Transform-based electromagnetic spectrum identification and tracking algorithm

Abstract

The invention discloses an electromagnetic spectrum identification and tracking algorithm based on transform. Comprising the following steps: step 1, electromagnetic spectrum signal acquisition, step 2, electromagnetic spectrum data preprocessing, step 3, Transform model construction, step 4, target identification, step 5, frequency hopping logic analysis, step 6, target signal parameter acquisition, step 7, target signal type customization, step 8, alarm and countering processing, step 9, historical information playback, and step 10, electromagnetic spectrum signal acquisition, step 5, frequency hopping logic analysis, step 6, target signal parameter acquisition, step 7, target signal type customization, step 8, alarm and countering processing, step 9, historical information playback, and step 10. And optimizing and updating the algorithm. According to the method, interlaced signals in a complex electromagnetic environment are accurately identified, long-distance weak signals are detected with high sensitivity, frequency hopping signals are accurately tracked, the false alarm rate and the missing report rate are reduced, preprocessing is optimized through a band-pass filter and wavelet transform, the feature extraction capacity is improved through a Transform model, the type of a user-defined target signal is supported, continuous optimization is carried out according to feedback, and the detection accuracy is improved. And high performance and adaptability are maintained.

Description

Technical Field

[0001] The present invention relates to the field of electromagnetic spectrum, and particularly to a transform-based electromagnetic spectrum identification and tracking algorithm. Background Art

[0002] In the fields of modern electronic warfare and information confrontation, the monitoring and management of the electromagnetic spectrum are particularly important. With the wide application of unmanned devices such as drones and unmanned ships, as well as the popularization of various electromagnetic emission devices, the electromagnetic environment has become increasingly complex. Traditional electromagnetic signal detection algorithms, such as those based on spectrum analysis, although can meet the basic signal detection requirements to a certain extent, have obvious deficiencies in detection performance and stability when facing complex electromagnetic environments, weak signals at long distances, and frequency-hopping signals.

[0003] Traditional electromagnetic detection algorithms often rely on fixed spectrum features for signal identification, which is effective in relatively simple electromagnetic environments with obvious signal features. However, in complex electromagnetic environments, multiple signals are intertwined and interfere with each other, making it difficult for traditional algorithms to accurately distinguish and identify target signals. At the same time, for weak signals at long distances, the detection sensitivity of traditional algorithms is limited, and they often cannot effectively capture them. In addition, due to the constantly changing frequency characteristics of frequency-hopping signals, traditional algorithms have great difficulties in tracking and identification, easily leading to high false alarm rates and high miss rates. Therefore, a transform-based electromagnetic spectrum identification and tracking algorithm is proposed. Summary of the Invention Aiming at the deficiencies of the prior art, the present invention provides a transform-based electromagnetic spectrum identification and tracking algorithm to solve the problems raised in the above background art.

[0004] To achieve the above object, the present invention provides the following technical solution: A transform-based electromagnetic spectrum identification and tracking algorithm, including the following steps: Step 1: Electromagnetic spectrum signal acquisition; Collect complete electromagnetic spectrum signals in the surrounding environment in real time through electromagnetic acquisition devices to obtain electromagnetic spectrum data, providing basic data support for subsequent target identification and tracking; Step 2: Preprocessing of electromagnetic spectrum data; Convert the electromagnetic spectrum data collected in Step 1 into a single-channel image format to obtain an electromagnetic spectrum image, and process the electromagnetic spectrum image through an image detection algorithm, where the image detection algorithm includes signal filtering, denoising, and normalization preprocessing; Step 3: Construction of the Transformer model; Build a Transformer-based target recognition model, which adopts deep learning technology capable of automatically extracting feature information from electromagnetic spectrum images; Step Four: Target Recognition; Use the target recognition model constructed in Step Three to perform target recognition on the preprocessed electromagnetic spectrum image in Step Two. Extract and classify the features of the electromagnetic spectrum image through the target recognition model, and identify electromagnetic signal targets that are different from the feature information of conventional electromagnetic spectrum images; Step Five: Frequency Hopping Logic Analysis; Analyze the frequency hopping characteristics of the electromagnetic signal targets obtained in Step Four, and combine the frequency hopping logic to achieve continuous tracking of the electromagnetic signal targets; Step Six: Acquisition of Target Signal Parameters; After identifying the electromagnetic signal targets, automatically obtain the basic parameters of the frequency points, bandwidths, and coding methods of the electromagnetic signal targets. The basic parameters are used for signal analysis and processing; Step Seven: Customize Target Signal Types; Dynamically add target information to be matched to the electromagnetic signal targets according to custom requirements; Step Eight: Alarm and Countermeasure Processing; Automatically execute an alarm operation when an electromagnetic signal target is detected, and jointly perform automatic countermeasure processing on the target using countermeasure means, where the countermeasure means includes electromagnetic interference; Step Nine: Historical Information Playback; Record the historical data of electromagnetic spectrum signals, detect and analyze the historical data of electromagnetic spectrum signals, obtain the changing trend of the electromagnetic environment of electromagnetic spectrum signals, and use the changing trend of the electromagnetic environment as a reference for subsequent detection and analysis; Step Ten: Algorithm Optimization and Update; Continuously optimize and update the methods in Step One to Step Nine in real time according to the usage situation and user feedback; By collecting and preprocessing electromagnetic spectrum data in real time, the electromagnetic information of the surrounding environment can be accurately obtained, providing a solid foundation for subsequent target recognition and tracking. Using the Transformer model for target recognition not only improves the recognition accuracy but also enhances the generalization ability of the model. The frequency hopping logic analysis realizes continuous tracking of electromagnetic signal targets, improving the real-time performance and effectiveness of monitoring. At the same time, the functions of automatically obtaining target signal parameters and customizing target signal types meet diverse requirements and improve the flexibility of signal processing. The alarm and countermeasure processing mechanism ensures timely detection and response to potential threats. The historical information playback function helps analyze the changing trend of the electromagnetic environment and provides a reference for future decision-making. Finally, the optimization and update of the algorithm ensure the continuous optimization and improvement of the entire process, adapting to the ever-changing electromagnetic environment.

