Electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology

Abstract

The invention belongs to the technical field of communication interference cancellation, and discloses an electromagnetic interference real-time monitoring and cancellation method based on a multi-antenna technology, which can realize comprehensive perception of an electromagnetic environment from a spatial dimension by utilizing the multi-antenna technology and through cooperative work of a main antenna and an auxiliary antenna. And an accurate basis is provided for timely coping with electromagnetic interference. Besides, an interference identification model is further constructed and is suitable for monitoring various different types of electromagnetic interferences, interference cancellation parameters and strategies are dynamically adjusted according to the interference types, high adaptability and flexibility are achieved, a cancellation algorithm is matched for the interference types, signals with the same amplitude and opposite phases as interference signals are generated, and the interference cancellation accuracy is improved. Therefore, the interference is quickly and effectively counteracted, and the normal operation of a satellite navigation communication system is ensured. In addition, according to the method disclosed by the invention, the anti-interference capability of the satellite navigation communication system in a complex electromagnetic environment can be remarkably enhanced, and the anti-interference tolerance is greatly improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of communication interference cancellation, and particularly relates to a method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology. BACKGROUND

[0002] With the rapid advancement of new power system construction and digital power grid transformation, significant progress has been made in the application and research of satellite navigation communication technology in recent years. Through the widespread deployment of global navigation satellite systems such as Beidou and GPS, the power system has achieved high-precision time synchronization in digital power transmission, transformation, and distribution systems. In terms of intelligentization of power production and operation, through the large-scale application of satellite navigation technology, a series of innovative applications such as unmanned aerial vehicle joint autonomous inspection, robot intelligent inspection, worker positioning, intelligent safety monitoring, digital twin, substation and transmission tower geological disaster monitoring, and power emergency communication have been implemented, which not only consolidates the foundation of power grid safety, but also improves the efficiency of power grid production and operation.

[0003] However, electromagnetic interference has been one of the important factors affecting the accuracy and reliability of satellite navigation communication. With the development and large-scale application of Beidou satellite navigation communication technology developed independently by China, Harbin Institute of Technology, Beijing University of Science and Technology, National University of Defense Technology, and multiple research institutes have conducted research on Beidou satellite navigation communication anti-jamming technology. Harbin Institute of Technology proposed a joint space-time processing algorithm in 2006, which uses the direction of arrival of the interference source and the navigation satellite information as the constraint condition to maximize the preservation of satellite signals approaching the direction of arrival of the interference source and suppress the interference signals. Beijing University of Science and Technology implemented a digital anti-jamming system on an FPGA platform in 2012 by using resource reuse strategies, solving the problem of hardware resource shortage. National University of Defense Technology established a calculation method for anti-jamming evaluation indicators in 2013 and completed the design and implementation of the evaluation system. The 54th Institute of China Electronics Technology Group explored multi-dimensional joint anti-jamming technology under polarization space-time and space-frequency in 2022, improving the degree of freedom of anti-jamming.

[0004] Although there are currently various commonly used anti-jamming technologies such as spread spectrum, frequency hopping, and hybrid spread spectrum technology, these anti-jamming technologies are mostly designed for specific interference scenarios and interference sources, have high requirements for receiver carrier capture and timing synchronization, and usually require more spectrum resources or increase system complexity, which requires high design requirements for system software and hardware, and has relatively insufficient adaptability and flexibility. In addition, machine learning methods show potential in the judgment of electromagnetic interference, but current research on satellite navigation communication anti-jamming based on machine learning methods to judge interference types in complex electromagnetic environments is still insufficient.

[0005] Therefore, studying a method for identifying multiple types of electromagnetic interference types by using a machine learning technique and effectively suppressing the electromagnetic interference types has become a technical problem to be solved in the current satellite navigation communication field. SUMMARY

[0006] The purpose of the present application is to propose a multi-antenna technology-based electromagnetic interference real-time monitoring and cancellation method, which uses multi-antenna technology to perceive the electromagnetic environment in all directions from the spatial dimension, judges whether there is electromagnetic interference and the interference type through an interference identification model, matches a cancellation algorithm according to the interference type, and finally realizes effective suppression of electromagnetic interference.

