Ground-space time-frequency electromagnetic data real-time processing and intelligent analysis system

Abstract

The invention discloses a ground-air time-frequency electromagnetic data real-time processing and intelligent analysis system, relates to the technical field of information processing, and solves the problems that a non-stationary broadband signal is difficult to decompose into a plurality of sub-bands, more refined time-frequency analysis is inconvenient to carry out on an electromagnetic signal, and the analysis accuracy is high. And data points obviously different from the normal mode are difficult to effectively identify. The signal is analyzed under different time and frequency resolutions through wavelet packet transformation, so that the non-stationary broadband signal can be conveniently decomposed into a plurality of sub-bands; and in combination with wavelet packet transformation and STFT, features of each frequency band are effectively extracted, and a high-quality information basis is provided for subsequent data analysis. The time-frequency analysis result can be displayed through a time-frequency graph and other visual modes, so that complex signal features are more visual, and subsequent data interpretation and decision support are facilitated. Similar data points are divided into one group through a clustering algorithm, so that unsupervised classification is realized, manual intervention is reduced, and classification efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of information processing, and specifically relates to a real-time processing and intelligent analysis system for ground-air-time-frequency electromagnetic data. Background Art

[0002] With the rapid development of technologies such as wireless communication, radar detection, Internet of Things (IoT), and intelligent transportation, the generation and application of ground-air-time-frequency electromagnetic data are increasing day by day. These electromagnetic signals cover a wide frequency range and involve multiple application fields. Traditional data processing methods often fail to meet the requirements of real-time and accuracy. Therefore, there is an urgent need for an efficient system to process and analyze these complex electromagnetic data. The real-time processing and intelligent analysis system for ground-air-time-frequency electromagnetic data realizes the efficient processing and in-depth analysis of complex electromagnetic signals by combining cloud computing technology, providing strong data support and decision-making basis for various industries.

[0003] Most of the current electromagnetic data real-time processing and analysis systems perform unified analysis on signals, making it difficult to decompose non-stationary broadband signals into multiple sub-bands, which is not convenient for more refined time-frequency analysis of electromagnetic signals, and it is difficult to effectively identify data points significantly different from the normal mode and timely discover potential problems. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; for this purpose, the present invention proposes a real-time processing and intelligent analysis system for ground-air-time-frequency electromagnetic data, which is used to solve the technical problems that it is difficult to decompose non-stationary broadband signals into multiple sub-bands, not convenient for more refined time-frequency analysis of electromagnetic signals, and difficult to effectively identify data points significantly different from the normal mode.

[0005] To solve the above problems, the first aspect of the present invention provides a real-time processing and intelligent analysis system for ground-air-time-frequency electromagnetic data, including: A multi-sensor module, which is used to set sensor nodes in the ground and air environments of the target area, and deploy electromagnetic signal sensors in the ground environment, aviation environment, and semi-aviation environment at the sensor nodes; A front-end data processing module, which is used to divide the ground and air environments of the target area into different detection blocks, set edge computing nodes in each detection block, and perform preliminary electromagnetic signal processing and feature extraction through the edge computing nodes; A data planning module, which evaluates the levels of each detection block according to the feature extraction of electromagnetic signals by the edge computing nodes, determines the criticality of the detection signals according to the levels of the detection blocks, and determines whether to send the data to the cloud data center or retain it at the edge node according to the criticality of the detection signals; The time-frequency analysis module is used to decompose broadband signals into multiple sub-bands by means of wavelet packet transform in the cloud data center, perform time-frequency analysis on the electromagnetic signals of each sub-band through the short-time Fourier transform (STFT) method, and extract the characteristics of each frequency sub-band; The prediction analysis module is used to receive key data from edge nodes through the established cloud data center, and classify, predict, and detect anomalies in the key data through the K-means clustering algorithm and deep learning algorithm.

[0006] As a further solution of the present invention: the front-end data processing module processes and extracts features of preliminary electromagnetic signals through edge computing nodes, including the following steps: Preprocess the electromagnetic signals in the ground environment, aviation environment, and semi-aviation environment, and remove background noise and interference signals through a band-pass filter; Extract features of the electromagnetic signals, including time-domain feature extraction and frequency-domain feature extraction; Among them, various electromagnetic signals are respectively extracted for time-domain feature extraction, including instantaneous amplitude, mean, and variance data; Perform frequency-domain feature extraction on various electromagnetic signals respectively. Through the fast Fourier transform (FFT), the time-domain signal is converted into the frequency domain. Calculate the squared amplitude of the frequency components of various electromagnetic signals through spectrum analysis to obtain the power spectrum, and divide the power spectrum by the total duration of the sampling time interval to obtain the energy distribution of various electromagnetic signals.

[0007] Specifically, in this embodiment, the basic distribution characteristics of the signal can be described through features such as instantaneous amplitude, mean, and variance, which is convenient for identifying different types of signals; by calculating the power spectral density, the energy distribution of the signal in different frequency ranges can be understood, which helps to discover potential frequency features.

