, and the value is obtained from a table; S is the sample standard deviation, which is calculated by the following formula, (x1, x2, …, x N N parameter samples Step 4. Dynamic weight calculation Firstly, the subjective weight vector is obtained according to the experience of experts Wherein is the subjective weight of the carrier frequency parameter, is the subjective weight of the pulse repetition interval parameter, is the subjective weight of the pulse width parameter, is the subjective weight of the pulse-in parameter; and then the objective weight result vector is determined according to different systems of each parameter in the working mode of each radar radiation source Wherein is the objective weight of the carrier frequency parameter, is the objective weight of the pulse repetition interval parameter, is the objective weight of the pulse width parameter, is the objective weight of the pulse-in parameter; the objective weight obtained by the objective weighting method is determined according to the specific steps of the four characteristic parameters, and the specific steps are: Step 1. Determine the weight of the intra-pulse characteristic parameter wherein, is an intrapulse parameter objective weight, which is 0 if there is no modulation of the intrapulse parameter in the mode, and 1 if there is modulation of the intrapulse parameter in the mode. Step 2. Determine the weight of the pulse width characteristic parameter In the formula, is the pulse width parameter objective weight, and if the pulse width in the mode is fixed, 0.5 is taken; if the pulse width in the mode is other systems, such as a pulse width jitter system, 1 is taken. Step 3. Determine the weight of the pulse repetition interval characteristic parameter In the formula, is a pulse repetition interval parameter objective weight value, different values are determined according to different systems of the identification library radar pulse interval, if the identification library radar is a fixed frequency, then 1 is taken; if the identification library radar frequency is other systems, such as a frequency difference system, then 0.5 is taken. Step 4. Determine the weight of the carrier frequency characteristic parameter wherein, is a carrier frequency parameter objective weight, different values are determined according to different systems of the identification library radar carrier frequency, if the identification library radar is carrier frequency fixed, 1 is taken; if the identification library radar carrier frequency is other system, such as carrier frequency agile system, 0.5 is taken; The subjective weighting result vector and the objective weighting result vector are normalized by using the multiplication synthesis method, that is a dynamic weight vector is obtained; wherein, a dynamic weight of the pth parameter, a subjective weight value of the pth parameter, an objective weight value of the pth parameter, wherein p is 1, 2,...M, and M represents the number of parameters: 3 in mode 1 and 4 in mode 2. Step 5. Confidence calculation First, the matching confidence calculation formula corresponding to different types of characteristic parameters is given: a) Recognition confidence calculation of scalar type characteristic parameters Let the parameter value of the pth feature in the ith mode of a template radar be f ip , the measurement deviation be Δf ip , the mean square error be σ ip , and the measured parameter value of the pth feature of a radar emitter to be identified be x p . The identification confidence f(x p ) of the pth feature of the emitter is obtained by using fuzzy set theory. wherein the measurement deviation Δf ip The mean square deviation σ ip is determined by the measurement noise and the system noise; b) Recognition confidence calculation of interval type characteristic parameters Let the parameter value of the pth feature in the ith mode of a template radar be the lower limit and the upper limit of the parameter value, respectively. The identification confidence of the pth feature of the radiation source can be obtained by using fuzzy set theory: c) Recognition confidence calculation of sequence type characteristic parameters Let the parameter value of the pth feature in the ith mode of a template radar be in a sequence form The possible values of the pth feature parameter in the ith mode are The measured values of the pth feature of the to-be-detected radar radiation source are also in a sequence form The r measured values of the parameter p are respectively, and the identification confidence of the pth feature of the radiation source is obtained by "XOR": Wherein, β(·) is a judgment function; For the intrapulse characteristic parameter type, its confidence calculation is simplified as operation matching, and "0" represents no intrapulse modulation type, "1" represents linear frequency modulation, "2" represents binary coding, and "4" represents quad coding. The parameter value f of the template radar is set as f ip ∈{0, 1, 2, 4}, and the parameter of the to-be-tested radar radiation source is represented as x p Then, the identification confidence of the parameter can be obtained by using "exclusive OR" and "NOT". f(x p ) = NOT(XOR(f ip , x p )) (18) The following can be divided into the following four cases when calculating the confidence: (1) Fixed tolerance, fixed weight; 1) Calculate the confidence of each parameter f(x p ); wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern ku in the library and the test to be identified, the given tolerance and the device measurement error, see formulas (14-16); the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18; 2) Set the weight of the pth feature of the to-be-tested radar radiation source as w p The confidence of test and ku matching is obtained by weighted average of each parameter confidence using formula 19; here, the fixed weight is considered, so the subjective weight determined by human is determined; In the formula, F is the confidence degree of test and ku matching, represents the subjective weight of the pth feature of the radar radiation source to be measured; (2) Dynamic tolerance, fixed weight; 1) Calculate the confidence of each parameter f(x p ) respectively; wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern ku in the library and the test to be identified, dynamic tolerance and equipment measurement error, see formulas (14-16); the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18; the dynamic tolerance of parameters RF, PRI, PW is calculated according to the typical sample corresponding to each parameter, see step 3 for specific calculation method, and then compared with the given tolerance to take the smaller value between the two, that is, the calculated dynamic tolerance is compared with the artificially given fixed tolerance, and the smaller value between the two is taken; 2) Weighted average of each parameter confidence with formula 19 to get the confidence F of test and ku matching; here the fixed weight is considered, so the subjective weight decided by human is determined; (3) Fixed tolerance, dynamic weight; 1) Calculate the confidence of each parameter f(x p ); wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern ku in the library and the test to be identified, the given tolerance and the measurement error of the device, see formulas (14-16); the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18; 2) First, calculate the objective weight of parameters RF, PRI, PW, MOP Then, combine the given subjective weight Calculate the dynamic weight of each parameter by using formula 13 See step 4 for specific calculation method; 3) The confidence F of test and ku matching is obtained by weighted average of each parameter confidence using formula 20; In the formula, represents the dynamic weight of the pth parameter of the radar radiation source to be measured; (4) Dynamic tolerance, dynamic weight; 1) Calculate the confidence of each parameter f(x p ) respectively; wherein the confidence of RF, PRI, PW is calculated according to the typical value or interval of the radar pattern ku in the library and the test to be identified, dynamic tolerance and equipment measurement error, see formulas (14-16); the confidence of MOP is calculated according to the modulation type of test and ku, see formula 18; the dynamic tolerance of parameters RF, PRI, PW is calculated according to the typical sample corresponding to each parameter, see step 3 for specific calculation method, and then compared with the given tolerance to take the smaller value between the two, that is, the calculated dynamic tolerance is compared with the artificially given fixed tolerance, and the smaller value between the two is taken; 2) First, the objective weight of the parameters (RF, PRI, PW, MOP) is calculated and the given subjective weight is combined The dynamic weight of each parameter is calculated by using formula 13 The specific calculation method is shown in step 4; 3) The confidence F of test and ku matching is obtained by weighted average of each parameter confidence using formula 20; Step 6. Result output The output results include radiation source identification results: radiation source corresponding style serial number, identified library radar ID and corresponding confidence, style identification results: identified library radar style serial number and corresponding confidence.

2. The radar emitter signal identification method based on dynamic parameter registration of claim 1, wherein, In step 4, a set of subjective weight vectors W is obtained according to expert experience a = (0.25, 0.25, 0.2, 0.3).

Images