[0007] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0008] A multi-antenna technology-based electromagnetic interference real-time monitoring and cancellation method comprises the following steps:

[0009] Step 1. Collect satellite signal data in electromagnetic interference scenes and electromagnetic interference-free scenes, and construct a training data set;

[0010] Step 2. Build an interference identification model, which comprises the following structure:

[0011] A convolutional neural network is used to extract the frequency domain features and time-frequency features of the satellite signal data;

[0012] A long short-term memory network is used to extract the time domain features of the satellite signal data;

[0013] A support vector machine is used to determine the interference type according to the frequency domain features, time-frequency features and time domain features of the satellite signal data;

[0014] Step 3. Based on the training data set, the interference identification model is trained in a supervised learning manner;

[0015] Step 4. Receive a main signal through a main antenna, the main signal being a baseband signal obtained by converting a target signal and an interference signal through radio frequency processing; receive an auxiliary signal through an auxiliary antenna, the auxiliary signal being a baseband signal obtained by converting the interference signal through radio frequency processing;

[0016] Input the auxiliary signal into the trained interference identification model to obtain a judgment result of the interference type;

[0017] Step 5. Match a cancellation algorithm according to the judgment result of the interference type, generate a cancellation signal with the same amplitude and opposite phase as the interference signal, and obtain a satellite signal after interference cancellation by superimposing the main signal and the cancellation signal.

[0018] Further, based on the above-mentioned electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology, the application further provides a computer device, which comprises a memory and one or more processors.

[0019] The memory stores executable code, and the processor executes the executable code to implement the steps of the above-mentioned electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology.

[0020] Further, based on the above-mentioned electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology, the application further provides a computer readable storage medium, which stores a program; the program is executed by a processor to implement the steps of the above-mentioned electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology.

[0021] The application has the following advantages:

[0022] As described above, the application provides an electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology. The method utilizes multi-antenna technology, cooperates the main antenna and the auxiliary antenna, receives the main signal including the target signal and the interference signal by the main antenna, and receives the interference signal by the auxiliary antenna, can realize comprehensive perception of the electromagnetic environment from the spatial dimension, provides accurate basis for timely response to electromagnetic interference. In addition, the method of the application also builds an interference identification model, which is suitable for monitoring of various types of electromagnetic interference, the method of the application can dynamically adjust the interference cancellation parameters and strategies according to the type of interference, has high adaptability and flexibility, once the interference signal is detected, the cancellation algorithm is matched according to the type of interference, a signal with the same amplitude and opposite phase as the interference signal is generated, so as to realize rapid and effective cancellation of the interference, thereby ensuring the normal operation of the satellite navigation communication system. In addition, the method of the application can significantly enhance the anti-interference ability of the satellite navigation communication system in complex electromagnetic environment, and greatly improve the anti-interference tolerance. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 The flowchart of the electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology in the embodiment of the application.

[0024] Figure 2 The schematic diagram of the interference identification model in the embodiment of the application.

[0025] Figure 3 The signal processing flowchart of the electromagnetic interference real-time monitoring and cancellation method based on multi-antenna technology in the application. DETAILED DESCRIPTION

[0026] The application will be further described in detail below in combination with the drawings and specific embodiments:

[0027] Embodiment 1

[0028] The embodiment proposes a multi-antenna technology-based electromagnetic interference real-time monitoring and cancellation method. The method is based on multi-antenna reception and combination of main signals, and through the cooperative work of the main antenna and the auxiliary antenna, it can realize rapid and accurate situation awareness of unknown interference signals, thereby realizing real-time monitoring of electromagnetic interference. The method can also match appropriate cancellation algorithms according to different interference types, and has high adaptability and flexibility.