[0005] Preferably, in the first step: electromagnetic spectrum signal acquisition, there are also a multi-band antenna and a high-speed data acquisition card. The multi-band antenna covers the electromagnetic spectrum range, and the high-speed data acquisition card acquires electromagnetic spectrum signals; The combination of the multi-band antenna and the high-speed data acquisition card brings significant benefits. The multi-band antenna can cover a wide range of the electromagnetic spectrum, ensuring that electromagnetic signals in different frequency bands can be captured, providing a comprehensive and rich data basis for subsequent analysis and processing. At the same time, the high-speed data acquisition card has high-speed and high-precision acquisition capabilities, can acquire electromagnetic spectrum signals in real time and accurately, and avoids signal loss and distortion. The combined use not only improves the efficiency and accuracy of signal acquisition, but also enhances the flexibility and adaptability of the system, enabling the electromagnetic spectrum signal acquisition system to better cope with complex and changing electromagnetic environments. Therefore, the application of the multi-band antenna and the high-speed data acquisition card provides strong support for the processing and analysis of electromagnetic spectrum signals and is an indispensable important part of the electromagnetic spectrum monitoring and identification system.

[0006] Preferably, in the second step: electromagnetic spectrum data preprocessing, there is also a band-pass filter. The passband range of the band-pass filter is dynamically adjusted according to the target electromagnetic spectrum range, and noise and interference signals are filtered out. The denoising uses the wavelet transform method to perform multi-level decomposition and reconstruction on the electromagnetic spectrum image to remove the noise in the image; the normalization preprocessing normalizes the pixel values of the electromagnetic spectrum image to the interval [0, 1] to improve the accuracy of subsequent target recognition; In the process of electromagnetic spectrum data preprocessing, introducing steps such as band-pass filter, wavelet transform denoising, and normalization preprocessing brings many benefits. The band-pass filter can be dynamically adjusted according to its passband range, effectively filtering out noise and interference signals outside the target electromagnetic spectrum range, improving the purity of the data. The wavelet transform denoising method accurately removes the noise in the electromagnetic spectrum image through multi-level decomposition and reconstruction of the image, retaining more useful information. The normalization preprocessing normalizes the pixel values of the electromagnetic spectrum image to the interval [0, 1] uniformly, eliminating the dimensional difference between different images and improving the accuracy and stability of subsequent target recognition. These preprocessing steps together improve the quality of electromagnetic spectrum data and provide a more reliable data basis for subsequent target recognition, tracking, and analysis.

