Embedded electromagnetic signal identification processing method and device

Abstract

The invention provides an embedded electromagnetic signal identification processing method and device. The method comprises the following steps: acquiring to-be-identified digital electromagnetic signal waveform data; identifying the digital electromagnetic signal waveform data to be identified based on an electromagnetic signal intelligent sensing and analysis model to obtain an electromagnetic signal identification result output by the electromagnetic signal intelligent sensing and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed by fusion based on a convolutional neural network layer and an autocorrelation convolution operation layer; and taking the sample digital electromagnetic signal waveform data and the sample electromagnetic signal identification result corresponding to the sample digital electromagnetic signal waveform data as a training data set to carry out migration training. According to the embedded electromagnetic signal identification processing method provided by the invention, the accuracy and efficiency of electromagnetic signal sensing and identification processing can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of electromagnetic technology, and in particular to an embedded electromagnetic signal recognition and processing method and device, and also to an electronic device, a non-transitory computer-readable storage medium and a computer program product. Background Art

[0002] In recent years, with the rapid development of radio technology, electromagnetic wave signals generated by various types of radiation sources overlap and change over time in wireless space. The increasingly complex electromagnetic environment has brought major challenges to applications such as communication and reconnaissance. Intelligent communication technology has gradually become a research hotspot in the electromagnetic field. In this context, analyzing the essential characteristic mechanism of signals and realizing the perception and recognition of electromagnetic signals in complex environments are of far-reaching significance for improving the utilization rate of spectrum resources and the anti-interference performance of communications, promoting the effective combination of artificial intelligence methods and digital signal theory, and realizing electromagnetic space situational awareness. However, the accuracy and efficiency of existing electromagnetic signal recognition and processing have great limitations, which greatly limits the perception and analysis of electromagnetic signals. Therefore, how to design a more efficient electromagnetic signal recognition and processing solution has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention provides an embedded electromagnetic signal recognition and processing method to solve the defect in the prior art that the accuracy and efficiency of electromagnetic signal recognition and processing are greatly limited, resulting in poor electromagnetic signal perception and analysis effects.

[0004] The present invention provides an embedded electromagnetic signal recognition and processing method, comprising: Acquire digital electromagnetic signal waveform data to be identified; The digital electromagnetic signal waveform data to be identified is identified based on the electromagnetic signal intelligent perception and analysis model to obtain the electromagnetic signal identification result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of the convolutional neural network layer and the autocorrelation convolution operation layer, and is obtained by migration training based on the sample digital electromagnetic signal waveform data and the sample electromagnetic signal identification result corresponding to the sample digital electromagnetic signal waveform data as the training data set; the electromagnetic signal intelligent perception and analysis model is used to extract the periodic attributes of the sample digital electromagnetic signal waveform data and obtain the representation characteristics of the periodic attributes, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

[0005] According to the embedded electromagnetic signal recognition and processing method proposed in the present invention, the digital electromagnetic signal waveform data to be recognized is recognized and processed based on the electromagnetic signal intelligent perception and analysis model to obtain the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model, specifically including: Inputting the digital electromagnetic signal waveform data to be identified into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain a first convolution output result and a second convolution output result output by the convolutional neural network layer; Wherein, the digital electromagnetic signal waveform data to be identified includes first sampled digital electromagnetic signal waveform data and second sampled digital electromagnetic signal waveform data; a sampling time point of the first sampled digital electromagnetic signal waveform data and a sampling time point of the second sampled digital electromagnetic signal waveform data satisfy a preset delay time interval; the first convolution output result is a convolution feature obtained by convolution processing the first sampled digital electromagnetic signal waveform data; the second convolution output result is a convolution feature obtained by convolution processing the second sampled digital electromagnetic signal waveform data; Input the first convolution output result and the second convolution output result into the autocorrelation convolution operation layer of the electromagnetic signal intelligent perception and analysis model for autocorrelation operation processing to obtain a third convolution output result output by the autocorrelation convolution operation layer; determine the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be identified based on the third convolution output result; the electromagnetic signal recognition result is the electromagnetic signal interference category corresponding to the digital electromagnetic signal waveform data to be identified; Among them, the third convolution output result is the convolution feature of the electromagnetic signal obtained after suppressing the background noise signal in the digital electromagnetic signal waveform data to be identified.

[0006] According to the embedded electromagnetic signal recognition and processing method proposed in the present invention, the digital electromagnetic signal waveform data to be recognized is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer, which specifically includes: The data stream of the digital electromagnetic signal waveform data to be identified is segmented to obtain a plurality of data blocks, and the data blocks are stored as data files to be read according to preset custom parameters; the data files respectively store the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data; wherein the custom parameters include the number of sampling points of a single data file, the number of data files, and the location parameters for storing the data files; The data file is read by the electromagnetic signal intelligent perception and analysis model, and the data file is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer.

[0007] According to the embedded electromagnetic signal recognition and processing method proposed by the present invention, before the digital electromagnetic signal waveform data to be recognized is recognized based on the electromagnetic signal intelligent perception and analysis model, and the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model is obtained, the method further includes: Obtain the constructed initial electromagnetic signal intelligent perception and analysis model; Obtain sample digital electromagnetic signal waveform data and a sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data; use the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data as a training data set, and perform transfer learning training on the initial electromagnetic signal intelligent perception and analysis model based on a model parameter migration strategy to obtain the trained electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model includes a convolutional neural network layer and an autocorrelation convolution operation layer embedded in the convolutional neural network layer.