[0008] As a further solution of the present invention: the data planning module evaluates the levels of each detection block according to the feature extraction of electromagnetic signals by edge computing nodes, including the following steps: Divide the electromagnetic signal data of the detected ground environment, aviation environment, and semi-aviation environment into several detection time intervals, obtain the electromagnetic signals of each environment, the time-domain features in each detection time interval, including instantaneous amplitude, mean, and variance data; and the frequency-domain features of the electromagnetic signals, including the energy distribution of the electromagnetic signals; Evaluate the levels of each detection block through the following formula: Among them, K is the hierarchical evaluation value of each detection block, μei is the mean value of the energy distribution of the i-th electromagnetic signal in each detection block within each detection time interval, Sei is the variance of the energy distribution of the i-th electromagnetic signal in each detection block within each detection time interval, μai is the mean value of the peak values of the i-th electromagnetic signal in each detection block within each detection time interval, Sai is the variance of the peak values of the i-th electromagnetic signal in each detection block within each detection time interval, μpi is the mean data of the time-domain characteristics of the i-th electromagnetic signal in the detection block, and Spi is the variance data in the time-domain characteristics of the i-th electromagnetic signal in the detection block; Sort the hierarchical evaluation values of each detection block from high to low. Set the detection blocks corresponding to the top 20% of the hierarchical evaluation values as high-level detection blocks, set the detection blocks corresponding to the last 20% of the hierarchical evaluation values as low-level detection blocks, and set the remaining detection blocks as middle-level detection blocks.

[0009] As a further solution of the present invention: The data planning module determines the criticality of the detection signal according to the level of the detection block, and determines whether to send the data to the cloud data center or retain it at the edge node according to the criticality of the detection signal, including the following steps: Calculate the criticality coefficient of the detection signal through the following formula: Among them, α is the criticality coefficient of the signal, μmax is the maximum value of the mean value of the energy distribution of the electromagnetic signal in the detection block, and Smax is the maximum value of the variance data in the time-domain characteristics of the electromagnetic signal in the detection block; Judge the criticality of the detection signal according to the level of the detection block and the criticality coefficient of the electromagnetic signal in the detection block: For high-level detection blocks, set the detection signals in the top 20% of the criticality coefficients of the electromagnetic signals in the detection block as critical detection signals, and set the detection signals in the last 5% as non-critical detection signals; For middle-level detection blocks, set the detection signals in the top 15% of the criticality coefficients of the electromagnetic signals in the detection block as critical detection signals, and set the detection signals in the last 10% as non-critical detection signals; For low-level detection blocks, set the detection signals in the top 10% of the criticality coefficients of the electromagnetic signals in the detection block as critical detection signals, and set the detection signals in the last 15% as non-critical detection signals; Retain the non-critical detection signal data in the detection block at the edge node, send the critical detection signal data in the detection block to the cloud data center in real time, and send the other signal data in the detection block to the cloud data center at preset time intervals.

[0010] As a further solution of the present invention: the time-frequency analysis module decomposes the wideband signal into multiple sub-bands by means of wavelet packet transform, performs time-frequency analysis on the electromagnetic signals of each sub-band by means of STFT short-time Fourier transform, and extracts the characteristics of each frequency segment sub-band, including the following steps: The time-frequency analysis module constructs a wavelet packet tree according to the Haar wavelet basis by means of wavelet packet transform; Extract the decomposed sub-bands from the wavelet packet tree; Select the window function and window length by means of STFT short-time Fourier transform; in this embodiment, the Hamming window is selected as the window function, and the window length is 350 sample data.

[0011] Slide the selected window on the electromagnetic signals of each sub-band respectively, and perform Fourier transform at each window position; Obtain the time-frequency diagrams of the electromagnetic signals of each sub-band by performing Fourier transform on the signals within each window; Extract the characteristics of each frequency segment sub-band according to the time-frequency diagrams of the electromagnetic signals of each sub-band, including: amplitude spectrum, phase spectrum and energy spectrum; Among them, amplitude spectrum: calculate the amplitude of each frequency component; phase spectrum: calculate the phase of each frequency component; energy spectrum: calculate the energy of each frequency segment.

[0012] According to the energy spectrum of each frequency segment sub-band extracted, with the frequency component of the electromagnetic signal as the X-axis and the energy distribution on the frequency component of the electromagnetic signal as the Y-axis, draw the PSD power spectral density graph; Identify the main frequency components and their harmonics according to the power spectral density graph, and extract the main frequency, bandwidth and harmonic ratio data of the sub-band as the characteristics of each frequency segment sub-band.

[0013] Harmonics are usually integer multiples of the fundamental frequency and can be identified by observing the PSD graph. Feature extraction includes: extracting the main frequency, bandwidth and harmonic ratio.

