A data link interference identification method

Abstract

The application discloses a data link interference identification method, comprising the following steps: acquiring an interference signal to obtain time domain characteristic parameters of the interference signal; constructing an initial interference identification model according to the time domain characteristic parameters and by using an adaptive fuzzy neural network; training the initial interference identification model by using a learning algorithm to obtain an optimized interference identification model; and identifying interference of a data link under interference by using the optimized interference identification model. The application can accurately identify interference of a data link under interference based on time domain characteristic parameters of an interference signal, and has the characteristics of easy feature extraction, adaptability and self-learning.

Description

Technical Field

[0001] This invention relates to the field of data link communication technology, and in particular to a method for identifying data link interference. Background Technology

[0002] Data link communication is the main communication transmission method of C4ISR (command, control, communications, computers, intelligence, surveillance and reconnaissance) systems. With the help of data links, information flow from sensors to weapon launches and combat platforms can be communicated, thereby guiding weapon launches and realizing fire coordination among multiple platforms, and ultimately achieving seamless battlefield connectivity.

[0003] On future battlefields, the increasing prevalence of electromagnetic interference (including various natural and man-made interferences) is leading to a deteriorating electromagnetic environment. Therefore, electronic countermeasures capabilities are paramount in weapon performance indicators. For electronic countermeasures, the first priority is to prevent the enemy from detecting the operating frequency and waveform parameters of data link signals. Secondly, even if detected, the data link must possess various anti-jamming technologies to avoid interference and fulfill its tactical mission. Clearly, anti-jamming technology is a necessary condition for ensuring the normal operation of wartime data link communication systems, a crucial component of electronic warfare, and a significant factor influencing the course of war and even its outcome.

[0004] Because wartime electromagnetic interference can disrupt electronic equipment across the entire frequency band with precision down to the kilohertz and high tracking speed, and with the advent of intelligent jammers, data link communication equipment in future warfare must possess strong anti-jamming capabilities. Therefore, it is essential to adopt communication systems with strong anti-jamming capabilities and effective measures to enhance the anti-jamming capabilities of data link equipment to cope with the complex electromagnetic environment of wartime. However, multipath effects, electromagnetic compatibility, delay spread, and multiple access interference in communication, especially malicious interference from the enemy, are highly detrimental to equipment.

[0005] Previous interference recognition techniques could acquire instantaneous signal information, construct feature parameter histograms, and search for statistical peak points as signal classification criteria. However, the classification and recognition results were not ideal under low signal-to-noise ratio conditions. Power spectrum accumulation methods can detect noisy FM interference signals, but their applicability is limited. Extracting complexity and box-dimensional parameters and applying threshold decision methods can also classify and identify interference signals, but threshold adjustment is difficult under different noise environments. To overcome these difficulties, new interference recognition methods are needed. Summary of the Invention

[0006] The purpose of this invention is to provide a data link interference identification method. Based on the time-domain feature parameters of the interference signal, it can accurately identify the interference in the data link and has the characteristics of easy feature extraction, adaptability and self-learning.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0008] A data link interference identification method, comprising:

[0009] Acquire the interference signal to obtain the time-domain characteristic parameters of the interference signal;

[0010] Based on the aforementioned time-domain feature parameters, an initial interference recognition model is constructed using an adaptive fuzzy neural network;

[0011] The initial interference identification model is trained using a learning algorithm to obtain an optimized interference identification model; and

[0012] The optimized interference identification model is used to identify interference in the interfered data link.

[0013] Optionally, the time-domain feature parameters include time-domain moment skewness, time-domain moment kurtosis, and envelope undulation.

[0014] Optionally, the time-domain moment skewness is calculated using the following formula:

[0015]

[0016] Where F1 represents the time-domain moment skewness; y(t) represents the interference signal; t represents time; M represents the period duration; u represents the mean of the interference signal; and σ represents the standard deviation of the interference signal.

[0017] The time-domain moment kurtosis is calculated using the following formula:

[0018]

[0019] Where F2 represents the time-domain moment kurtosis;

[0020] The envelope undulation is calculated using the following formula:

[0021]

[0022] Where F3 represents the envelope undulation; σ e The variance of the squared envelope of the interference signal; u e This represents the mean of the square of the interference signal envelope.