[0008] According to the embedded electromagnetic signal recognition and processing method proposed in the present invention, determining the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be recognized according to the third convolution output result specifically includes: Inputting the third convolution output result into a preset loss function of the electromagnetic signal intelligent perception and analysis model to obtain a corresponding electromagnetic signal data classification label; The electromagnetic signal interference category is determined according to the electromagnetic signal data classification label, and the electromagnetic signal interference category is output as an electromagnetic signal recognition result.

[0009] According to the embedded electromagnetic signal recognition and processing method proposed in the present invention, after obtaining the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model, it also includes: Performing spectrum analysis on the digital electromagnetic signal waveform data to be identified to map the digital electromagnetic signal waveform data to be identified into electromagnetic spectrum data; Obtaining an electromagnetic signal interference category corresponding to the electromagnetic signal identification result; The electromagnetic spectrum data and the electromagnetic signal interference category are sent to a client for display in a graphical form.

[0010] The present invention also provides an embedded electromagnetic signal recognition and processing device, comprising: An electromagnetic signal acquisition unit, used to acquire waveform data of a digital electromagnetic signal to be identified; An electromagnetic signal recognition processing unit, used to recognize the digital electromagnetic signal waveform data to be recognized based on the electromagnetic signal intelligent perception and analysis model, and obtain the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; The electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as training data sets; The electromagnetic signal intelligent perception and analysis module is used to extract the representation characteristics of the periodic attributes of the sample digital electromagnetic signal waveform data, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

[0011] The electromagnetic signal intelligent perception and analysis module is used to extract the representation characteristics of the periodic attributes of the sample digital electromagnetic signal waveform data, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

[0012] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, the electromagnetic signal recognition and processing unit is specifically used for: Inputting the digital electromagnetic signal waveform data to be identified into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain a first convolution output result and a second convolution output result output by the convolutional neural network layer; Wherein, the digital electromagnetic signal waveform data to be identified includes first sampled digital electromagnetic signal waveform data and second sampled digital electromagnetic signal waveform data; a sampling time point of the first sampled digital electromagnetic signal waveform data and a sampling time point of the second sampled digital electromagnetic signal waveform data satisfy a preset delay time interval; the first convolution output result is a convolution feature obtained by convolution processing the first sampled digital electromagnetic signal waveform data; the second convolution output result is a convolution feature obtained by convolution processing the second sampled digital electromagnetic signal waveform data; Input the first convolution output result and the second convolution output result into the autocorrelation convolution operation layer of the electromagnetic signal intelligent perception and analysis model for autocorrelation operation processing to obtain a third convolution output result output by the autocorrelation convolution operation layer; determine the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be identified based on the third convolution output result; the electromagnetic signal recognition result is the electromagnetic signal interference category corresponding to the digital electromagnetic signal waveform data to be identified; Among them, the third convolution output result is the convolution feature of the electromagnetic signal obtained after suppressing the background noise signal in the digital electromagnetic signal waveform data to be identified.

[0013] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, the digital electromagnetic signal waveform data to be recognized is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer, specifically including: The data stream of the digital electromagnetic signal waveform data to be identified is segmented to obtain a plurality of data blocks, and the data blocks are stored as data files to be read according to preset custom parameters; the data files respectively store the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data; wherein the custom parameters include the number of sampling points of a single data file, the number of data files, and the location parameters for storing the data files; The data file is read by the electromagnetic signal intelligent perception and analysis model, and the data file is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer.

[0014] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, before the digital electromagnetic signal waveform data to be recognized is recognized based on the electromagnetic signal intelligent perception and analysis model, and the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model is obtained, the embedded electromagnetic signal recognition and processing device also includes: Model training unit, specifically used for: Obtain the constructed initial electromagnetic signal intelligent perception and analysis model; Obtain sample digital electromagnetic signal waveform data and a sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data; use the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data as a training data set, and perform transfer learning training on the initial electromagnetic signal intelligent perception and analysis model based on a model parameter migration strategy to obtain the trained electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model includes a convolutional neural network layer and an autocorrelation convolution operation layer embedded in the convolutional neural network layer.

[0015] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, determining the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be recognized according to the third convolution output result specifically includes: Inputting the third convolution output result into a preset loss function of the electromagnetic signal intelligent perception and analysis model to obtain a corresponding electromagnetic signal data classification label; The electromagnetic signal interference category is determined according to the electromagnetic signal data classification label, and the electromagnetic signal interference category is output as an electromagnetic signal recognition result.

[0016] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, after obtaining the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model, it also includes: Visual processing unit for: Performing spectrum analysis on the digital electromagnetic signal waveform data to be identified to map the digital electromagnetic signal waveform data to be identified into electromagnetic spectrum data; Obtaining an electromagnetic signal interference category corresponding to the electromagnetic signal identification result; The electromagnetic spectrum data and the electromagnetic signal interference category are sent to a client for display in a graphical form.

[0017] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the embedded electromagnetic signal recognition and processing method as described in any one of the above items is implemented.

[0018] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the embedded electromagnetic signal recognition and processing method as described in any one of the above items is implemented.

[0019] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the embedded electromagnetic signal recognition and processing method as described in any one of the above items is implemented.

[0020] The embedded electromagnetic signal recognition and processing method provided by the present invention obtains digital electromagnetic signal waveform data to be recognized; recognizes the digital electromagnetic signal waveform data to be recognized based on an electromagnetic signal intelligent perception and analysis model, and obtains an electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as training data sets; it can effectively improve the accuracy and efficiency of electromagnetic signal perception and recognition processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 This is one of the flow charts of the embedded electromagnetic signal recognition and processing method provided by the present invention.

[0023] Figure 2 This is the second flow chart of the embedded electromagnetic signal recognition and processing method provided by the prior art.