[0007] Preferably, in the third step: building the Transformer model, a multi-layer self-attention mechanism is further included. Each layer of the multi-layer self-attention mechanism includes multiple heads. Each head independently calculates the attention weights, and the outputs of multiple heads are concatenated and linearly transformed to improve the model's ability to extract feature information from electromagnetic spectrum images. During the training process of the Transformer model, the Adam optimization algorithm and the cross-entropy loss function are used to minimize the difference between the predicted labels and the actual labels. In the construction of the Transformer model, the use of a multi-layer self-attention mechanism significantly enhances the model's ability to process electromagnetic spectrum images. Each layer of the self-attention mechanism contains multiple heads, and each head independently calculates the attention weights. This design enables the model to capture different features in the image more meticulously. By concatenating the outputs of multiple heads and performing linear transformation, the model can integrate information from various aspects and improve its ability to extract feature information from electromagnetic spectrum images. In addition, during the model training process, the Adam optimization algorithm can accelerate convergence and optimize model parameters. At the same time, the cross-entropy loss function is used to minimize the difference between the predicted labels and the actual labels, ensuring the prediction accuracy of the model. These designs together enhance the performance of the Transformer model in the electromagnetic spectrum image target recognition task and provide strong support for subsequent electromagnetic signal analysis and processing.

[0008] Preferably, in the fourth step: in target recognition, a non-maximum suppression algorithm is further included. The non-maximum suppression algorithm performs screening on the target detection boxes to remove duplicate and redundant detection boxes, improving the accuracy and efficiency of target recognition. At the same time, the soft non-maximum suppression algorithm is used to attenuate the scores of the detection boxes to handle the overlap problem between detection boxes. During the target recognition process, the introduction of the non-maximum suppression algorithm significantly improves the accuracy and efficiency of recognition. This algorithm screens the target detection boxes, effectively removing duplicate and redundant detection boxes, and avoiding misjudgment and missed detection caused by the overlap of multiple detection boxes. At the same time, the soft non-maximum suppression algorithm is used to attenuate the scores of the detection boxes, skillfully solving the overlap problem between detection boxes. This processing method not only considers the confidence of the detection boxes but also takes into account the relationship between adjacent detection boxes, making the finally retained detection boxes more accurate and reliable. Therefore, the application of the non-maximum suppression algorithm and its soft variant not only improves the accuracy of target recognition but also optimizes the recognition process, making the entire target recognition system more efficient and stable, laying a solid foundation for subsequent electromagnetic signal tracking and analysis.

[0009] Preferably, in the fifth step: frequency hopping logic analysis, there is also a hidden Markov model, which models and analyzes the frequency hopping characteristics of the electromagnetic signal target, predicts the frequency hopping pattern and frequency hopping points of the electromagnetic signal target, and combines the Kalman filtering algorithm to estimate and predict the motion state of the electromagnetic signal target, so as to improve the continuous tracking ability of the electromagnetic signal target; In frequency hopping logic analysis, integrating the hidden Markov model significantly enhances our ability to understand the frequency hopping characteristics of electromagnetic signal targets. This model can accurately model and analyze the frequency hopping pattern and frequency hopping points of electromagnetic signal targets, thereby predicting the frequency hopping behavior of the targets. This predictive ability is crucial for grasping the dynamic changes of electromagnetic signal targets. At the same time, combined with the Kalman filtering algorithm, we can estimate and predict the motion state of electromagnetic signal targets in real time. This combined method not only improves the accuracy of tracking but also enhances the persistence of tracking. Even in the case of frequent frequency hopping or complex motion states of signal targets, stable tracking performance can be maintained. Therefore, the combined application of the hidden Markov model and the Kalman filtering algorithm greatly improves our continuous tracking ability of electromagnetic signal targets, providing strong support for electromagnetic spectrum management and security monitoring.

[0010] Preferably, in the sixth step: target signal parameter acquisition, there is also a fast Fourier transform algorithm, which performs spectrum analysis on the electromagnetic spectrum signal and identifies the frequency points of the electromagnetic signal target; In the process of target signal parameter acquisition, the application of the fast Fourier transform algorithm brings significant advantages. This algorithm can perform efficient spectrum analysis on the electromagnetic spectrum signal, convert the complex time-domain signal into an intuitive frequency-domain representation, and thus clearly show each frequency component in the signal. Through this conversion, we can quickly and accurately identify the frequency points of the electromagnetic signal target, which is a key step in signal analysis and processing. The efficiency and accuracy of the fast Fourier transform algorithm not only improve the efficiency of parameter acquisition but also ensure the reliability of the acquisition results. This enables us to quickly lock the target signal in a complex electromagnetic environment, providing a solid foundation for subsequent tasks such as signal decoding, interference analysis, or security monitoring. Therefore, the fast Fourier transform algorithm plays a crucial role in target signal parameter acquisition.