[0023] Optionally, the adaptive fuzzy neural network includes an input layer, a fully connected layer, a normalization layer, a fuzzification layer, an inference layer, and a defuzzification layer; and the step of constructing the initial interference recognition model includes:

[0024] The time-domain moment skewness, the time-domain moment kurtosis, and the envelope undulation are respectively denoted as the first input data, the second input data, and the third input data;

[0025] The first input data, the second input data, and the third input data are mapped to the first neuron group, the second neuron group, and the third neuron group, respectively, through the fully connected layer to obtain the first vector, the second vector, and the third vector.

[0026] The first vector, the second vector, and the third vector are mapped to the interval [0, 1] through the normalization layer to obtain the first normalized vector, the second normalized vector, and the third normalized vector;

[0027] The membership degrees of the first standardized vector, the second standardized vector, and the third standardized vector to different fuzzy subsets are calculated through the fuzzification layer to obtain the first membership value, the second membership value, and the third membership value.

[0028] Based on the first membership value, the second membership value, and the third membership value, calculate the excitation of the fuzzy rule corresponding to each node in the inference layer;

[0029] Based on the excitation of the fuzzy rule and the corresponding fuzzy rule, the actual output is calculated through the defuzzification layer calculation model.

[0030] Optionally, the first neuron group, the second neuron group, and the third neuron group each include a plurality of neurons.

[0031] Optionally, the first standardized vector is calculated using the following formula:

[0032]

[0033] Among them, s 1j f represents the first normalized vector; 1j Let a represent the first vector, and j∈[1, the total number of neurons in the first neuron group]; a1 and b1 represent the first parameter and the second parameter, respectively;

[0034] The second standardized vector is calculated using the following formula:

[0035]

[0036] Among them, s 2k f represents the second normalized vector;2k Let a represent the second vector, and k∈[1, the total number of neurons in the second neuron group]; a2 and b2 represent the third and fourth parameters, respectively;

[0037] The third standardized vector is calculated using the following formula:

[0038]

[0039] Among them, s 3l f represents the third normalized vector; 3l Let l represent the third vector, and l∈[1, the total number of neurons in the third neuron group]; a3 and b3 represent the fifth and sixth parameters, respectively.

[0040] Optionally, the first membership value is calculated using the following formula:

[0041] μ 1j =μ Aj (s 1j )

[0042] Where, μ 1j μ represents the first membership value. Aj represents the membership function of the first standardized variable; Aj represents the fuzzy subset of the time-domain moment skewness;

[0043] The second membership value is calculated using the following formula:

[0044] μ 2k =μ Bk (s 2k )

[0045] Where, μ 2k Indicates the second membership value; μ 2k Bk represents the membership function of the second standardized variable; Bk represents the fuzzy subset of the time-domain moment kurtosis.

[0046] The third membership value is calculated using the following formula:

[0047] μ 3l =μ Cl (s 3l )

[0048] Where, μ 3l Indicates the third membership value; μ 3l Cl represents the membership function of the third standardized variable; Cl represents the fuzzy subset of the envelope variability.

[0049] Optionally, the actual output of the model is calculated using the following formula:

[0050]

[0051] P=∑ωi p i

[0052] C=∑p i

[0053] Where y represents the actual output of the model; ω i Let p represent the excitation of the fuzzy rule corresponding to the i-th node in the inference layer, where i ∈ [1, total number of nodes in the inference layer]; i This represents the fuzzy rule corresponding to the i-th node in the inference layer.

[0054] Optionally, the membership functions of the first standardized variable, the second standardized variable, and the second standardized variable are all Gaussian functions.

[0055] Optionally, the learning algorithm employs a hybrid algorithm combining least squares and gradient descent.

[0056] Compared with the prior art, the present invention has at least one of the following advantages:

[0057] This invention provides a data link interference identification method, which constructs an initial interference identification model based on the time-domain feature parameters of the interference signal and using an adaptive fuzzy neural network; by training the initial interference identification model using a learning algorithm, an optimized interference identification model can be obtained, thereby accurately identifying interference in the interfered data link.