[0024] Figure 3 It is a schematic diagram of the hardware structure connection corresponding to the embedded electromagnetic signal recognition and processing method provided by the present invention.

[0025] Figure 4 It is a schematic diagram of the electromagnetic signal intelligent perception and analysis model provided by the present invention.

[0026] Figure 5 It is a schematic diagram of the electromagnetic signal recognition result provided by the present invention.

[0027] Figure 6 It is a structural schematic diagram of the embedded electromagnetic signal recognition and processing device provided by the present invention.

[0028] Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] Combine the following Figure 1-Figure 7 The embedded electromagnetic signal recognition and processing method and device of the present invention are described, and the embodiments thereof are described in detail.

[0031] The following describes in detail the embodiment of the embedded electromagnetic signal recognition and processing method according to the present invention. Figure 1 As shown, it is a flow chart of the embedded electromagnetic signal recognition and processing method provided by the present invention, and the specific implementation process includes the following steps: Step 101: Acquire waveform data of a digital electromagnetic signal to be identified.

[0032] In an embodiment of the present invention, before executing this step, it is necessary to pre-build a model and train the model. Specifically, obtain a constructed initial electromagnetic signal intelligent perception and analysis model; obtain sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data; use the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as a training data set, and perform transfer learning training on the initial electromagnetic signal intelligent perception and analysis model based on the model parameter migration strategy to obtain the trained electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model includes a convolutional neural network layer and an autocorrelation convolution operation layer embedded in the convolutional neural network layer.

[0033] It should be noted that based on the sample digital electromagnetic signal waveform data, the initial electromagnetic signal intelligent perception and analysis model is constructed using the deep learning method, and the network parameter training is completed after the simulation function fitting is completed. The initial electromagnetic signal intelligent perception and analysis model is obtained and can be put into practical application. In the process of offline algorithm training and verification, the deep neural network has been able to obtain high recognition accuracy on the open source classic data set, but the data set does not cover the real influencing factors enough, resulting in deviations between the training data and the real test data. The difference in data distribution leads to the insufficient adaptability of the intelligent model to the different types of electromagnetic interference. After the data distribution changes, if the data is re-collected, labeled, and then trained, the time cost and resource consumption are high. In order to solve the impact of the offline training model on the performance of signal instability factors such as interference type, transmission channel, and receiving equipment, the system parameters need to be adjusted to adapt to the current environment. Transfer learning migrates the trained model parameters to the new model to help the new model training. Considering that most data or tasks are related, the learned model parameters (i.e., the knowledge learned by the model) can be shared with the new model through transfer learning, thereby accelerating and optimizing the learning efficiency of the model. This technology allows the model to be trained without starting from scratch, and uses the pre-existing network architecture to quickly start the learning process, thereby obtaining a high-precision transfer learning model, which can effectively reduce the workload. The transfer learning process includes: at the beginning of the training phase, the training process only updates the parameters of the fully connected layer that are replaced twice, and selects a smaller learning rate for training, such as 0.0001, so that it will not have a significant impact on the overall parameters of the model; after the parameter learning of the fully connected layer is completed, the original network layer parameters are activated, and the model is trained as a whole. Therefore, in the present invention, the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data are used as training data sets, and the initial electromagnetic signal intelligent perception and analysis model is transferred based on the model parameter migration strategy. The trained electromagnetic signal intelligent perception and analysis model is obtained, thereby effectively improving the overall training efficiency of the model.

[0034] Step 102: Identify the digital electromagnetic signal waveform data to be identified based on the electromagnetic signal intelligent perception and analysis model, and obtain the electromagnetic signal identification result output by the electromagnetic signal intelligent perception and analysis model. The electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of the convolutional neural network layer and the autocorrelation convolution operation layer, and is obtained by migration training based on the sample digital electromagnetic signal waveform data and the sample electromagnetic signal identification result corresponding to the sample digital electromagnetic signal waveform data as the training data set; the electromagnetic signal intelligent perception and analysis model is used to extract the periodic attribute of the sample digital electromagnetic signal waveform data and obtain the periodic attribute representation feature, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the periodic attribute representation feature, and obtain the perception and analysis result.

[0035] In an embodiment of the present invention, the computer device may input the digital electromagnetic signal waveform data to be identified (i.e., the intermediate frequency digital signal) into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer. The digital electromagnetic signal waveform data to be identified includes the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data; the sampling time point of the first sampled digital electromagnetic signal waveform data and the sampling time point of the second sampled digital electromagnetic signal waveform data meet the preset delay time interval, that is, the second sampled digital electromagnetic signal waveform data is the digital electromagnetic signal waveform data collected after the preset delay time interval of collecting the first sampled digital electromagnetic signal waveform data. The first convolution output result is the convolution feature obtained by convolution processing the first sampled digital electromagnetic signal waveform data; the second convolution output result is the convolution feature obtained by convolution processing the second sampled digital electromagnetic signal waveform data. Furthermore, the first convolution output result and the second convolution output result can be input into the autocorrelation convolution operation layer of the electromagnetic signal intelligent perception and analysis model for autocorrelation operation processing to obtain a third convolution output result output by the autocorrelation convolution operation layer; according to the third convolution output result, the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be identified is determined. In the process of inputting the digital electromagnetic signal waveform data to be identified into the convolution neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolution neural network layer, the data stream of the digital electromagnetic signal waveform data to be identified can be segmented to obtain multiple data blocks, and the data blocks can be stored as data files to be read according to preset custom parameters; the data files respectively store the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data; wherein the custom parameters include the number of sampling points of a single data file, the number of data files, and the location parameters for saving data files. The data file is read through the electromagnetic signal intelligent perception and analysis model, and the data file is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer. In the process of determining the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be recognized according to the third convolution output result, the third convolution output result can be input into the preset loss function of the electromagnetic signal intelligent perception and analysis model to obtain the corresponding electromagnetic signal data classification label; the electromagnetic signal interference category is determined according to the electromagnetic signal data classification label, and the electromagnetic signal interference category is output as the electromagnetic signal recognition result.The electromagnetic signal recognition result is the electromagnetic signal interference category corresponding to the digital electromagnetic signal waveform data to be identified. The third convolution output result is the convolution feature of the electromagnetic signal obtained after suppressing the background noise signal in the digital electromagnetic signal waveform data to be identified. The convolutional neural network layer is a one-dimensional CNN (Convolutional Neural Networks) network. The computer device is a server.