[0014] As a further solution of the present invention: constructing a wavelet packet tree according to the Haar wavelet basis by means of wavelet packet transform includes the following steps: Perform a wavelet transform on the original signal to obtain the low-frequency part and the high-frequency part. Pass the low-frequency part through a low-pass filter to generate approximation coefficients, and pass the high-frequency part through a high-pass filter to generate detail coefficients; Perform DWT discrete wavelet transform through the Haar wavelet basis; Let the original signal be x(t). After performing a DWT discrete wavelet transform once, obtain the initial low-frequency coefficients and high-frequency coefficients; after performing a DWT discrete wavelet transform once, obtain: c0 = DWT(x(t)); where c0 is the initial low-frequency coefficient and d0 is the initial high-frequency coefficient.

[0015] Perform further wavelet transform on the low-frequency part. After the second DWT (Discrete Wavelet Transform) on the low-frequency part, secondary low-frequency coefficients and high-frequency coefficients are obtained; continue to perform DWT on the low-frequency part c0 to obtain finer low-frequency and high-frequency parts. After the second DWT, we get: c1, d1 = DWT(c0), where c1 and d1 are the secondary low-frequency coefficients and high-frequency coefficients.

[0016] Repeat this process to decompose each low-frequency subband until a predetermined number of layers is reached.

[0017] As a further solution of the present invention: The prediction analysis module classifies, predicts, and detects anomalies in key data through the K-means clustering algorithm and deep learning algorithm, including the following steps: Obtain the main frequency, bandwidth, and harmonic ratio data of the subbands extracted by the time-frequency analysis module; Integrate all features into a data table, where each row represents a sample and each column represents a feature; Classify the key data for similarity through the K-means clustering algorithm and divide the feature data of the subbands into different similarity classifications; Obtain the features of each frequency subband extracted by the time-frequency analysis module, including: the amplitude spectrum, phase spectrum, and energy spectrum of the subbands, and construct anomaly detection and prediction models for different similarity classifications through deep learning algorithms; Detect and predict the anomaly status of the subbands through the anomaly detection and prediction models of different similarity classifications.

[0018] As a further solution of the present invention: Classify the key data for similarity through the K-means clustering algorithm and divide the feature data of the subbands into different similarity classifications, including the following steps: S1: Standardize the feature data in the data table; S2: Calculate the silhouette coefficients under different K values through the silhouette coefficient method and select the K value with the highest silhouette coefficient as the final choice; S3: Randomly select K initial cluster centers; S4: For each subband feature data in the data table, calculate the distance between it and all cluster centers and assign it to the cluster where the nearest cluster center is located; S5: According to the positions of all subband feature data in the data table in the current cluster, calculate the mean of all subband feature data in the current cluster and update the position of the mean of all subband feature data in the current cluster to the cluster center; S6: Repeat steps S4 and S5 until the change in the position of the most recently updated cluster center is less than the preset distance, and use the sub-band feature data in each data table within each cluster as the data in the same similarity classification.

[0019] As a further solution of the present invention: By means of a deep learning algorithm, anomaly detection and prediction models for different similarity classifications are respectively constructed, including the following steps: Obtain the historical data of the sub-band feature data of different similarity classifications, as well as the image data of the amplitude spectrum, phase spectrum, and energy spectrum of the corresponding sub-bands in the sub-band feature data of different similarity classifications; Divide the historical data into data at the time of anomaly occurrence, data in a preset time period before the anomaly, and normal data, and add data labels to the corresponding data according to the classification; Train a neural network model through the image data of the amplitude spectrum, phase spectrum, and energy spectrum with added labels of different similarity classifications to construct anomaly detection and prediction models for different similarity classifications.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention analyzes signals at different time and frequency resolutions through wavelet packet transform, which is convenient for decomposing non-stationary wideband signals into multiple sub-bands; through the short-time Fourier transform (STFT) method, it is convenient to obtain the time and frequency information of the signal simultaneously, making the understanding of signal characteristics more comprehensive, especially when dealing with dynamically changing signals. Combining wavelet packet transform with STFT can effectively extract the characteristics of each frequency band, providing a high-quality information basis for subsequent data analysis, classification, or recognition. The time-frequency analysis results can be displayed in a visual way such as a time-frequency diagram, making complex signal characteristics more intuitive and facilitating subsequent data interpretation and decision support.

[0021] The present invention classifies similar data points into a group through the K-means clustering algorithm, thereby achieving unsupervised classification, reducing manual intervention, and improving classification efficiency. Through the deep learning model, data points significantly different from the normal mode can be effectively identified to achieve real-time anomaly detection, which helps to timely discover potential problems. The deep learning algorithm can capture complex non-linear relationships in the data, improving the accuracy and robustness of the prediction model. By performing real-time analysis on key data, the system status can be monitored in a timely manner, and feedback can be provided according to the analysis results to optimize the system operation. Description of the Drawings

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0023] Figure 1 It is a schematic diagram of the system framework of the present invention. Specific embodiments