[0011] Preferably, in the seventh step: customizing the target signal type, there are also signals within a specific frequency range and signals with a specific coding method or signals with a specific modulation method; During the process of customizing the target signal type, incorporating signals within a specific frequency range, signals with a specific coding method, and signals with a specific modulation method greatly enriches our target recognition scope and capabilities. This custom setting allows us to precisely lock in and focus on those signals that are of specific significance to us according to actual needs. Whether it is frequency-based screening or the identification of coding and modulation methods, it provides us with more detailed and flexible signal classification and analysis means. This not only improves the pertinence and efficiency of signal processing but also enhances our signal capture and analysis capabilities in complex electromagnetic environments. Therefore, this function of customizing the target signal type provides strong support for us to handle diverse signal monitoring and analysis tasks and is an indispensable part of electromagnetic spectrum management and security monitoring.

[0012] Preferably, in step eight: the alarm and countermeasure processing also includes sound alarm, light alarm, or network alarm; In the alarm and countermeasure processing system, integrating multiple alarm methods such as sound alarm, light alarm, and network alarm greatly improves the emergency response ability and flexibility of the system. The sound alarm can quickly attract the attention of on-site personnel to ensure the discovery of potential threats in the first place; the light alarm provides a clear warning sign for on-site personnel through intuitive visual signals, while the network alarm breaks through geographical restrictions and can transmit alarm information to the remote monitoring center or relevant personnel in real time to achieve rapid information transmission and sharing. These diverse alarm methods complement each other and jointly constitute a comprehensive and efficient alarm system, which not only improves the monitoring and response speed to electromagnetic signal targets but also enhances the reliability and security of the system, providing strong guarantee for electromagnetic spectrum management and security monitoring.

[0013] Preferably, in step nine: the historical information playback also includes a time series analysis algorithm. The time series analysis algorithm processes and analyzes the historical data of electromagnetic spectrum signals and reveals the changing trends and laws of the electromagnetic environment. The algorithm optimization and update adjust and improve the algorithm parameters, model structure, or training strategy according to usage conditions and user feedback to improve the performance and effect of the algorithm; In the historical information playback of Step Nine, the addition of the time series analysis algorithm provides a powerful tool for the processing and analysis of historical data of electromagnetic spectrum signals. This algorithm can deeply explore the laws and trends of electromagnetic spectrum signals changing over time, helping us better understand the changing characteristics of the electromagnetic environment. By revealing these hidden change patterns, we can provide a favorable basis for subsequent electromagnetic spectrum management and security monitoring. At the same time, the algorithm optimization and update mechanism flexibly adjusts algorithm parameters, model structures, or training strategies according to actual usage and user feedback to ensure that the algorithm always remains in the best state. This continuous improvement and optimization process not only improves the performance and effect of the algorithm but also enhances the adaptability and stability of the system, laying a solid foundation for the long-term development of electromagnetic spectrum identification and tracking algorithms.

[0014] In summary, compared with the prior art, the present invention provides a transform-based electromagnetic spectrum identification and tracking algorithm, which has the following beneficial effects: This invention can accurately distinguish and identify various intertwined signals in a complex electromagnetic environment. Even when the signal characteristics are not obvious or there is mutual interference, it can maintain a high recognition accuracy. At the same time, for weak signals at a long distance, the detection sensitivity of the algorithm is significantly improved, effectively solving the problem that traditional algorithms are difficult to capture weak signals. Secondly, for frequency-hopping signals, the algorithm realizes the precise tracking and identification of frequency-hopping signals through the hidden Markov model and the Kalman filter algorithm, significantly reducing the false alarm rate and the missed alarm rate. In addition, the algorithm adopts a band-pass filter and a wavelet transform method in the preprocessing stage, effectively improving the signal quality and providing more accurate data support for subsequent identification. The multi-layer self-attention mechanism of the Transformer model and the Adam optimization algorithm further enhance the feature extraction ability and training efficiency of the model. Finally, the algorithm also has a high degree of flexibility and scalability, supports customizing the types of target signals to meet diverse needs. At the same time, by revealing the changing trends of the electromagnetic environment through the time series analysis algorithm, it provides strong support for subsequent detection and analysis. The algorithm can also be continuously optimized and updated according to usage and user feedback to ensure long-term high performance and adaptability, thus solving the problems raised in the background technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a schematic diagram of the steps of the transform-based electromagnetic spectrum identification and tracking algorithm of this invention. DETAILED DESCRIPTION OF THE INVENTION