[0036] The present invention integrates a one-dimensional CNN network with statistical autocorrelation operations. Since the wireless electromagnetic signal is artificially generated, the variation pattern of the signal waveform contains symbol data information, and the symbol data is encoded in digital form and used cyclically in the communication system, so that the waveform of the wireless electromagnetic signal (i.e., digital electromagnetic signal waveform data) presents periodic characteristics. In addition, non-artificial noise data generally does not have periodic characteristics. Therefore, introducing a periodic processing method in the process of extracting wireless electromagnetic signal features can improve the ability to characterize the signal and suppress noise. In traditional signal statistical features, the signal cyclic spectrum and high-order cumulants extract the periodic characteristics of sample data based on correlation operations. Therefore, on the basis of extracting local features of signals through the classic one-dimensional CNN network, the present invention constructs the autocorrelation operation as a special network layer (i.e., the autocorrelation convolution operation layer) to realize the extraction of periodic local features of electromagnetic signals. Among them, the one-dimensional CNN network is constructed by size The convolution kernel is used to obtain the convolution output result The process is as follows:

[0037] in, Represents the original electromagnetic signal (i.e. the original digital electromagnetic signal waveform data) from Time to The sampling data at the time (for example, the digital electromagnetic signal waveform data to be identified includes the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data), is the bias, is a nonlinear activation function; W is the weight parameter. Since the nonlinear activation function will limit the commutativity of multiplication, in order to integrate the autocorrelation convolution operation layer in the one-dimensional CNN network, the convolution output of the data is nonlinearly activated after the autocorrelation convolution operation layer. Electromagnetic signal characteristic diagram The representation is as follows:

[0038] in, is the convolution operation; is the actual electromagnetic signal; is the background noise signal; y(n) is a mixed electromagnetic signal containing an actual electromagnetic signal and a background noise signal, that is, a signal with electromagnetic interference; and The electromagnetic signal characteristic graph is the convolution output result obtained after the electromagnetic signal is convolved, such as the first convolution output result and the second convolution output result.

[0039] The present invention designs the traditional signal statistical feature autocorrelation operation into a special network layer structure (i.e. Figure 4 The autocorrelation convolution operation layer shown in the figure, which performs the corresponding autocorrelation operation), is integrated into the one-dimensional CNN network to enable it to have periodic learning capabilities. Integrating electromagnetic field correlation knowledge into deep neural networks mainly considers two aspects: (1) For currently known communication systems, communication signals have periodic characteristics. Enabling neural networks to have periodic learning capabilities can improve the applicability of the network in the electromagnetic field and improve the electromagnetic signal characterization capabilities; (2) Non-artificial background noise in the electromagnetic environment does not have periodicity. Autocorrelation operations are used to achieve signal data feature mapping. This process can suppress non-periodic background noise. The autocorrelation convolution operation layer reflects the similarity between the original time series and its own sequence within a continuous time interval (i.e., the Pearson correlation coefficient). Unlike calculating the similarity between two different time series, autocorrelation uses the same time series, one of which is the original data (i.e., the first sampled digital electromagnetic signal waveform data) and the other is the delayed interval. The original time series (i.e. the second sampled digital electromagnetic signal waveform data) is , autocorrelation operation ( ) is defined as follows: In order to simplify the operation process and efficiently integrate it into the one-dimensional CNN network, the autocorrelation operation algorithm corresponding to the autocorrelation convolution operation layer in the present invention is simplified as follows:

[0040] in, The electromagnetic signal characteristic map output by the one-dimensional CNN network mapping the original electromagnetic signal waveform is the convolution output result obtained after the electromagnetic signal is convolved, such as the first convolution output result c (1) Second convolution output result c (2), The preset delay time interval ( ), is the data index of the electromagnetic signal characteristic graph, the result of autocorrelation operation It has the following properties: (1) and When the similarity is high, the autocorrelation operation result (i.e., the third convolution output result) will be displayed at a preset delay time interval of Generate a peak. The electromagnetic signal characteristic diagram (i.e. characteristic diagram ) has a period of ,but exist There are periodic peaks at is a positive integer; (2) the result of autocorrelation operation is an even function, that is Symmetric about the zero point of the coordinate axis, Therefore, when integrating the autocorrelation convolution operation layer into the convolutional neural network layer, only half of the results need to be output, that is, ,This process can simplify the network structure and reduce the amount of computation.

[0041] Therefore, the algorithm corresponding to the autocorrelation convolution operation layer proposed in the present invention is as follows (4):

[0042] For the electromagnetic signal at the receiving end, the autocorrelation operation result (that is, the third convolution output result) is as follows:

[0043] in, is the autocorrelation result of the electromagnetic signal, is the cross-correlation result between the electromagnetic signal and the background noise, is the autocorrelation result of the background noise. The electromagnetic signal and the background noise are obviously very similar. Can be ignored, that is ,In addition, non-artificial background noise is also not relevant.