[0024] The following will clearly and completely describe the technical solutions of the present invention in combination with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0025] Please refer to Figure 1 , an embodiment of the first aspect of the present invention provides a real-time processing and intelligent analysis system for ground-air time-frequency electromagnetic data, including: A multi-sensor module, used to set sensor nodes in the ground and air environments of the target area, and deploy electromagnetic signal sensors in the ground environment, aviation environment, and semi-aviation environment at the sensor nodes; A front-end data processing module, used to divide the ground and air environments of the target area into different detection blocks, set edge computing nodes in each detection block, and perform preliminary electromagnetic signal processing and feature extraction through the edge computing nodes; A data planning module, which evaluates the levels of each detection block according to the feature extraction of electromagnetic signals by the edge computing nodes, determines the criticality of the detection signals according to the levels of the detection blocks, and determines whether to send the data to the cloud data center or retain it at the edge node according to the criticality of the detection signals; A time-frequency analysis module, used to decompose the broadband signal into multiple sub-bands by the wavelet packet transform method in the cloud data center, perform time-frequency analysis on the electromagnetic signals of each sub-band by the STFT short-time Fourier transform method, and extract the features of each frequency sub-band; A prediction analysis module, used to receive key data from the edge nodes through the established cloud data center, and perform classification, prediction, and anomaly detection on the key data through the K-means clustering algorithm and the deep learning algorithm.

[0026] Specifically, in this embodiment, the front-end data processing module divides the ground and aerial environments of the target area into different detection blocks, sets edge computing nodes in each detection block, and processes and extracts features of preliminary electromagnetic signals through the edge computing nodes. The edge computing nodes are close to the data sources, facilitating low-latency data processing and promptly responding to the detection data of different detection blocks. By processing data locally and only transmitting necessary information to the central server, the consumption of network bandwidth is reduced, and the data transmission cost is lowered. By dispersing computing tasks and storage tasks to multiple edge nodes, the computing load can be effectively shared, improving the overall performance of the system. The data planning module evaluates the levels of each detection block according to the feature extraction of electromagnetic signals by the edge computing nodes, determines the criticality of the detection signals based on the levels of the detection blocks, and determines whether to send the data to the cloud data center or retain it at the edge node according to the criticality of the detection signals. Through the evaluation of signal criticality, more intelligent data processing and decision-making are achieved, ensuring the timely transmission of important information while unimportant information can be processed locally. Only sending critical data to the cloud reduces unnecessary data transmission, thus saving network bandwidth and improving transmission efficiency. Through the screening and hierarchical management of data, computing resources can be effectively utilized, avoiding cloud overload and reducing storage costs at the same time.

[0027] Dynamically adjust the data transmission strategy according to the real-time evaluation results, enabling the system to better adapt to environmental changes and demand fluctuations. By focusing on critical signals, the quality of the data uploaded to the cloud can be improved, contributing to subsequent data analysis and decision support.

[0028] In the cloud data center, the time-frequency analysis module decomposes the broadband signal into multiple sub-bands through the wavelet packet transform method, performs time-frequency analysis on the electromagnetic signals of each sub-band through the STFT (Short-Time Fourier Transform) method, and extracts the features of each frequency sub-band. The wavelet packet transform can analyze signals at different time and frequency resolutions, facilitating the decomposition of non-stationary broadband signals into multiple sub-bands. Through the STFT method, it is convenient to obtain both the time and frequency information of the signal simultaneously, enabling a more comprehensive understanding of the signal features, especially when dealing with dynamically changing signals. Combining the wavelet packet transform with the STFT can effectively extract the features of each frequency band, providing a high-quality information basis for subsequent data analysis, classification, or recognition.

[0029] The time-frequency analysis results can be displayed in a visual way such as a time-frequency diagram, making the complex signal features more intuitive and facilitating subsequent data interpretation and decision support.

[0030] The prediction analysis module receives the critical data from the edge nodes through the established cloud data center, and classifies, predicts, and detects anomalies for the critical data through the K-means clustering algorithm and deep learning algorithm.

[0031] The cloud data center has powerful computing capabilities, which can quickly process and analyze a large amount of key data, improve data processing efficiency. By combining K-means clustering and deep learning, it can achieve more complex data analysis tasks, identify potential patterns and trends, and enhance intelligent decision-making capabilities.

[0032] Through the K-means clustering algorithm, similar data points are grouped together to achieve unsupervised classification, reduce manual intervention, and improve classification efficiency. Through the deep learning model, data points significantly different from the normal pattern can be effectively identified to achieve real-time anomaly detection, which helps to discover potential problems in a timely manner.

[0033] The deep learning algorithm can capture complex non-linear relationships in the data, improving the accuracy and robustness of the prediction model. By performing real-time analysis on key data, the system status can be monitored in a timely manner, and feedback can be provided according to the analysis results to optimize the system operation.