[0016] The present invention provides a technical solution, a transform-based electromagnetic spectrum identification and tracking algorithm. Please refer to Figure 1 , including the following steps: Step One: Electromagnetic spectrum signal acquisition; Collect the complete electromagnetic spectrum signals in the surrounding environment in real time through an electromagnetic acquisition device to obtain electromagnetic spectrum data, providing basic data support for subsequent target recognition and tracking; Step 2: Preprocessing of electromagnetic spectrum data; Convert the electromagnetic spectrum data collected in Step 1 into a single-channel image format to obtain an electromagnetic spectrum image, and process the electromagnetic spectrum image through an image detection algorithm. The image detection algorithm includes signal filtering, denoising, and normalization preprocessing; Step 3: Construction of the Transformer model; Construct a target recognition model based on Transformer. The target recognition model adopts deep learning technology with the ability to automatically extract feature information in the electromagnetic spectrum image; Step 4: Target recognition; Use the target recognition model constructed in Step 3 to perform target recognition on the preprocessed electromagnetic spectrum image in Step 2. Extract and classify the features of the electromagnetic spectrum image through the target recognition model to identify electromagnetic signal targets that are different from the feature information of the conventional electromagnetic spectrum image; Step 5: Frequency hopping logic analysis; Analyze the frequency hopping characteristics of the electromagnetic signal targets obtained in Step 4, and combine the frequency hopping logic to achieve continuous tracking of the electromagnetic signal targets; Step 6: Acquisition of target signal parameters; After identifying the electromagnetic signal targets, automatically obtain the basic parameters of the frequency points, bandwidths, and coding methods of the electromagnetic signal targets. The basic parameters are used for signal analysis and processing; Step 7: Customize the target signal type; Dynamically add target information to be matched to the electromagnetic signal targets according to custom requirements; Step 8: Alarm and countermeasure processing; Automatically execute an alarm operation when an electromagnetic signal target is detected, and jointly use countermeasure means to perform automatic countermeasure processing on the target. The countermeasure means include electromagnetic interference; Step 9: Historical information playback; Record the historical data of the electromagnetic spectrum signals, detect and analyze the historical data of the electromagnetic spectrum signals to obtain the changing trend of the electromagnetic environment of the electromagnetic spectrum signals, and use the changing trend of the electromagnetic environment as a reference for subsequent detection and analysis; Step 10: Algorithm optimization and update; Continuously optimize and update the methods in Step 1 to Step 9 in real time according to the usage situation and user feedback; By collecting and preprocessing electromagnetic spectrum data in real time, the electromagnetic information of the surrounding environment can be accurately obtained, providing a solid foundation for subsequent target recognition and tracking. Using the Transformer model for target recognition not only improves the recognition accuracy but also enhances the generalization ability of the model. The frequency hopping logic analysis realizes the continuous tracking of electromagnetic signal targets, improving the real-time performance and effectiveness of monitoring. At the same time, the functions of automatically obtaining target signal parameters and customizing target signal types meet diverse requirements and improve the flexibility of signal processing. The alarm and countermeasure processing mechanism ensures the timely discovery and response to potential threats. The historical information playback function helps analyze the changing trend of the electromagnetic environment and provides a reference for future decision-making. Finally, the optimization and update of the algorithm ensure the continuous optimization and improvement of the entire process, adapting to the ever-changing electromagnetic environment.

[0017] Please refer to Figure 1 , Step 1: The electromagnetic spectrum signal acquisition also includes a multi-band antenna and a high-speed data acquisition card. The multi-band antenna covers the electromagnetic spectrum range, and the high-speed data acquisition card acquires the electromagnetic spectrum signal. The combination of a multi-band antenna and a high-speed data acquisition card brings significant benefits. The multi-band antenna can cover a wide range of the electromagnetic spectrum, ensuring that electromagnetic signals in different frequency bands can be captured, providing a comprehensive and rich data basis for subsequent analysis and processing. At the same time, the high-speed data acquisition card has high-speed and high-precision acquisition capabilities, capable of acquiring electromagnetic spectrum signals in real time and accurately, avoiding signal loss and distortion. The combined use not only improves the efficiency and accuracy of signal acquisition but also enhances the flexibility and adaptability of the system, enabling the electromagnetic spectrum signal acquisition system to better cope with complex and changing electromagnetic environments. Therefore, the application of the multi-band antenna and the high-speed data acquisition card provides strong support for the processing and analysis of electromagnetic spectrum signals and is an indispensable important part of the electromagnetic spectrum monitoring and recognition system.