[0044] Therefore, the above formula (5) can be simplified as follows: ( )= C [ ( )] (6)

[0045] The fusion process of the autocorrelation convolution operation layer and the convolutional neural network layer (i.e., convolutional neural network) is as follows Figure 4As shown. The fusion process of the autocorrelation convolution operation layer and the convolutional neural network layer. The convolutional neural network is a one-dimensional CNN network. Integrating the autocorrelation convolution operation network layer into the one-dimensional CNN network can retain the learning ability of the one-dimensional CNN network for local features of the data, and at the same time, based on the essential periodic properties of electromagnetic signals, improve the electromagnetic signal intelligent perception and analysis model's ability to characterize periodic electromagnetic signals. In addition, the autocorrelation convolution operation can suppress random background noise signals, making the electromagnetic signal intelligent perception and analysis model suitable for situations with low signal-to-noise ratio. Among them, Figure 4 401 is a two-dimensional convolution kernel, 402 to 404 are the corresponding two-dimensional data, 405 to 407 are the corresponding one-dimensional electromagnetic data, 408 is a one-dimensional convolution kernel, and 409 is the electromagnetic feature data obtained after convolution (i.e., the convolution feature of the electromagnetic signal obtained after suppressing the background noise signal in the digital electromagnetic signal waveform data to be identified).

[0046] In addition, after obtaining the electromagnetic signal identification result output by the electromagnetic signal intelligent perception and analysis model, spectral analysis can be performed on the digital electromagnetic signal waveform data to be identified to map the digital electromagnetic signal waveform data to be identified into electromagnetic spectrum data; obtain the electromagnetic signal interference category corresponding to the electromagnetic signal identification result; and send the electromagnetic spectrum data and the electromagnetic signal interference category to the client for display in a graphical form.

[0047] Because the intelligence and interpretability of features in the existing technology restrict each other: (1) Traditional electromagnetic features are driven by predefined models, and the interpretability of this feature is high, but it relies on prior knowledge, resulting in insufficient representation capabilities of traditional signal features, and is only applicable to ideal situations where interference factors are controllable; (2) Intelligent electromagnetic features are driven by massive data, and the intelligence of this feature is high, but it does not consider the unique properties of electromagnetic signals, the interpretability is low, and the adaptability needs to be improved. Traditional electromagnetic features are based on artificially predefined data mapping models, and the lack of a closed-loop mechanism leads to insufficient representation capabilities and poor robustness of traditional electromagnetic features; while deep neural networks rely on data-driven parameter fitting, do not consider the unique properties of electromagnetic signals, and have low adaptability when applied to the electromagnetic field. In order to solve the problems of poor performance of traditional electromagnetic features and insufficient interpretability of intelligent electromagnetic features, the present invention starts from the perspective of the essential properties of electromagnetic signals themselves, analyzes the characteristic mechanisms contained in the essential properties of electromagnetic signals, and explores the intelligent feature extraction method of electromagnetic signals under the premise of knowledge constraints in the electromagnetic field, realizes "periodic cyclic learning" of electromagnetic features in the time domain dimension, constructs an electromagnetic intelligent feature description model with a high degree of interpretability (i.e., an electromagnetic signal intelligent perception and analysis model), promotes the effective combination of artificial intelligence methods and digital signal theory, and provides a theoretical basis for signal perception and individual identification.

[0048] In terms of the corresponding hardware structure, such as Figure 3 As shown in the figure, the electromagnetic interference intelligent perception and analysis equipment consists of three parts: electromagnetic signal data acquisition module, embedded hardware board and visual touch LCD screen. The electromagnetic interference intelligent perception and analysis equipment has the ability to generate electromagnetic signals, collect analog signal data, embed the electromagnetic signal intelligent perception and analysis model, monitor the time / frequency domain waveform in real time and control the visual LCD touch screen. The working process of the electromagnetic interference intelligent perception and analysis equipment is as follows Figure 2 As shown, the intelligent recognition model under the core processing module is an electromagnetic signal intelligent perception and analysis model.

[0049] In the intelligent perception process of electromagnetic interference, the embedded hardware board consists of two parts: (1) a high-speed analog / digital signal data acquisition module (i.e., electromagnetic signal data acquisition module); (2) an embedded core intelligent processing module, in which the electromagnetic signal intelligent perception and analysis model is run. The working process of the electromagnetic interference intelligent perception and analysis equipment is as follows: (1) The electromagnetic signal data acquisition module is based on the SMA hardware interface, and connects the medium-frequency electromagnetic signal to the hardware device to the outside, collects the medium-frequency analog electromagnetic signal data in real time and caches it; (2) The embedded core intelligent processing module reads the digital electromagnetic signal waveform data to be identified in the acquisition card through the network port protocol, and after pre-processing by the digital signal processing algorithm, visualizes the electromagnetic signal time / frequency domain waveform on the LCD screen; (3) The embedded core intelligent processing module identifies and analyzes the interference type in real time based on the deployed electromagnetic signal intelligent perception and analysis model, and the electromagnetic signal recognition result (i.e., the perception result) is synchronously displayed in the main area of ​​the LCD screen and the electromagnetic signal recognition result is transmitted to the outside through the serial port. The hardware of the electromagnetic interference intelligent perception and analysis equipment is as follows: Figure 3 shown.

[0050] The embedded electromagnetic signal recognition and processing method provided by the present invention obtains digital electromagnetic signal waveform data to be recognized; recognizes the digital electromagnetic signal waveform data to be recognized based on an electromagnetic signal intelligent perception and analysis model, and obtains an electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as training data sets; it can effectively improve the accuracy and efficiency of electromagnetic signal perception and recognition processing.