[0029] Asymmetrical electromagnetic interference, i.e. strong suppression electromagnetic interference, is usually generated by specific wireless transmitting devices, radars, communication systems or malicious interference sources as a powerful electromagnetic interference signal. The signals of the wireless communication system are usually distributed in a specific frequency band, for example, the GPS L1 carrier is located at 1575.42 MHz, with a bandwidth of about 2.046 MHz. Asymmetrical interference is usually one or more high-power interference signals, and if its frequency falls within the bandwidth of the target signal, it will directly cover part or all of the target signal, making it difficult for the receiver to normally demodulate. The method of the present application is particularly suitable for asymmetrical electromagnetic interference, and using the method of the present application can also greatly improve the anti-interference margin of the satellite navigation system.

[0030] As shown in Figure 1 A multi-antenna technology-based electromagnetic interference real-time monitoring and cancellation method, specifically comprising the following steps:

[0031] Step 1. Collect satellite signal data in electromagnetic interference scenarios and electromagnetic interference-free scenarios, and construct a training data set.

[0032] The satellite signal data collected in the electromagnetic interference scenario is a mixed signal containing the target signal and the interference signal, and in fact the interference signal strength is much greater than the target signal strength at this time, and the target signal will be submerged in the interference. The main purpose of collecting satellite signal data in the electromagnetic interference scenario is to collect interference signals, convert them into baseband signals, and use them for subsequent interference identification model to judge the interference type and match the cancellation algorithm.

[0033] In this embodiment, the training data set constructed in step 1 includes satellite signal data and interference type labels, wherein the satellite signal can be a GPS satellite signal or a Beidou satellite signal.

[0034] The satellite signal data includes time domain data, frequency domain data and time-frequency domain data.

[0035] The time domain data includes time domain amplitude and phase.

[0036] The frequency domain data includes power spectral density (PSD), which can be calculated by fast Fourier transform (FFT).

[0037] The time-frequency domain data includes a time-frequency spectrogram obtained by a short-time Fourier transform (STFT).

[0038] The interference type label includes no electromagnetic interference, narrowband interference, wideband interference, and impulse interference.

[0039] Step 2. Building an interference identification model, which includes a convolutional neural network (CNN), a long short-term memory (LSTM) network, and a support vector machine (SVM).

[0040] The convolutional neural network (CNN) is used to extract frequency domain features and time-frequency features of satellite signal data, that is, the frequency domain data and time-frequency data in the satellite signal data are processed through the convolutional neural network, and the frequency domain features and time-frequency features are learned.

[0041] The processing flow of the signal in the convolutional neural network is as follows:

[0042] The data input into the convolutional neural network is a power spectral density (PSD) matrix and a time-frequency spectrogram of the signal. First, the input data is preprocessed, which is normalization processing, that is, the input data is normalized first. Then, a convolution operation is performed, the convolution layer slides along the frequency axis through the first layer of convolution kernels to extract local frequency domain patterns, and a ReLU activation function is used to introduce nonlinearity. Then, a pooling operation is performed, the pooling layer retains significant features through maximum pooling and reduces the frequency dimension. Then, a full connection operation is performed, the bottom layer features and information obtained by the convolution layer and the pooling layer are summarized and flattened through the full connection layer, and the output is processed through a Softmax function. Finally, the output result of the convolutional neural network is a frequency domain feature vector and a time-frequency feature vector.

[0043] The long short-term memory (LSTM) network is used to extract time domain features of satellite signal data. The long short-term memory network is used to model time series and extract time series features of satellite signal data, i.e., time domain features.

[0044] The processing flow of the signal in the long short-term memory network is as follows:

[0045] The data input into the long short-term memory network is a two-dimensional sequence composed of time domain amplitude and phase of the satellite signal data, which is operated through a gating mechanism and updated through memory, and the output result is a time domain feature vector. The processing process specifically includes:

[0046] The gating mechanism operates, the forgetting gate decides how much historical state to retain through a Sigmoid function, the input gate filters current time effective information such as amplitude mutation / phase jump detection, and the output gate controls the hidden state output. Memory update, cell state accumulates long-term features such as signal envelope trend, and hidden state captures short-term dynamics such as amplitude transient change.