[0034] In one embodiment of the present invention, the front-end data processing module processes and extracts features of preliminary electromagnetic signals through edge computing nodes, including the following steps: Preprocess the electromagnetic signals in the ground environment, aviation environment, and semi-aviation environment, and remove background noise and interference signals through a band-pass filter; Extract features of the electromagnetic signals, including time-domain feature extraction and frequency-domain feature extraction; Among them, various electromagnetic signals are respectively subjected to time-domain feature extraction, including instantaneous amplitude, mean, and variance data; Various electromagnetic signals are respectively subjected to frequency-domain feature extraction. Through the FFT (Fast Fourier Transform), the time-domain signal is converted into the frequency domain. By calculating the squared amplitude of the frequency components of various electromagnetic signals through spectrum analysis, the power spectrum is obtained. Divide the power spectrum by the total duration of the sampling time interval to obtain the energy distribution of various electromagnetic signals.

[0035] Specifically, in this embodiment, the basic distribution characteristics of the signal can be described through features such as instantaneous amplitude, mean, and variance, which is convenient for identifying different types of signals; by calculating the power spectral density, the energy distribution of the signal in different frequency ranges can be understood, which helps to discover potential frequency features.

[0036] In one embodiment of the present invention, the data planning module evaluates the levels of each detection block according to the feature extraction of electromagnetic signals by edge computing nodes, including the following steps: Divide the detected electromagnetic signal data of the ground environment, aviation environment, and semi-aviation environment into several detection time intervals, obtain the electromagnetic signals of each environment, and the time-domain characteristics in each detection time interval, including instantaneous amplitude, mean, and variance data; as well as the frequency-domain characteristics of the electromagnetic signals, including the energy distribution of the electromagnetic signals. Evaluate the levels of each detection block through the following formula: where K is the level evaluation value of each detection block, μei is the mean of the energy distribution of the i-th electromagnetic signal in the detection block within each detection time interval, Sei is the variance of the energy distribution of the i-th electromagnetic signal in the detection block within each detection time interval, μai is the mean of the peak values of the i-th electromagnetic signal in the detection block within each detection time interval, Sai is the variance of the peak values of the i-th electromagnetic signal in the detection block within each detection time interval, μpi is the mean data of the time-domain characteristics of the i-th electromagnetic signal in the detection block, and Spi is the variance data in the time-domain characteristics of the i-th electromagnetic signal in the detection block. Sort the level evaluation values of each detection block from high to low, set the detection blocks corresponding to the top 20% of the level evaluation values as high-level detection blocks, set the detection blocks corresponding to the last 20% of the level evaluation values as low-level detection blocks, and set the remaining detection blocks as medium-level detection blocks.

[0037] Specifically, in this embodiment, the entire monitoring area is divided into multiple detection blocks, and the importance score of each block is calculated. Then, sort according to the scores. According to actual needs, the monitoring area is divided into several small blocks, and each block corresponds to a set of electromagnetic signal data. Apply the importance scoring model in the previous step to each block to obtain the importance score of each block. Sort all blocks according to the scores to determine which areas are the most critical and need to be prioritized for attention and processing.

[0038] In one embodiment of the present invention, the data planning module determines the criticality of the detection signal according to the level of the detection block, and determines whether to send the data to the cloud data center or retain it at the edge node according to the criticality of the detection signal, including the following steps: Calculate the criticality coefficient of the detection signal through the following formula: where α is the criticality coefficient of the signal, μmax is the maximum value of the mean of the energy distribution of the electromagnetic signal in the detection block, and Smax is the maximum value of the variance data in the time-domain characteristics of the electromagnetic signal in the detection block. Judge the criticality of the detection signal according to the level of the detection block and the criticality coefficient of the electromagnetic signal in the detection block: For high-level detection blocks, set the top 20% of the detection signals with key coefficients of the electromagnetic signals within the detection block as key detection signals, and set the bottom 5% of the detection signals as non-key detection signals; For mid-level detection blocks, set the top 15% of the detection signals with key coefficients of the electromagnetic signals within the detection block as key detection signals, and set the bottom 10% of the detection signals as non-key detection signals; For low-level detection blocks, set the top 10% of the detection signals with key coefficients of the electromagnetic signals within the detection block as key detection signals, and set the bottom 15% of the detection signals as non-key detection signals; Retain the non-key detection signal data within the detection block at the edge node, send the key detection signal data within the detection block to the cloud data center in real time, and send the other signal data within the detection block to the cloud data center every preset time period.

[0039] In one embodiment of the present invention, the time-frequency analysis module decomposes the broadband signal into multiple sub-bands by the wavelet packet transform method, performs time-frequency analysis on the electromagnetic signals of each sub-band by the STFT short-time Fourier transform method, and extracts the characteristics of each frequency segment sub-band, including the following steps: The time-frequency analysis module constructs a wavelet packet tree according to the Haar wavelet basis by the wavelet packet transform method; Extract the decomposed sub-bands from the wavelet packet tree; Select the window function and window length by the STFT short-time Fourier transform; in this embodiment, select the Hamming window as the window function and the window length is 350 sample data.