[0051] The embedded electromagnetic signal recognition and processing device provided by the present invention is described below. The embedded electromagnetic signal recognition and processing device described below and the embedded electromagnetic signal recognition and processing method described above can be referred to each other. Figure 6As shown, it is a schematic diagram of the structure of the embedded electromagnetic signal recognition and processing device provided by the present invention. The embedded electromagnetic signal recognition and processing device of the present invention specifically includes the following parts: The electromagnetic signal acquisition unit 601 is used to acquire the waveform data of the digital electromagnetic signal to be identified.

[0052] The electromagnetic signal recognition processing unit 602 is used to identify the digital electromagnetic signal waveform data to be identified based on the electromagnetic signal intelligent perception and analysis model, and obtain the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of the convolutional neural network layer and the autocorrelation convolution operation layer, and is obtained by migration training based on the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as the training data set; the electromagnetic signal intelligent perception and analysis model is used to extract the periodic attributes of the sample digital electromagnetic signal waveform data and obtain the representation characteristics of the periodic attributes, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

[0053] It should be noted that the electromagnetic signal recognition and processing unit 602 (i.e., embedded core intelligent processing module) is developed on python, C++ and pyqt platforms, the operation page is developed based on the Linux system, the main algorithms involved in each module are written in python, and QT GUI (i.e., Graphical User Interface) is used to implement the human-computer interaction interface design, and the system is tested on the embedded end to verify the usability of the system. The embedded core intelligent processing module transmits data streams from the electromagnetic signal acquisition unit (i.e., electromagnetic signal data acquisition module). This process requires limiting the flow of data streams, so as to control the entire system 601 (i.e., embedded electromagnetic signal recognition and processing device) to run in a smooth and stable state. At the same time, the parameters required by the electromagnetic signal data acquisition module, such as the transmission channel, sampling rate, and data volume, are set.

[0054] Based on the data stream of the parameters required by the electromagnetic signal data acquisition module, the embedded core intelligent processing module divides the data stream. The embedded core intelligent processing module reads any binary data, can create individual files (i.e., data files) for the data stream, and saves the data in the data file. For multiple types of electromagnetic signal interference, the system needs to collect and save multiple data files, and for ease of use, the number of data files needs to be limited. Therefore, the present invention develops a custom Python OOT module built into the grc flow graph, and the Python OOT module inherits gr.sync_block to realize the synchronization function, obtains the data block, and sets custom parameters including the number of sampling points of a single data file, the number of data files, and the location parameters for saving the data file. Since the intermediate frequency analog signal can collect a large amount of signal data in a short time, in order to reduce resource waste, collect data for a longer time as much as possible, and set the sampling interval time additionally, the collected signal data (i.e., the digital electromagnetic signal waveform data to be identified) will be saved in a format recognizable by Numpy, which is convenient for waveform visualization and use of deep learning frameworks such as Tensorflow / Pytorch (i.e., embedded core intelligent processing module).

[0055] In the process of visualizing the electromagnetic signal time / frequency domain waveform, the electromagnetic signal time / frequency domain waveform and electromagnetic interference type are visualized based on the intelligent perception results (i.e., electromagnetic signal recognition results) output by the electromagnetic signal data acquisition module and the electromagnetic signal recognition processing unit. Among them, the electromagnetic signal time domain waveform can be directly displayed through the visualization control based on the output results of the electromagnetic signal data acquisition module. Both the electromagnetic spectrum data and the intelligent perception results need to be calculated based on the time domain waveform information and then visualized.

[0056] The 10 common interference signals of digital electromagnetic signal waveform data include single-tone interference, dual-tone interference, deceptive interference (4ASK, 4FSK, 4PSK), broadband frequency sweep, mixed interference 1, mixed interference 2, broadband interference and narrowband interference. The generated data includes various frequency points, bandwidths, and Gaussian white noise is added. In an embodiment of the present invention, it is possible to set the training data set to have 2000 data for each type of interference signal, and set the test set to have 2000 data for each type of interference signal. After the initial iterative training of the electromagnetic signal intelligent perception and analysis model, the obtained electromagnetic signal intelligent perception and analysis model was tested, and the recognition accuracy of the training data set was 99.1%, and the accuracy of the test data set was 93.45%. The classification confusion matrix is ​​as follows Figure 5 shown.

[0057] The electromagnetic signal intelligent perception and analysis module is used to extract the representation characteristics of the periodic attributes of the sample digital electromagnetic signal waveform data, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

[0058] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, the electromagnetic signal recognition and processing unit is specifically used for: Inputting the digital electromagnetic signal waveform data to be identified into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain a first convolution output result and a second convolution output result output by the convolutional neural network layer; Wherein, the digital electromagnetic signal waveform data to be identified includes first sampled digital electromagnetic signal waveform data and second sampled digital electromagnetic signal waveform data; a sampling time point of the first sampled digital electromagnetic signal waveform data and a sampling time point of the second sampled digital electromagnetic signal waveform data satisfy a preset delay time interval; the first convolution output result is a convolution feature obtained by convolution processing the first sampled digital electromagnetic signal waveform data; the second convolution output result is a convolution feature obtained by convolution processing the second sampled digital electromagnetic signal waveform data; Input the first convolution output result and the second convolution output result into the autocorrelation convolution operation layer of the electromagnetic signal intelligent perception and analysis model for autocorrelation operation processing to obtain a third convolution output result output by the autocorrelation convolution operation layer; determine the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be identified based on the third convolution output result; the electromagnetic signal recognition result is the electromagnetic signal interference category corresponding to the digital electromagnetic signal waveform data to be identified; Among them, the third convolution output result is the convolution feature of the electromagnetic signal obtained after suppressing the background noise signal in the digital electromagnetic signal waveform data to be identified.