[0018] Please refer to Figure 1 , Step 2: The electromagnetic spectrum data preprocessing also includes a band-pass filter. The passband range of the band-pass filter is dynamically adjusted according to the target electromagnetic spectrum range, and noise and interference signals are filtered out. Wavelet transform method is used for denoising, and the electromagnetic spectrum image is decomposed and reconstructed at multiple levels to remove the noise in the image. The normalization preprocessing normalizes the pixel values of the electromagnetic spectrum image to the interval [0,1] to improve the accuracy of subsequent target recognition. In the process of electromagnetic spectrum data preprocessing, introducing steps such as band-pass filtering, wavelet transform denoising, and normalization preprocessing brings many benefits. The band-pass filter can be dynamically adjusted according to its passband range, effectively filtering out noise and interference signals outside the target electromagnetic spectrum range, improving the purity of the data. The wavelet transform denoising method accurately removes noise in the electromagnetic spectrum image through multi-level decomposition and reconstruction of the image, retaining more useful information. The normalization preprocessing unifies the pixel values of the electromagnetic spectrum image to the [0,1] interval, eliminating the dimensional difference between different images and improving the accuracy and stability of subsequent target recognition. These preprocessing steps together improve the quality of electromagnetic spectrum data, providing a more reliable data basis for subsequent target recognition, tracking, and analysis.

[0019] Please refer to Figure 1 Step 3: The construction of the Transformer model also includes a multi-layer self-attention mechanism. Each layer of the multi-layer self-attention mechanism includes multiple heads. Each head independently calculates the attention weights, and the outputs of multiple heads are concatenated and linearly transformed to improve the model's ability to extract feature information from electromagnetic spectrum images. During the training process of the Transformer model, the Adam optimization algorithm and the cross-entropy loss function are used to minimize the difference between the predicted labels and the actual labels. In the construction of the Transformer model, the adoption of a multi-layer self-attention mechanism significantly improves the model's ability to process electromagnetic spectrum images. Each layer of the self-attention mechanism contains multiple heads, and each head independently calculates the attention weights. This design enables the model to more carefully capture different features in the image. By concatenating the outputs of multiple heads and performing a linear transformation, the model can integrate information from various aspects and improve its ability to extract feature information from electromagnetic spectrum images. In addition, during the model training process, the Adam optimization algorithm can accelerate convergence and optimize the model parameters. At the same time, the cross-entropy loss function is used to minimize the difference between the predicted labels and the actual labels, ensuring the prediction accuracy of the model. These designs together improve the performance of the Transformer model in the task of electromagnetic spectrum image target recognition, providing strong support for subsequent electromagnetic signal analysis and processing.

[0020] Please refer to Figure 1 Step 4: The target recognition also includes a non-maximum suppression algorithm. The non-maximum suppression algorithm performs the screening work of removing duplicate and redundant detection boxes for the target detection boxes, improving the accuracy and efficiency of target recognition. At the same time, the soft non-maximum suppression algorithm is used to attenuate the scores of the detection boxes to handle the overlapping problem between detection boxes. During the target recognition process, the introduction of the non-maximum suppression algorithm significantly improves the accuracy and efficiency of recognition. This algorithm screens the target detection boxes, effectively removing duplicate and redundant detection boxes, avoiding misjudgment and missed detection caused by the overlap of multiple detection boxes. At the same time, the soft non-maximum suppression algorithm is used to attenuate the scores of the detection boxes, cleverly solving the problem of overlap between detection boxes. This processing method not only considers the confidence of the detection boxes but also takes into account the relationship between adjacent detection boxes, making the finally retained detection boxes more accurate and reliable. Therefore, the application of the non-maximum suppression algorithm and its soft variant not only improves the accuracy of target recognition but also optimizes the recognition process, making the entire target recognition system more efficient and stable, laying a solid foundation for subsequent electromagnetic signal tracking and analysis.

[0021] Please refer to Figure 1 , Step Five: The frequency hopping logic analysis also includes a hidden Markov model. The hidden Markov model models and analyzes the frequency hopping characteristics of the electromagnetic signal target, predicts the frequency hopping pattern and frequency hopping points of the electromagnetic signal target, and combines the Kalman filtering algorithm to estimate and predict the motion state of the electromagnetic signal target to improve the continuous tracking ability of the electromagnetic signal target; In the frequency hopping logic analysis, the incorporation of the hidden Markov model significantly enhances our ability to understand the frequency hopping characteristics of the electromagnetic signal target. This model can accurately model and analyze the frequency hopping pattern and frequency hopping points of the electromagnetic signal target, thereby predicting the frequency hopping behavior of the target. This predictive ability is crucial for grasping the dynamic changes of the electromagnetic signal target. At the same time, combined with the Kalman filtering algorithm, we can estimate and predict the motion state of the electromagnetic signal target in real time. This combined method not only improves the accuracy of tracking but also enhances the persistence of tracking. Even in the case of frequent frequency hopping or complex motion states of the signal target, it can maintain stable tracking performance. Therefore, the combined application of the hidden Markov model and the Kalman filtering algorithm greatly improves our continuous tracking ability of the electromagnetic signal target, providing strong support for electromagnetic spectrum management and security monitoring.