[0047] Support vector machine (SVM) is used to determine the interference type based on the frequency domain, time-frequency, and time domain characteristics of satellite signal data. The signal processing flow in the support vector machine is as follows:

[0048] The input data of the support vector machine are the frequency domain feature vector and time-frequency feature vector output by the convolutional neural network, and the time domain feature vector output by the long short-term memory network.

[0049] First, feature fusion is performed to concatenate the time domain feature vector, frequency domain feature vector and time-frequency domain feature vector into a single vector to obtain a fused feature vector, and the fused feature is standardized.

[0050] Then select the kernel function and use the radial basis kernel function RBF kernel for nonlinear separable features.

[0051] Then, interference classification is performed, and the categories include no electromagnetic interference, narrowband interference, broadband interference, and pulse interference. A one-to-many strategy is used for multi-classification, so multiple SVMs need to be trained. In this embodiment, the input fusion feature vector is calculated by four SVMs for confidence, such as Figure 2 As shown, SVM1, SVM2, SVM3, and SVM4 represent the four support vector machines, and the category with the highest confidence is selected as the judgment result of the interference type output by the support vector machine.

[0052] The confidence calculation formula is as follows:

[0053]

[0054] Among them, k represents the kth classifier, f k (x) represents the confidence that the sample belongs to the category corresponding to the k-th classifier, x i Represents the i-th sample, i∈[1,N], N is the total number of samples, α i,k is the support vector, y i,k is the label, b k is the bias, κ(x i ,x) is the kernel function.

[0055] The method of the present invention can determine whether electromagnetic interference exists and classify the interference type. The interference identification model inputs the time domain data, frequency domain data, and time-frequency domain data of the satellite signal. CNN processes the frequency domain data and the time-frequency domain data, while LSTM processes the time domain data to obtain time domain feature vectors, frequency domain feature vectors, and time-frequency domain feature vectors. After feature vector fusion, SVM is used to further classify the specific interference type, which includes no electromagnetic interference, narrowband interference, broadband interference, and pulse interference.

[0056] Step 3. Based on the training dataset, the interference recognition model is trained using supervised learning.

[0057] The training data set is the data set of known interference types constructed in step 1, and the main purpose of training the interference identification model with satellite signal data received by multiple antennas is to enable the interference identification model to learn and distinguish interference signals and improve the accuracy of interference detection and classification, so as to quickly identify and suppress interference in actual application. In the training process, a supervised learning method is used, and the classification accuracy and misjudgment rate are calculated to adjust the parameters and optimize the model.

[0058] In order to perform end-to-end joint training of the interference identification model, a multi-task loss function needs to be designed, and the total loss function is:

[0059] L total =αL CNN +βL LSTM +γL SVM .

[0060] Wherein, L total is the total loss, L CNN is the cross-entropy loss, L LSTM is the mean square error loss, L SVM is the hinge loss, and α, β, γ are task weights.

[0061] Specifically, the cross-entropy loss function is represented as:

[0062]

[0063] Wherein, c represents the class, c∈[1,C], C represents the total number of classes, y i,c represents the true label of the i-th sample in the c-th class, and p i,c represents the class probability output by the Softmax function for the i-th sample in the c-th class.

[0064] The mean square error loss function is represented as:

[0065]

[0066] Wherein, t represents the time sequence, t∈[1,T], T represents the total time, represents the reconstructed signal output by the long short-term memory network, and x t represents the original signal input into the long short-term memory network.

[0067] The hinge loss function is represented as:

[0068]

[0069] Wherein, represents the classification output by the interference identification model, w is the SVM weight vector, b is the bias, κ(x i ) is the kernel function, and λ is the regularization coefficient of SVM.

[0070] The interference identification model is trained in step 3 by using a joint training strategy, and the joint training strategy specifically includes:

[0071] Forward propagation is performed, and the CNN and LSTM branches are calculated in parallel, and the features are fused and input to the SVM.