[0058] This invention maps temporal feature parameters with multidimensional data into vectors by building a fully connected layer. In the intermediate layer, a fuzzification layer and an inference layer are designed to generate fuzzy rules and their excitation strengths. Then, a hybrid algorithm is used to automatically update the membership function and fuzzy rules. After defuzzification, the output layer outputs the interference type and its confidence level. This invention is applied to the field of interference recognition and has the advantages of easy feature extraction, adaptability, and self-learning. Attached Figure Description

[0059] Figure 1 This is a flowchart of a data link interference identification method provided in an embodiment of the present invention;

[0060] Figure 2 This is a schematic diagram of an interference identification model in a data link interference identification method provided in an embodiment of the present invention;

[0061] Figure 3 This is an experimental diagram illustrating the accuracy of a data link interference identification method provided in an embodiment of the present invention. Detailed Implementation

[0062] The following detailed description, in conjunction with the accompanying drawings and specific embodiments, provides a further detailed explanation of the data link interference identification method proposed in this invention. The advantages and features of this invention will become clearer from the following description. It should be noted that the drawings are in a very simplified form and use non-precise proportions, used only to facilitate and clearly illustrate the embodiments of this invention. Please refer to the drawings to make the objectives, features, and advantages of this invention more apparent and understandable. It should be understood that the structures, proportions, sizes, etc., depicted in the accompanying drawings are only for illustrative purposes to aid those skilled in the art and are not intended to limit the implementation conditions of this invention. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in proportions, or adjustments to the size, without affecting the effects and objectives achieved by this invention, should still fall within the scope of the technical content disclosed in this invention.

[0063] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0064] Combination Figures 1-3 As shown, this embodiment provides a data link interference identification method, including: step S1, acquiring an interference signal to obtain the time-domain feature parameters of the interference signal; step S2, constructing an initial interference identification model based on the time-domain feature parameters and using an adaptive fuzzy neural network; step S3, training the initial interference identification model using a learning algorithm to obtain an optimized interference identification model; and step S4, using the optimized interference identification model to identify interference in the interfered data link.

[0065] Specifically, in this embodiment, in step S1, three typical interference signals can be selected: noise amplitude modulation interference signal, noise frequency modulation interference signal, and noise phase modulation interference signal. Optionally, the noise mean μ in the interference signal is 0 and the variance σ is... 2 1. Carrier frequency f e 35MHz, sampling frequency f s 150MHz, effective amplitude modulation factor m Ae0.1, frequency modulation slope K FM 10MHz / V, phase modulation coefficient K PM The value is 10MHz / V; and each interference signal generates 400 sets of characteristic data, i.e., time-domain feature parameters, of which 70% of the data is used as training data when constructing the interference identification model and 30% of the data is used as test data when constructing the interference identification model, but the present invention is not limited thereto.

[0066] More specifically, in this embodiment, each of the time-domain feature parameters includes time-domain moment skewness, time-domain moment kurtosis, and envelope ripple; the time-domain moment skewness describes the degree of asymmetry in the signal's time-domain waveform, and its calculation formula is as follows:

[0067]

[0068] Where F1 represents the time-domain moment skewness; y(t) represents the interference signal; t represents time; M represents the period duration; u represents the mean of the interference signal; and σ represents the standard deviation of the interference signal.

[0069] The time-domain moment kurtosis describes the sharpness of the signal's time-domain waveform, and its calculation formula is as follows:

[0070]

[0071] Where F2 represents the time-domain moment kurtosis;

[0072] The envelope undulation is calculated using the following formula:

[0073]

[0074] Where F3 represents the envelope undulation; σ e The variance of the squared envelope of the interference signal; u e This represents the mean of the square of the interference signal envelope.

[0075] Please also refer to Figure 1 and Figure 2The adaptive fuzzy neural network includes an input layer, a fully connected layer, a normalization layer, a fuzzification layer, an inference layer, and a defuzzification layer; and step S2 includes: step S21, recording the temporal moment skewness, the temporal moment kurtosis, and the envelope variability as first input data, second input data, and third input data, respectively; step S22, mapping the first input data, the second input data, and the third input data through the fully connected layer to obtain a first vector, a second vector, and a third vector; step S23, standardizing the first vector, the second vector, and the third vector through the normalization layer. Step S24: Calculate the membership degrees of the first, second, and third standardized vectors to different fuzzy subsets through the fuzzification layer to obtain the first, second, and third membership values; Step S25: Calculate the excitation of each fuzzy rule through the inference layer based on the first, second, and third membership values; Step S26: Calculate the actual output of the model through the defuzzification layer based on the excitation of the fuzzy rule and the corresponding fuzzy rule.