[0040] Slide the selected window on the electromagnetic signals of each sub-band respectively, and perform Fourier transform at each window position; Obtain the time-frequency diagrams of the electromagnetic signals of each sub-band by performing Fourier transform on the signals within each window; Extract the characteristics of each frequency segment sub-band according to the time-frequency diagrams of the electromagnetic signals of each sub-band, including: amplitude spectrum, phase spectrum and energy spectrum; Among them, amplitude spectrum: calculate the amplitude of each frequency component; phase spectrum: calculate the phase of each frequency component; energy spectrum: calculate the energy of each frequency segment.

[0041] According to the energy spectra of each frequency segment sub-band extracted, with the frequency components of the electromagnetic signal as the X-axis and the energy distribution on the frequency components of the electromagnetic signal as the Y-axis, draw the PSD power spectral density graph; Identify the main frequency components and their harmonics according to the power spectral density graph, and extract the main frequency, bandwidth and harmonic ratio data of the sub-band as the characteristics of each frequency segment sub-band.

[0042] Harmonics are usually integer multiples of the fundamental frequency and can be identified by observing the PSD graph. Feature extraction includes: extracting the main frequency, bandwidth, and harmonic ratio.

[0043] In one embodiment of the present invention, by using the wavelet packet transform method, a wavelet packet tree is constructed according to the Haar wavelet basis, including the following steps: Perform a wavelet transform on the original signal to obtain a low-frequency part and a high-frequency part. Pass the low-frequency part through a low-pass filter to generate approximation coefficients, and pass the high-frequency part through a high-pass filter to generate detail coefficients; Perform a DWT (Discrete Wavelet Transform) using the Haar wavelet basis; Let the original signal be x(t). After performing a DWT discrete wavelet transform once, initial low-frequency coefficients and high-frequency coefficients are obtained; after one DWT discrete wavelet transform, we get: c0 = DWT(x(t)); where c0 is the initial low-frequency coefficient and d0 is the initial high-frequency coefficient.

[0044] Perform a further wavelet transform on the low-frequency part. After performing a second DWT discrete wavelet transform on the low-frequency part, secondary low-frequency coefficients and high-frequency coefficients are obtained; continue to perform DWT on the low-frequency part c0 to obtain finer low-frequency and high-frequency parts. After the second DWT, we get: c1, d1 = DWT(c0), where c1 and d1 are the secondary low-frequency coefficients and high-frequency coefficients.

[0045] Repeat this process to decompose each low-frequency subband until a predetermined number of layers is reached.

[0046] Specifically, in this embodiment, each node of the wavelet packet tree constructed by the Haar wavelet basis represents a frequency band, and the depth of the tree determines the number of decomposition layers.

[0047] For each layer, pass the signal through a low-pass filter to generate approximation coefficients and through a high-pass filter to generate detail coefficients; After each layer, the same operations can be continued on the approximation coefficients and detail coefficients to obtain more subbands.

[0048] Through multi-level decomposition, several subbands are obtained, and each subband corresponds to signal components within a different frequency range.

[0049] Generally, after decomposing to the Nth layer, we will get subbands.

[0050] In one embodiment of the present invention, the prediction analysis module classifies, predicts, and performs anomaly detection on key data through the K-means clustering algorithm and the deep learning algorithm, including the following steps: Obtain the main frequency, bandwidth, and harmonic ratio data of the subbands extracted by the time-frequency analysis module; All features are integrated into a data table, where each row represents a sample and each column represents a feature; The key data is classified by similarity through the K-means clustering algorithm, and the feature data of the subbands is divided into different similarity classifications; Obtain the features of each frequency subband extracted by the time-frequency analysis module, including: the amplitude spectrum, phase spectrum, and energy spectrum of the subband, and respectively construct models for anomaly detection and prediction of different similarity classifications through deep learning algorithms; Detect and predict the abnormal conditions of the subbands through the anomaly detection and prediction models of different similarity classifications.

[0051] In one embodiment of the present invention, the key data is classified by similarity through the K-means clustering algorithm, and the feature data of the subbands is divided into different similarity classifications, including the following steps: S1: Standardize the feature data in the data table; S2: Calculate the silhouette coefficients under different K values through the silhouette coefficient method, and select the K value with the highest silhouette coefficient as the final choice; S3: Randomly select K initial cluster centers; S4: For each subband feature data in the data table, calculate the distance between it and all cluster centers, and assign it to the cluster where the nearest cluster center is located; S5: According to the positions of each subband feature data in all data tables in the current cluster, calculate the mean value of each subband feature data in all data tables in the current cluster, and update the position of the mean value of each subband feature data in all data tables in the current cluster to the cluster center; S6: Repeat steps S4 and S5 until the change in the position of the latest updated cluster center is less than the preset distance, and take each subband feature data in all data tables in each cluster as the data in the same similarity classification.