[0072] Back propagation is performed, and the gradient is propagated back to the fusion layer through the SVM, and the CNN and LSTM parameters are updated respectively.

[0073] Part of the parameters are frozen, the SVM is frozen in the first 10 rounds, only the feature extractor composed of CNN and LSTM is trained, and all layers are fine-tuned from the 11th round.

[0074] Step 3 also includes performance evaluation and parameter adjustment, which specifically includes:

[0075] Evaluation index calculation, including classification accuracy and misjudgment rate analysis, and the formula of classification accuracy is:

[0076]

[0077] Where N is the total number of samples, represents the model output classification, y i represents the label classification, represents the model output classification is the probability of label classification y i , and Acc represents the probability of the number of samples with correct model output classification in all samples, i.e. the classification accuracy. The misjudgment rate is Mis = 1-Acc.

[0078] Optimize the key parameters of the model, such as optimizing the CNN convolution kernel size by grid search, optimizing the LSTM hidden unit number by early stopping mechanism, and optimizing the SVM kernel parameter by Bayesian optimization.

[0079] Step 4. Receive the main signal through the main antenna, the main signal is the baseband signal obtained by converting the target signal and the interference signal through radio frequency processing; receive the auxiliary signal through the auxiliary antenna, the auxiliary signal is the baseband signal obtained by converting the interference signal through radio frequency processing; input the auxiliary signal into the trained interference identification model to obtain the judgment result of the interference type.

[0080] The application receives signals by the main antenna and the auxiliary antenna in cooperation, the main antenna selects a high-gain antenna, and the auxiliary antenna selects a low-gain antenna. In fact, the received signals of the main antenna and the auxiliary antenna are mixed signals containing target signals and interference signals, but the interference signals are much stronger than the target signals, the target signals are submerged in the interference, and the interference signals play a major role. Therefore, the main antenna is used for receiving the main signal, and the auxiliary antenna is used for receiving the auxiliary signal to collect spatial interference information and for subsequent interference identification and cancellation.

[0081] The satellite signals received by the main antenna and the auxiliary antenna are converted into baseband digital signals, i.e., baseband signals, through radio frequency processing, wherein the main antenna corresponds to the main signal, and the auxiliary antenna corresponds to the auxiliary signal.

[0082] In step 4, the radio frequency processing mainly includes low-noise amplification, frequency conversion, and analog-to-digital conversion.

[0083] The auxiliary signal is preprocessed to obtain the time-domain signal amplitude and phase, power spectral density, and time-frequency spectrum.

[0084] The preprocessed data is input into the interference identification model trained in steps 1 to 3.

[0085] The time-domain feature vector, the frequency-domain feature vector, and the time-frequency domain feature vector are obtained through the CNN and the LSTM processing, the feature vectors are fused, and then processed by the SVM to calculate the confidence, and the interference type judged by the interference identification model is output.

[0086] The judgment basis is that if the f k (x) of a certain classifier is greater than 0 and is the maximum value, it is determined as the corresponding category. For example, the confidence f k (x), k = 1, 2, 3, 4, corresponds to no electromagnetic interference, narrowband interference, wideband interference, and pulse interference, and the calculation result is [f1(x) = 0.5, f2(x) = 5.6, f3(x) = -2.2, f4(x) = 1.8], and the judgment result of the interference type is determined as narrowband interference.

[0087] Step 5. According to the judgment result of the interference type, the cancellation algorithm is matched to generate a cancellation signal with the same amplitude and opposite phase as the interference signal, and the satellite signal after interference cancellation is obtained by superimposing the main signal and the cancellation signal, so as to realize the suppression of electromagnetic interference.

[0088] The interference cancellation algorithm includes beamforming algorithm, power inversion algorithm, sidelobe cancellation algorithm and space-time adaptive filtering (STAP) algorithm. If the judgment result of the interference type is no electromagnetic interference, the beamforming algorithm is adopted; if the judgment result of the interference type is pulse interference, the power inversion algorithm is adopted; if the judgment result of the interference type is narrowband interference, the sidelobe cancellation algorithm is adopted; and if the judgment result of the interference type is wideband interference, the STAP algorithm is adopted.