[0076] Specifically, in this embodiment, regarding the simulation experiment scenario, in step S21, the input variables are [x1 = time-domain moment skewness, x2 = time-domain moment kurtosis, x3 = envelope ripple]. Since the input is the time-domain characteristic of the signal, the first input data x1 = [x 11 ,x 12 ,...,x 1n The second input data x2 = [x 21 ,x 22 ,...,x 2n The third input data x3 = [x 31 ,x 32 ,...,x 3n ]; and the first input data, the second input data, and the third input data are all multidimensional data. In step S22, the fully connected layer can map the first input data to the first neuron group, the second input data to the second neuron group, and the third input data to the third neuron group, and each of the first neuron group, the second neuron group, and the third neuron group includes a plurality of neurons; optionally, each of the first neuron group, the second neuron group, and the third neuron group includes 3 neurons, then the first vector obtained in step S22 is [f 11 ,f 12 ,f 13 ], the second vector is [f 21 ,f22 ,f 23 The third vector is [f] 31 ,f 32 ,f 33 More specifically, all three sets of vectors have actual semantic features; wherein the semantic features of the first vector are [symmetric, relatively symmetric, asymmetric], the semantic features of the second vector are [gentle, normal, steep], and the semantic features of the third vector are [large, medium, small], but the present invention is not limited thereto.

[0077] Specifically, in this embodiment, since the ranges of the first input data, the second input data, and the third input data are quite large, in order to better perform fuzzification processing on the input data later, the neuron values ​​(i.e., the first vector, the second vector, and the third vector) obtained in the dimensionality reduction in step S22 can be mapped to the range [0,1] through a nonlinear function in the normalization layer, thereby obtaining the first normalized vector, the second normalized vector, and the third normalized vector. More specifically, the first normalized vector is calculated using the following formula:

[0078]

[0079] Among them, s 1j f represents the first normalized vector; 1j Let a represent the first vector, and j∈[1, the total number of neurons in the first neuron group]. In this embodiment, j=1,2,3; a1 and b1 represent the first parameter and the second parameter, respectively, which can be set according to the specific feature parameters.

[0080] The second standardized vector is calculated using the following formula:

[0081]

[0082] Among them, s 2k f represents the second normalized vector; 2k Let a represent the second vector, and k∈[1, the total number of neurons in the second neuron group]. In this embodiment, k=1,2,3; a2 and b2 represent the third and fourth parameters, respectively, which can be set according to the specific feature parameters.

[0083] The third standardized vector is calculated using the following formula:

[0084]

[0085] Among them, s 3l f represents the third normalized vector; 3lLet l represent the third vector, and l∈[1, the total number of neurons in the third neuron group]. In this embodiment, l=1,2,3; a3 and b3 represent the fifth and sixth parameters, respectively, which can be set according to the specific feature parameters.

[0086] Specifically, in this embodiment, in step S24, the fuzzification layer plays a fuzzification role, that is, it calculates the fuzzy membership values ​​of each standardized vector belonging to different fuzzy subsets, and the first membership value is calculated using the following formula:

[0087] μ 1j =μ Aj (s 1j (7)

[0088] Where, μ 1j μ represents the first membership value. Aj s represents the first standardized variable 1j The membership function; Aj represents the fuzzy subset of the time-domain moment skewness;

[0089] The second membership value is calculated using the following formula:

[0090] μ 2k =μ Bk (s 2k (8)

[0091] Where, μ 2k Indicates the second membership value; μ 2k The second standardized variable s 2k The membership function; Bk represents a fuzzy subset of the time-domain moment kurtosis;

[0092] The third membership value is calculated using the following formula:

[0093] μ 3l =μ Cl (s 3l (9)

[0094] Where, μ 3l Indicates the third membership value; μ 3l Represents the third standardized variable s 3l The membership function; Cl represents a fuzzy subset of the envelope undulation.

[0095] More specifically, the membership functions of the first standardized variable, the second standardized variable, and the second standardized variable can all be Gaussian functions, but this invention is not limited thereto.

[0096] Specifically, in this embodiment, each node in the inference layer corresponds to a fuzzy rule, for example, x1isA j andx2isB jandx3isC j , thenisY; and the excitation w of the fuzzy rule corresponding to each node in the inference layer is the product of the membership degrees in the previous layer, i.e., the fuzzification layer. Optionally, the inference layer has a total of 27 nodes p1, p2, ..., p 27 More specifically, the excitation of the fuzzy rule corresponding to the i-th node in the inference layer, i.e., the i-th fuzzy rule, is calculated using the following formula:

[0097] ω i =μ 1j μ 2k μ 3l (10)

[0098] Where, ω i Let i represent the incentive of the i-th fuzzy rule, and i∈[1, total number of nodes in the inference layer]. In this embodiment, i=1,2,…,27.