[0052] In one embodiment of the present invention, models for anomaly detection and prediction of different similarity classifications are respectively constructed through deep learning algorithms, including the following steps: Obtain the historical data of the subband feature data of different similarity classifications, as well as the image data of the amplitude spectrum, phase spectrum, and energy spectrum of the corresponding subbands in the subband feature data of different similarity classifications; Divide the historical data into data during anomaly occurrence, data in a preset time period before the anomaly, and normal data, and add data labels to the corresponding data according to the classification; Train the neural network model through the image data of the amplitude spectrum, phase spectrum, and energy spectrum with added labels of different similarity classifications, and construct models for anomaly detection and prediction of different similarity classifications.

[0053] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A real-time processing and intelligent analysis system for ground-to-air time-frequency electromagnetic data, characterized in that: include: A multi-sensor module is used to set up sensor nodes in the ground and air environments of the target area, and to deploy electromagnetic signal sensors for ground environments, aviation environments, and semi-aviation environments at the sensor nodes; The front-end data processing module is used to divide the ground and air environments of the target area into different detection blocks, set up edge computing nodes in each detection block, and perform preliminary electromagnetic signal processing and feature extraction through the edge computing nodes; The data planning module extracts the features of electromagnetic signals from edge computing nodes, evaluates the levels of each detection block, determines the criticality of the detection signal based on the level of the detection block, and determines whether to send the data to the cloud data center or keep it in the edge node based on the criticality of the detection signal; The time-frequency analysis module is used to decompose the broadband signal into multiple sub-bands by using the wavelet packet transform method in the cloud data center, perform time-frequency analysis on the electromagnetic signal of each sub-band by using the STFT short-time Fourier transform method, and extract the characteristics of each frequency sub-band; The predictive analysis module is used to receive key data from edge nodes through the established cloud data center, and classify, predict and detect anomalies of the key data through K-means clustering algorithm and deep learning algorithm.

2. A system for real-time processing and intelligent analysis of ground-space time-frequency electromagnetic data according to claim 1, characterized in that: The front-end data processing module performs preliminary electromagnetic signal processing and feature extraction through the edge computing node, including the following steps: Pre-process the electromagnetic signals in the ground environment, aviation environment and semi-aviation environment, and remove the background noise and interference signals through the bandpass filter; Extract features of electromagnetic signals, including time domain feature extraction and frequency domain feature extraction; Among them, various electromagnetic signals are extracted for time domain feature extraction, including instantaneous amplitude, mean and variance data; Frequency domain features are extracted for various electromagnetic signals respectively. The time domain signals are converted into frequency domain through FFT fast Fourier transform. The amplitude squares of the frequency components of various electromagnetic signals are calculated through spectrum analysis to obtain the power spectrum. The power spectrum is divided by the total duration of the sampling time interval to obtain the energy distribution of various electromagnetic signals.

3. The real-time processing and intelligent analysis system for ground-space time-frequency electromagnetic data according to claim 1 is characterized in that: The data planning module evaluates the level of each detection block according to the feature extraction of the electromagnetic signal by the edge computing node, including the following steps: The electromagnetic signal data of the detected ground environment, aviation environment and semi-aviation environment are divided into several detection time intervals, and the electromagnetic signal of each environment is obtained, and the time domain characteristics in each detection time interval, including instantaneous amplitude, mean and variance data; and the frequency domain characteristics of the electromagnetic signal, including the energy distribution of the electromagnetic signal; The level of each detection block is evaluated by the following formula: Among them, K is the hierarchical evaluation value of each detection block, μei is the mean value of the energy distribution of the ith electromagnetic signal in the detection block in each detection time interval, Sei is the variance of the energy distribution of the ith electromagnetic signal in the detection block in each detection time interval, μai is the mean value of the peak value of the ith electromagnetic signal in the detection block in each detection time interval, Sai is the variance of the peak value of the ith electromagnetic signal in the detection block in each detection time interval, μpi is the mean data of the time domain characteristics of the ith electromagnetic signal in the detection block, and Spi is the variance data in the time domain characteristics of the ith electromagnetic signal in the detection block; The hierarchical evaluation values ​​of each detection block are sorted from high to low, and the detection blocks corresponding to the first 20% of the hierarchical evaluation values ​​are set as high-level detection blocks, the detection blocks corresponding to the last 20% of the hierarchical evaluation values ​​are set as low-level detection blocks, and the remaining detection blocks are set as medium-level detection blocks.