[0059] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, the digital electromagnetic signal waveform data to be recognized is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer, specifically including: The data stream of the digital electromagnetic signal waveform data to be identified is segmented to obtain a plurality of data blocks, and the data blocks are stored as data files to be read according to preset custom parameters; the data files respectively store the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data; wherein the custom parameters include the number of sampling points of a single data file, the number of data files, and the location parameters for storing the data files; The data file is read by the electromagnetic signal intelligent perception and analysis model, and the data file is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer.

[0060] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, before the digital electromagnetic signal waveform data to be recognized is recognized based on the electromagnetic signal intelligent perception and analysis model, and the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model is obtained, the embedded electromagnetic signal recognition and processing device also includes: Model training unit, specifically used for: Obtain the constructed initial electromagnetic signal intelligent perception and analysis model; Obtain sample digital electromagnetic signal waveform data and a sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data; use the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data as a training data set, and perform transfer learning training on the initial electromagnetic signal intelligent perception and analysis model based on a model parameter migration strategy to obtain the trained electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model includes a convolutional neural network layer and an autocorrelation convolution operation layer embedded in the convolutional neural network layer.

[0061] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, determining the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be recognized according to the third convolution output result specifically includes: Inputting the third convolution output result into a preset loss function of the electromagnetic signal intelligent perception and analysis model to obtain a corresponding electromagnetic signal data classification label; The electromagnetic signal interference category is determined according to the electromagnetic signal data classification label, and the electromagnetic signal interference category is output as an electromagnetic signal recognition result.

[0062] According to the embedded electromagnetic signal recognition and processing device mentioned in the present invention, after obtaining the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model, it also includes: Visual processing unit for: Performing spectrum analysis on the digital electromagnetic signal waveform data to be identified to map the digital electromagnetic signal waveform data to be identified into electromagnetic spectrum data; Obtaining an electromagnetic signal interference category corresponding to the electromagnetic signal identification result; The electromagnetic spectrum data and the electromagnetic signal interference category are sent to a client for display in a graphical form.

[0063] The embedded electromagnetic signal recognition and processing device provided by the present invention obtains digital electromagnetic signal waveform data to be recognized; recognizes the digital electromagnetic signal waveform data to be recognized based on an electromagnetic signal intelligent perception and analysis model, and obtains an electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as training data sets; it can effectively improve the accuracy and efficiency of electromagnetic signal perception and recognition processing.

[0064] Figure 7 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 7 As shown, the electronic device (i.e., computer device) may include: a processor (processor) 701, a communication interface (Communications Interface) 704, a memory (memory) 702 and a communication bus 703, wherein the processor 701, the communication interface 704, and the memory 702 communicate with each other through the communication bus 703. The processor 701 can call the logic instructions in the memory 702 to execute the embedded electromagnetic signal recognition and processing method, which includes: obtaining digital electromagnetic signal waveform data to be recognized; recognizing the digital electromagnetic signal waveform data to be recognized based on the electromagnetic signal intelligent perception and analysis model, and obtaining the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of the convolutional neural network layer and the autocorrelation convolution operation layer, and is obtained by migration training based on the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data as the training data set; the electromagnetic signal intelligent perception and analysis model is used to extract the periodic attributes of the sample digital electromagnetic signal waveform data based on the representation characteristics of the periodic attributes, and perceive and analyze the digital electromagnetic signal waveform data to be recognized based on the representation characteristics of the periodic attributes.

[0065] In addition, the logic instructions in the above-mentioned memory 702 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0066] On the other hand, the present application also provides a computer program product, which includes a computer program, which can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the embedded electromagnetic signal recognition and processing method provided by the above-mentioned methods, and the method includes: obtaining digital electromagnetic signal waveform data to be identified; identifying the digital electromagnetic signal waveform data to be identified based on an electromagnetic signal intelligent perception and analysis model, and obtaining an electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as a training data set; the electromagnetic signal intelligent perception and analysis model is used to extract the periodic attributes of the sample digital electromagnetic signal waveform data to obtain the representation characteristics of the periodic attributes, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

[0067] On the other hand, the present application also provides a computer-readable storage medium, which includes a stored program, wherein the program, when running, executes the embedded electromagnetic signal recognition and processing method provided by the above-mentioned methods, the method comprising: obtaining digital electromagnetic signal waveform data to be recognized; recognizing the digital electromagnetic signal waveform data to be recognized based on an electromagnetic signal intelligent perception and analysis model, and obtaining an electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as a training data set; the electromagnetic signal intelligent perception and analysis model is used to extract the periodic attributes of the sample digital electromagnetic signal waveform data based on the representation characteristics of the periodic attributes, and perceive and analyze the digital electromagnetic signal waveform data to be recognized based on the representation characteristics of the periodic attributes.