[0089] Specifically, for no electromagnetic interference, the beamforming algorithm is adopted. The beamforming algorithm weights all antennas based on the target signal direction, so as to realize the gain improvement of the target signal direction. The implementation process of the algorithm is as follows: first, the main signal direction is extracted. Then, the main lobe direction steering vector is constructed. Then, the weighting coefficient is designed to suppress the signals in other directions. Finally, the satellite signal after beam enhancement is output.

[0090] For pulse interference, the power inversion algorithm is adopted. The power inversion algorithm is an interference cancellation algorithm with the minimum output power as the target. The implementation process of the algorithm is as follows: first, the instantaneous high-energy point of the pulse interference is detected in the time domain. Then, the power spectrum matrix is constructed and inverted. Then, the weight corresponding to the minimum output power direction is taken. Finally, the original signal is weighted and summed to suppress the pulse interference.

[0091] For narrowband interference, the sidelobe cancellation algorithm is adopted. The sidelobe cancellation algorithm weights the signal based on the interference information to realize the nulling of the interference direction, so as to suppress the interference. The implementation process of the algorithm is as follows: first, the spatial steering vector corresponding to the interference direction is extracted. Then, the interference direction nulling weight is constructed. Then, the array signal is weighted to form a sidelobe suppression beam. Finally, the narrowband interference is suppressed, and the main signal quality is improved.

[0092] For wideband interference, the STAP algorithm is adopted. The STAP algorithm extends the spatial filtering to space-time two-dimensional filtering by increasing the time delay, increases the degree of freedom of interference suppression, and is suitable for the elimination of wideband interference. The implementation process of the algorithm is as follows: first, the space-time joint data matrix is constructed. Then, the covariance matrix is estimated and the weight vector is calculated. Then, the space-time filter is designed to suppress the wideband interference component. Finally, the filtering result of the signal reserved in the target direction is output.

[0093] Through the matched interference cancellation algorithm, the interference cancellation signal with the same amplitude and opposite phase as the interference signal is generated, and the satellite signal after interference cancellation is obtained by superimposing the main antenna signal and the interference cancellation signal, so as to improve the satellite signal quality. Finally, the electromagnetic interference is quickly and accurately cancelled, and the signal demodulation and positioning accuracy are improved.

[0094] The electromagnetic interference real-time monitoring and cancellation method based on the multi-antenna technology can realize real-time and accurate monitoring of electromagnetic interference, utilizes the multi-antenna technology to receive signals from multiple spatial angles, can accurately perceive various parameters of the interference signal in real time, such as frequency and intensity, and provides accurate basis for timely response to interference. In addition, the method can also realize fast and intelligent cancellation. Once the interference signal is detected, the cancellation algorithm can be matched and responded quickly to generate a signal with the same amplitude and opposite phase of the interference signal, so that the interference is quickly and effectively cancelled, thereby ensuring the normal operation of the satellite navigation communication system. In addition, the method can greatly improve the anti-interference margin. By effectively suppressing the interference signal, the anti-interference capability of the satellite navigation communication system in a complex electromagnetic environment is significantly enhanced, and the system can better cope with extreme situations such as hostile and malicious interference, thereby ensuring the stable operation of the three safety production defense line equipment of the power grid.

[0095] The method uses multi-antenna spatial perception, which innovatively uses multi-antenna technology to realize all-round perception of the electromagnetic environment from the spatial dimension by arranging multiple antennas. The high-gain antenna is used as the main antenna, which is mainly used for receiving interference signals and target signals, i.e., satellite signals. The low-gain antenna is used as the auxiliary antenna, which is mainly used for sensing the electromagnetic environment and receiving interference signals to provide a reference for subsequent interference feature extraction and cancellation. The cooperative work of the main antenna and the auxiliary antenna can simultaneously receive signals from different directions, effectively utilize the differences in signal propagation in space, and provide a rich information basis for accurately obtaining the characteristics of the interference signal, thereby ensuring accurate identification and cancellation of the interference signal.