[0022] Please refer to Figure 1 , Step Six: The target signal parameter acquisition also includes the fast Fourier transform algorithm. The fast Fourier transform algorithm performs spectral analysis on the electromagnetic spectrum signal and identifies the frequency points of the electromagnetic signal target; In the process of obtaining target signal parameters, the application of the Fast Fourier Transform (FFT) algorithm brings significant advantages. This algorithm can perform efficient spectral analysis on electromagnetic spectrum signals, converting complex time-domain signals into intuitive frequency-domain representations, thereby clearly showing each frequency component in the signal. Through this transformation, we can quickly and accurately identify the frequency points of electromagnetic signal targets, which is a crucial step in signal analysis and processing. The efficiency and accuracy of the FFT algorithm not only improve the efficiency of parameter acquisition but also ensure the reliability of the acquisition results. This enables us to quickly lock in on target signals in a complex electromagnetic environment, providing a solid foundation for subsequent tasks such as signal decoding, interference analysis, or security monitoring. Therefore, the FFT algorithm plays a vital role in obtaining target signal parameters.

[0023] Please refer to Figure 1 , Step Seven: The custom target signal types also include signals within a specific frequency range, signals with a specific coding method, or signals with a specific modulation method. In the process of customizing target signal types, incorporating signals within a specific frequency range, signals with a specific coding method, and signals with a specific modulation method greatly enriches our target recognition scope and capabilities. This custom setting allows us to precisely lock in and focus on those signals that are of specific significance to us according to actual needs. Whether it is frequency-based screening or the identification of coding and modulation methods, it provides us with more detailed and flexible signal classification and analysis means. This not only improves the pertinence and efficiency of signal processing but also enhances our signal capture and analysis capabilities in a complex electromagnetic environment. Therefore, this function of customizing target signal types provides strong support for us to handle diverse signal monitoring and analysis tasks and is an indispensable part of electromagnetic spectrum management and security monitoring.

[0024] Please refer to Figure 1 , Step Eight: The alarm and countermeasure processing also includes sound alarms, light alarms, or network alarms. In the alarm and countermeasure processing system, integrating multiple alarm methods such as sound alarms, light alarms, and network alarms greatly enhances the system's emergency response capabilities and flexibility. Sound alarms can quickly attract the attention of on-site personnel to ensure the discovery of potential threats in a timely manner; light alarms provide clear warning signs for on-site personnel through intuitive visual signals, while network alarms break through geographical limitations and can transmit alarm information to remote monitoring centers or relevant personnel in real time to achieve rapid information transmission and sharing. These diverse alarm methods complement each other and jointly form a comprehensive and efficient alarm system, which not only improves the monitoring and response speed to electromagnetic signal targets but also enhances the reliability and security of the system, providing strong guarantees for electromagnetic spectrum management and security monitoring.

[0025] Please refer to Figure 1 Step 9: The historical information playback also includes a time series analysis algorithm. The time series analysis algorithm processes and analyzes the historical data of electromagnetic spectrum signals, and reveals the changing trends and patterns of the electromagnetic environment. Algorithm optimization and update adjust and improve algorithm parameters, model structures, or training strategies according to usage scenarios and user feedback to enhance the performance and effectiveness of the algorithm; In the historical information playback of Step 9, the addition of the time series analysis algorithm provides a powerful tool for processing and analyzing the historical data of electromagnetic spectrum signals. This algorithm can deeply explore the laws and trends of electromagnetic spectrum signals over time, helping us better understand the changing characteristics of the electromagnetic environment. By revealing these hidden changing patterns, we can provide a favorable basis for subsequent electromagnetic spectrum management and security monitoring. At the same time, the algorithm optimization and update mechanism flexibly adjusts algorithm parameters, model structures, or training strategies according to actual usage scenarios and user feedback to ensure that the algorithm always maintains its optimal state. This continuous improvement and optimization process not only enhances the performance and effectiveness of the algorithm but also strengthens the adaptability and stability of the system, laying a solid foundation for the long-term development of electromagnetic spectrum identification and tracking algorithms.