[0099] Specifically, in this embodiment, in step S26, the defuzzification layer calculates the product of all inputs to each node in the inference layer to form the output of each node in that layer, and calculates the actual output of the model based on the output of each node. More specifically, the actual output of the model is calculated using the following formula:

[0100]

[0101] P=∑ω i p i (12)

[0102] C=∑p i (13)

[0103] Where y represents the actual output of the model; p i Let i represent the i-th fuzzy rule.

[0104] Please continue to refer to this. Figure 1 In this embodiment, in step S2, the initial interference identification model pre-sets a fuzzy membership function and fuzzy rules. In step S3, the learning algorithm corrects the parameters of the fuzzy membership function and the fuzzy rules based on a large amount of known data; the learning algorithm employs a hybrid algorithm combining least squares and gradient descent. More specifically, the error function of the initial interference identification model uses a mean squared error function, the specific expression of which is as follows:

[0105]

[0106] Where Y is the desired output, and the value of Y is 0, 1 and 2 according to the three typical interference signals selected (0 is noise amplitude modulation interference, 1 is noise frequency modulation interference, and 2 is noise phase modulation interference).

[0107] Specifically, in this embodiment, in the learning algorithm, input data and functional signals are propagated forward along each network in the initial interference identification model, and the output of each node in each network is calculated. At this time, the antecedent parameters are fixed, and the consequent parameters are adjusted using a least squares estimation algorithm until the actual output of the initial interference identification model is obtained. An error signal is calculated based on the actual output and expected output of the initial interference identification model. The error signal propagates backward along the network, and the antecedent parameters are updated using gradient descent. The antecedent and consequent parameters are iteratively adjusted using the learning algorithm until the iteration stopping condition is reached, at which point the optimized interference identification model is obtained. Optionally, the antecedent parameters are semantic features, such as symmetry, gradualness, or largeness; the consequent parameters are output values, such as amplitude-modulated noise interference; and the iteration stopping condition is reaching a set accuracy, but this invention is not limited thereto.

[0108] Furthermore, in this embodiment, the optimized interference identification model is used to identify the three typical interference signals to be identified. To verify the interference identification performance of the data link interference identification method provided in this embodiment, a simulation experiment was conducted under a general noise background, i.e., SNR = -3dB, to test the correct identification rate of the optimized interference identification model under different interference-to-signal ratios. The simulation results are as follows: Figure 3 As shown. From Figure 3 As can be seen: (1) The interference identification rate of the data link interference identification method provided in this embodiment is generally very high, above 96%; (2) When the interference-to-signal ratio is less than the system interference tolerance (23dB), the identification rate is between 96% and 99.6%, and when the interference-to-signal ratio is greater than the interference tolerance, the identification rate is close to 100%; (3) In a fairly large dynamic range, the identification rate does not change with the interference-to-signal ratio, and has good robustness and stability.

[0109] In summary, this embodiment provides a data link interference identification method. It constructs an initial interference identification model based on the temporal feature parameters of the interference signal and utilizes an adaptive fuzzy neural network. By training the initial interference identification model with a learning algorithm, an optimized interference identification model can be obtained, thereby accurately identifying interference in the interfered data link. This embodiment maps the temporal feature parameters with multidimensional data to vectors through a fully connected layer. A fuzzification layer and an inference layer are designed in the intermediate layers to generate fuzzy rules and their excitation strengths. A hybrid algorithm is then used to automatically update the membership function and fuzzy rules. The output layer defuzzifies the data and outputs the interference type and its confidence level. Applying this method to the field of interference recognition, it has advantages such as easy feature extraction, adaptability, and self-learning capabilities.

[0110] Although the present invention has been described in detail through the preferred embodiments above, it should be understood that the above description should not be considered as a limitation of the present invention. Various modifications and substitutions to the present invention will be apparent to those skilled in the art after reading the above description. Therefore, the scope of protection of the present invention should be defined by the appended claims.