[0096] The method also establishes a real-time and dynamic interference monitoring mechanism, which can continuously track the changes of the interference signal. Through real-time monitoring of the interference signal, the dynamic changes of the key parameters such as intensity, frequency and phase are grasped in time, real-time and accurate data support is provided for intelligent matching of the cancellation algorithm, and the effectiveness of the cancellation strategy is ensured.

[0097] The cancellation algorithm of the method has high adaptability and flexibility. It can dynamically adjust the interference cancellation parameters and strategies according to the real-time monitoring of the interference, and ensure the high quality of the received satellite signals. When facing different types, intensities and change rules of interference, the algorithm can automatically optimize the generation of cancellation signals to realize accurate cancellation of the interference and minimize the impact of the interference on the satellite navigation communication system. By matching the cancellation algorithm and generating cancellation signals, the interference is suppressed. It can quickly analyze various parameters of the signal, accurately identify the interference signal and useful signal, realize accurate positioning, classification and evaluation of the interference signal, and overcome the limitations of traditional methods in processing complex electromagnetic environment signals.

[0098] Example 2

[0099] The embodiment 2 discloses a computer device, which comprises a memory and one or more processors.

[0100] The executable code is stored in the memory, and when the processor executes the executable code, steps of the electromagnetic interference real-time monitoring and cancellation method based on the multi-antenna technology in the above embodiment 1 are implemented.

[0101] The computer device in the embodiment is any device or apparatus with data processing capability, which will not be described here.

[0102] Embodiment 3

[0103] The embodiment 3 discloses a computer readable storage medium, which stores a program, and when the program is executed by a processor, steps of the electromagnetic interference real-time monitoring and cancellation method based on the multi-antenna technology are implemented.

[0104] The computer readable storage medium can be an internal storage unit of any device or apparatus with data processing capability, such as a hard disk or a memory, or can be an external storage device of any device with data processing capability, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc.

[0105] Of course, the above description is only for the preferred embodiments of the present application, and the present application is not limited to the above-mentioned embodiments. It should be noted that any skilled person in the art can make all equivalent replacements and obvious modifications under the teaching of the present application, and all the replacements and modifications fall within the scope of the present application, and should be protected by the present application.

Claims

1. A method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology, characterized in that: The steps include: Step 1. Collect satellite signal data in electromagnetic interference scenarios and non-electromagnetic interference scenarios to construct a training dataset; Step 2. Build an interference recognition model, which includes the following structure: Convolutional neural network, used to extract frequency domain features and time-frequency features of satellite signal data; Long short-term memory network, used to extract time domain features of satellite signal data; Support vector machine, used to determine the interference type based on the frequency domain characteristics, time-frequency characteristics, and time domain characteristics of satellite signal data; Step 3. Based on the training data set, the interference recognition model is trained using supervised learning. Step 4. Receive a primary signal through the primary antenna. The primary signal is a baseband signal obtained by converting the target signal and the interference signal through RF processing. Receive an auxiliary signal through the auxiliary antenna. The auxiliary signal is a baseband signal obtained by converting the interference signal through RF processing. Input the auxiliary signal into the trained interference recognition model to obtain the judgment result of the interference type; Step 5. Match the cancellation algorithm based on the interference type judgment result to generate a cancellation signal with the same amplitude and opposite phase as the interference signal. By superimposing the main signal and the cancellation signal, the interference-free satellite signal is obtained.

2. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 1, characterized in that: In step 1, the constructed training data set includes satellite signal data and interference type labels; Satellite signal data includes time domain data, frequency domain data, and time-frequency domain data; Time domain data includes time domain amplitude and phase; Frequency domain data includes power spectral density PSD; The time-frequency domain data include the time-frequency spectrum obtained by short-time Fourier transform (STFT); Interference type labels include no electromagnetic interference, narrowband interference, broadband interference, and pulse interference.

3. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 2, characterized in that: In step 2, the signal processing flow in the convolutional neural network is as follows: The data input into the convolutional neural network is the power spectrum density matrix and time-frequency spectrum of the satellite signal data; First, the input data is preprocessed, and the preprocessing is normalization; Then the convolution operation is performed. The convolution layer slides along the frequency axis through the first layer of convolution kernel to extract the local frequency domain pattern and introduces nonlinearity using the ReLU activation function. Next, the pooling operation is performed. The pooling layer retains significant features through maximum pooling and reduces the frequency dimension. After that, a full connection operation is performed, and the underlying features and information obtained by the flattened convolution layer and pooling layer are summarized through the fully connected layer, and processed by the Softmax function before output; The final output of the convolutional neural network is the frequency domain feature vector and the time-frequency feature vector.

4. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 3, characterized in that: In step 2, the data input into the long short-term memory network is a two-dimensional sequence consisting of time domain amplitude and phase in the satellite signal data, which is operated by a gating mechanism and memory updated, and the output result is a time domain feature vector.

5. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 4, characterized in that: In step 2, the signal processing flow in the support vector machine (SVM) is as follows: The data input to the support vector machine are the frequency domain feature vector and time-frequency feature vector output by the convolutional neural network, and the time domain feature vector output by the long short-term memory network; First, feature fusion is performed to concatenate the time domain feature vector, frequency domain feature vector, and time-frequency domain feature vector into a single vector to obtain a fused feature vector, and the fused feature vector is normalized. Then select the kernel function and use the radial basis kernel function RBF kernel for nonlinear separable features; Then, interference classification is performed, including no electromagnetic interference, narrowband interference, broadband interference, and pulse interference. The fused feature vector is input into four SVMs for confidence calculation. The confidence calculation formula is: Among them, k represents the kth classifier, f k (x) represents the confidence that the sample belongs to the category corresponding to the k-th classifier, x i Represents the i-th sample, i∈[1,N], N is the total number of samples, α i,k is the support vector, y i,k is the label, b k is the bias, κ(x i ,x) is the kernel function; Finally, the category with the highest confidence is selected as the judgment result of the interference type output by the support vector machine.

6. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 5, characterized in that: In step 3, the total loss function is designed as: L total =αL CNN +βL LSTM +γL SVM ; Among them, L total is the total loss, L CNN is the cross entropy loss, L LSTM is the mean square error loss, L SVM is the hinge loss, α, β, γ are weights; The cross entropy loss function is expressed as: Among them, c represents the category, c∈[1,C], C represents the total number of categories, y i,c represents the true label of the i-th sample in the c-th class, p i,c represents the category probability of the i-th sample in the c-th category; The mean square error loss function is expressed as: Among them, t represents the time sequence, t∈[1,T], T represents the total time, Represents the reconstructed signal output by the long short-term memory network, x t Represents the original signal input to the long short-term memory network; The hinge loss function is expressed as: in, represents the output classification of the interference recognition model, w is the SVM weight vector, b is the bias, κ(x i ) is the kernel function, and λ is the regularization coefficient of SVM.

7. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 1, characterized in that: In step 4, the main antenna uses a high-gain antenna, and the auxiliary antenna uses a low-gain antenna.

8. The method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology according to claim 5, characterized in that: In step 5, the cancellation algorithm includes a beamforming algorithm, a power inversion algorithm, a sidelobe cancellation algorithm, and a space-time adaptive filtering STAP algorithm; If the interference type is determined to be no electromagnetic interference, the beamforming algorithm is used; If the interference type is determined to be pulse interference, the power inversion algorithm is used; If the interference type is narrowband interference, the sidelobe cancellation algorithm is used; If the interference type is determined to be broadband interference, the STAP algorithm is used.

9. A computer device comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, the steps of the method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by the processor, the steps of the method for real-time monitoring and cancellation of electromagnetic interference based on multi-antenna technology as described in any one of claims 1 to 8 are implemented.

Images