[0026] It should be noted that in this document, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising", or any other variation thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0027] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made therein without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A transform-based electromagnetic spectrum recognition and tracking algorithm, characterized in that: The steps include: Step 1: Electromagnetic spectrum signal acquisition; Through electromagnetic acquisition equipment, complete electromagnetic spectrum signals in the surrounding environment are collected in real time to obtain electromagnetic spectrum data, providing basic data support for subsequent target identification and tracking; Step 2: electromagnetic spectrum data preprocessing; Converting the electromagnetic spectrum data collected in step 1 into a single-channel image format to obtain an electromagnetic spectrum image, and processing the electromagnetic spectrum image through an image detection algorithm, wherein the image detection algorithm includes signal filtering, denoising and normalization preprocessing; Step 3: Transformer model construction; Constructing a Transformer-based target recognition model, which uses deep learning technology capable of automatically extracting feature information from electromagnetic spectrum images; Step 4: Target identification; The target recognition model constructed in step three is used to perform target recognition on the electromagnetic spectrum image preprocessed in step two, and the features of the electromagnetic spectrum image are extracted and classified by the target recognition model to identify electromagnetic signal targets that are different from the feature information of conventional electromagnetic spectrum images; Step 5: Frequency hopping logic analysis; Analyze the frequency hopping characteristics of the electromagnetic signal target obtained in step 4, and combine the frequency hopping logic to achieve continuous tracking of the electromagnetic signal target; Step 6: Obtain target signal parameters; After identifying the electromagnetic signal target, automatically obtaining the basic parameters of the frequency, bandwidth and encoding method of the electromagnetic signal target, the basic parameters are used for signal analysis and processing; Step 7: Customize the target signal type; Dynamically add target information to be matched in the electromagnetic signal target according to custom requirements; Step 8: Alarm and countermeasure processing; Automatically perform an alarm operation when an electromagnetic signal target is found, and automatically counter the target with countermeasures, wherein the countermeasures include electromagnetic interference; Step 9: Playback of historical information; Record the historical data of electromagnetic spectrum signals, detect and analyze the historical data of electromagnetic spectrum signals, obtain the electromagnetic environment change trend of electromagnetic spectrum signals, and use the electromagnetic environment change trend as a reference for subsequent detection and analysis; Step 10: Algorithm optimization and update; The methods of steps one to nine are continuously optimized and updated in real time based on usage and user feedback.

2. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step 1: electromagnetic spectrum signal acquisition also includes a multi-band antenna and a high-speed data acquisition card, the multi-band antenna covers the electromagnetic spectrum range, and the high-speed data acquisition card acquires electromagnetic spectrum signals.

3. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step 2: the electromagnetic spectrum data preprocessing also includes a bandpass filter, the passband range of the bandpass filter is dynamically adjusted according to the target electromagnetic spectrum range, and noise and interference signals are filtered out.

4. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step three: the construction of the Transformer model also includes a multi-layer self-attention mechanism, each layer of the multi-layer self-attention mechanism includes multiple heads, each head independently calculates the attention weight, and the outputs of the multiple heads are spliced ​​and linearly transformed; the Adam optimization algorithm and the cross entropy loss function are used in the training process of the Transformer model.

5. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step 4: target recognition also includes a non-maximum suppression algorithm, which screens the target detection frame to remove repeated and redundant detection frames; at the same time, a soft non-maximum suppression algorithm is used to attenuate the score of the detection frame.

6. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step five: the frequency hopping logic analysis also includes a hidden Markov model, which models and analyzes the frequency hopping characteristics of the electromagnetic signal target, predicts the frequency hopping mode and frequency hopping point of the electromagnetic signal target, and estimates and predicts the motion state of the electromagnetic signal target in combination with the Kalman filter algorithm.

7. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step six: the acquisition of target signal parameters also includes a fast Fourier transform algorithm, which performs spectrum analysis on the electromagnetic spectrum signal and identifies the frequency point of the electromagnetic signal target.

8. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step seven: the customized target signal type also includes signals within a specific frequency range and signals in a specific coding method or signals in a specific modulation method.

9. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step eight: the alarm and countermeasure processing also includes sound alarm, light alarm or network alarm.

10. The transform-based electromagnetic spectrum identification and tracking algorithm according to claim 1, characterized in that: The step nine: the historical information playback also includes a time series analysis algorithm, which processes and analyzes the historical data of the electromagnetic spectrum signal and reveals the changing trends and laws of the electromagnetic environment.

Images