[0068] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0069] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An embedded electromagnetic signal recognition and processing method, characterized in that: include: Acquire digital electromagnetic signal waveform data to be identified; Identify the digital electromagnetic signal waveform data to be identified based on the electromagnetic signal intelligent perception and analysis model, and obtain the electromagnetic signal identification result output by the electromagnetic signal intelligent perception and analysis model; The electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as training data sets; The electromagnetic signal intelligent perception and analysis module is used to extract the representation characteristics of the periodic attributes of the sample digital electromagnetic signal waveform data, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

2. The embedded electromagnetic signal recognition and processing method according to claim 1, characterized in that: The step of identifying the digital electromagnetic signal waveform data to be identified based on the electromagnetic signal intelligent perception and analysis model to obtain the electromagnetic signal identification result output by the electromagnetic signal intelligent perception and analysis model specifically includes: Inputting the digital electromagnetic signal waveform data to be identified into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain a first convolution output result and a second convolution output result output by the convolutional neural network layer; Wherein, the digital electromagnetic signal waveform data to be identified includes first sampled digital electromagnetic signal waveform data and second sampled digital electromagnetic signal waveform data; a sampling time point of the first sampled digital electromagnetic signal waveform data and a sampling time point of the second sampled digital electromagnetic signal waveform data satisfy a preset delay time interval; the first convolution output result is a convolution feature obtained by convolution processing the first sampled digital electromagnetic signal waveform data; the second convolution output result is a convolution feature obtained by convolution processing the second sampled digital electromagnetic signal waveform data; Input the first convolution output result and the second convolution output result into the autocorrelation convolution operation layer of the electromagnetic signal intelligent perception and analysis model for autocorrelation operation processing to obtain a third convolution output result output by the autocorrelation convolution operation layer; determine the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be identified based on the third convolution output result; the electromagnetic signal recognition result is the electromagnetic signal interference category corresponding to the digital electromagnetic signal waveform data to be identified; Among them, the third convolution output result is the convolution feature of the electromagnetic signal obtained after suppressing the background noise signal in the digital electromagnetic signal waveform data to be identified.

3. The embedded electromagnetic signal recognition and processing method according to claim 2, characterized in that: The step of inputting the digital electromagnetic signal waveform data to be identified into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain a first convolution output result and a second convolution output result output by the convolutional neural network layer specifically includes: The data stream of the digital electromagnetic signal waveform data to be identified is segmented to obtain a plurality of data blocks, and the data blocks are stored as data files to be read according to preset custom parameters; the data files respectively store the first sampled digital electromagnetic signal waveform data and the second sampled digital electromagnetic signal waveform data; wherein the custom parameters include the number of sampling points of a single data file, the number of data files, and the location parameters for storing the data files; The data file is read by the electromagnetic signal intelligent perception and analysis model, and the data file is input into the convolutional neural network layer of the electromagnetic signal intelligent perception and analysis model to obtain the first convolution output result and the second convolution output result output by the convolutional neural network layer.

4. The embedded electromagnetic signal recognition and processing method according to claim 1, characterized in that: Before identifying the digital electromagnetic signal waveform data to be identified based on the electromagnetic signal intelligent perception and analysis model and obtaining the electromagnetic signal identification result output by the electromagnetic signal intelligent perception and analysis model, the method further includes: Obtain the constructed initial electromagnetic signal intelligent perception and analysis model; Obtain sample digital electromagnetic signal waveform data and a sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data; use the sample digital electromagnetic signal waveform data and the sample electromagnetic signal recognition result corresponding to the sample digital electromagnetic signal waveform data as a training data set, and perform transfer learning training on the initial electromagnetic signal intelligent perception and analysis model based on a model parameter migration strategy to obtain the trained electromagnetic signal intelligent perception and analysis model; wherein the electromagnetic signal intelligent perception and analysis model includes a convolutional neural network layer and an autocorrelation convolution operation layer embedded in the convolutional neural network layer.

5. The embedded electromagnetic signal recognition and processing method according to claim 2, characterized in that: Determining the electromagnetic signal recognition result corresponding to the digital electromagnetic signal waveform data to be recognized according to the third convolution output result specifically includes: Inputting the third convolution output result into a preset loss function of the electromagnetic signal intelligent perception and analysis model to obtain a corresponding electromagnetic signal data classification label; The electromagnetic signal interference category is determined according to the electromagnetic signal data classification label, and the electromagnetic signal interference category is output as an electromagnetic signal recognition result.

6. The embedded electromagnetic signal recognition and processing method according to claim 1, characterized in that: After obtaining the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model, the method further includes: Performing spectrum analysis on the digital electromagnetic signal waveform data to be identified to map the digital electromagnetic signal waveform data to be identified into electromagnetic spectrum data; Obtaining an electromagnetic signal interference category corresponding to the electromagnetic signal identification result; The electromagnetic spectrum data and the electromagnetic signal interference category are sent to a client for display in a graphical form.

7. An embedded electromagnetic signal recognition and processing device, characterized in that: include: An electromagnetic signal acquisition unit, used to acquire waveform data of a digital electromagnetic signal to be identified; An electromagnetic signal recognition processing unit, used to recognize the digital electromagnetic signal waveform data to be recognized based on the electromagnetic signal intelligent perception and analysis model, and obtain the electromagnetic signal recognition result output by the electromagnetic signal intelligent perception and analysis model; The electromagnetic signal intelligent perception and analysis model is constructed based on the fusion of a convolutional neural network layer and an autocorrelation convolution operation layer, and is obtained by migration training based on sample digital electromagnetic signal waveform data and sample electromagnetic signal recognition results corresponding to the sample digital electromagnetic signal waveform data as training data sets; The electromagnetic signal intelligent perception and analysis module is used to extract the representation characteristics of the periodic attributes of the sample digital electromagnetic signal waveform data, and perceive and analyze the digital electromagnetic signal waveform data to be identified based on the representation characteristics of the periodic attributes.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the embedded electromagnetic signal recognition and processing method as claimed in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the embedded electromagnetic signal recognition and processing method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the embedded electromagnetic signal recognition and processing method according to any one of claims 1 to 6 is implemented.

Images