Claims

1. A data link interference identification method, characterized in that, include: Acquire the interference signal to obtain the time-domain characteristic parameters of the interference signal; Based on the aforementioned time-domain feature parameters, an initial interference recognition model is constructed using an adaptive fuzzy neural network; The initial interference identification model is trained using a learning algorithm to obtain an optimized interference identification model; and The optimized interference identification model is used to identify interference in the interfered data link; The time-domain feature parameters include time-domain moment skewness, time-domain moment kurtosis, and envelope undulation. The adaptive fuzzy neural network includes an input layer, a fully connected layer, a normalization layer, a fuzzification layer, an inference layer, and a defuzzification layer; and the steps for constructing the initial interference recognition model include: The time-domain moment skewness, the time-domain moment kurtosis, and the envelope undulation are respectively denoted as the first input data, the second input data, and the third input data; and the first input data, the second input data, and the third input data are all multidimensional data; The first input data, the second input data, and the third input data are mapped to the first neuron group, the second neuron group, and the third neuron group, respectively, through the fully connected layer to obtain the first vector, the second vector, and the third vector; each of the first neuron group, the second neuron group, and the third neuron group includes a plurality of neurons; The first vector, the second vector, and the third vector are mapped to the interval [0, 1] through the normalization layer to obtain the first normalized vector, the second normalized vector, and the third normalized vector; The membership degrees of the first standardized vector, the second standardized vector, and the third standardized vector to different fuzzy subsets are calculated through the fuzzification layer to obtain the first membership value, the second membership value, and the third membership value. Based on the first membership value, the second membership value, and the third membership value, calculate the excitation of the fuzzy rule corresponding to each node in the inference layer; Based on the excitation of the fuzzy rule and the corresponding fuzzy rule, the actual output is calculated through the defuzzification layer calculation model.

2. The data link interference identification method as described in claim 1, characterized in that, The time-domain moment skewness is calculated using the following formula: in, F 1 indicates the time-domain moment skewness; y ( t () indicates an interference signal; t Indicates time, M Indicates the duration of the period; u Indicates the mean value of the interference signal; σ This represents the standard deviation of the interference signal; The time-domain moment kurtosis is calculated using the following formula: in, F 2 represents the time-domain moment kurtosis; The envelope undulation is calculated using the following formula: in, F 3 indicates envelope variability; σ e The variance of the squared envelope of the interference signal; u e This represents the mean of the square of the interference signal envelope.

3. The data link interference identification method as described in claim 1, characterized in that, The first standardized vector is calculated using the following formula: in, s 1j Represents the first normalized vector; f 1j Let represent the first vector, and j ∈[1, the total number of neurons in the first neuron group]; a 1 and b 1 represents the first parameter and the second parameter respectively; The second standardized vector is calculated using the following formula: in, s 2k Represents the second normalized vector; f 2k Denotes the second vector, and k ∈[1, total number of neurons in the second neuron group]; a 2 and b 2 represents the third and fourth parameters respectively; The third standardized vector is calculated using the following formula: in, s 3l Represents the third normalized vector; f 3l Denotes the third vector, and l ∈[1, total number of neurons in the third neuron group]; 𝑎3 and 𝑏3 represent the fifth and sixth parameters, respectively.

4. The data link interference identification method as described in claim 3, characterized in that, The first membership value is calculated using the following formula: in, μ 1j Indicates the first membership value; μ Aj Represents the first standardized variable s 1j Membership function; Aj A fuzzy subset representing the time-domain moment skewness; The second membership value is calculated using the following formula: in, μ 2k Indicates the second membership value; μ 2k Represents the second standardized variable s 2k Membership function; Bk A fuzzy subset representing the kurtosis of time-domain moments; The third membership value is calculated using the following formula: in, μ 3l Indicates the third membership value; μ 3l Represents the third standardized variable s 3l Membership function; Cl A fuzzy subset representing the envelope undulation.

5. The data link interference identification method as described in claim 4, characterized in that, The actual output of the model is calculated using the following formula: in, y This represents the actual output of the model; ω i Indicating the first inference layer i The excitation of the fuzzy rules corresponding to each node, and i ∈[1, total number of nodes in the inference layer]; p i Indicating the first inference layer i The fuzzy rules corresponding to each node.

6. The data link interference identification method as described in claim 4, characterized in that, The membership functions of the first standardized variable, the second standardized variable, and the second standardized variable are all Gaussian functions.

7. The data link interference identification method as described in claim 1, characterized in that, The learning algorithm employs a hybrid algorithm combining least squares and gradient descent.

Images