4. The real-time processing and intelligent analysis system for ground-space time-frequency electromagnetic data according to claim 3 is characterized in that: The data planning module determines the criticality of the detection signal according to the level of the detection block, and determines whether to send the data to the cloud data center or keep it in the edge node according to the criticality of the detection signal, including the following steps: The criticality coefficient of the detection signal is calculated by the following formula: Among them, α is the critical coefficient of the signal, μmax is the maximum value of the energy distribution mean of the electromagnetic signal of the detection block, and Smax is the maximum value of the variance data in the time domain characteristics of the electromagnetic signal of the detection block; According to the level of the detection block and the criticality coefficient of the electromagnetic signal in the detection block, the criticality of the detection signal is judged: For high-level detection blocks, the detection signals of the first 20% of the critical coefficients of the electromagnetic signals in the detection blocks are set as critical detection signals, and the detection signals of the last 5% are set as non-critical detection signals; For the middle-level detection block, the detection signals of the first 15% of the critical coefficients of the electromagnetic signals in the detection block are set as critical detection signals, and the detection signals of the last 10% are set as non-critical detection signals; For low-level detection blocks, the detection signals of the first 10% of the critical coefficients of the electromagnetic signals in the detection blocks are set as critical detection signals, and the detection signals of the last 15% are set as non-critical detection signals; The non-critical detection signal data in the detection block is retained in the edge node, the critical detection signal data in the detection block is sent to the cloud data center in real time, and other signal data in the detection block is sent to the cloud data center at preset time intervals.

5. The real-time processing and intelligent analysis system for ground-space time-frequency electromagnetic data according to claim 1 is characterized in that: The time-frequency analysis module decomposes the broadband signal into multiple sub-bands by wavelet packet transform, performs time-frequency analysis on each sub-band electromagnetic signal by STFT short-time Fourier transform, and extracts the characteristics of each frequency sub-band, including the following steps: The time-frequency analysis module constructs a wavelet packet tree based on the Haar wavelet basis using the wavelet packet transform method; Extract decomposed subbands from the wavelet packet tree; Select the window function and window length through STFT short-time Fourier transform; Slide the selected window on each sub-band electromagnetic signal respectively, and perform Fourier transform at each window position; By performing Fourier transform on the signal in each window, the time-frequency diagram of the electromagnetic signal of each sub-band is obtained; According to the time-frequency diagram of each sub-band electromagnetic signal, the characteristics of each frequency sub-band are extracted, including: amplitude spectrum, phase spectrum and energy spectrum; According to the energy spectrum of each frequency sub-band extracted, the PSD power spectrum density graph is drawn with the frequency component of the electromagnetic signal as the X-axis and the energy distribution on the frequency component of the electromagnetic signal as the Y-axis; The main frequency components and their harmonics are identified based on the power spectrum density graph, and the main frequency, bandwidth and harmonic ratio data of the sub-band are extracted as the characteristics of each frequency sub-band.

6. A system for real-time processing and intelligent analysis of ground-space time-frequency electromagnetic data according to claim 5, characterized in that: By using the wavelet packet transform method, a wavelet packet tree is constructed based on the Haar wavelet basis, including the following steps: Perform a wavelet transform on the original signal to obtain the low-frequency part and the high-frequency part. The low-frequency part is passed through a low-pass filter to generate an approximate coefficient, and the high-frequency part is passed through a high-pass filter to generate a detail coefficient. After performing a DWT discrete wavelet transform on the original signal, the initial low-frequency coefficients and high-frequency coefficients are obtained; The low-frequency part is further subjected to wavelet transformation, and the low-frequency part is subjected to a second DWT discrete wavelet transformation to obtain secondary low-frequency coefficients and high-frequency coefficients; This process is repeated to decompose each low-frequency subband until the predetermined number of layers is reached.

7. The real-time processing and intelligent analysis system for ground-space time-frequency electromagnetic data according to claim 1 is characterized in that: The prediction and analysis module classifies, predicts and detects anomalies of key data through K-means clustering algorithm and deep learning algorithm, including the following steps: Obtain the main frequency, bandwidth and harmonic ratio data of the sub-band extracted by the time-frequency analysis module; All features are integrated into a data table, where each row represents a sample and each column represents a feature; The key data is classified by similarity through the K-means clustering algorithm, and the feature data of the sub-band is divided into different similarity categories; Obtain the features of each frequency sub-band extracted by the time-frequency analysis module, including the amplitude spectrum, phase spectrum, and energy spectrum of the sub-band. Use deep learning algorithms to build models for anomaly detection and prediction of different similarity classifications. The abnormal conditions of the sub-band are detected and predicted through the anomaly detection and prediction models of different similarity classifications.

8. The real-time processing and intelligent analysis system for ground-space time-frequency electromagnetic data according to claim 7 is characterized in that: The key data is classified by similarity through the K-means clustering algorithm, and the feature data of the sub-band is divided into different similarity categories, including the following steps: S1: Standardize the characteristic data in the data table; S2: Calculate the silhouette coefficients under different K values ​​by using the silhouette coefficient method, and select the K value with the highest silhouette coefficient as the final choice; S3: Randomly select K initial cluster centers; S4: For each sub-band feature data in the data table, calculate the distance between it and all cluster centers, and assign it to the cluster where the nearest cluster center is located; S5: according to the position of each sub-band feature data in all data tables in the current cluster, calculate the mean value of each sub-band feature data in all data tables in the current cluster, and update the position of the mean value of each sub-band feature data in all data tables in the current cluster as the cluster center; S6: Repeat steps S4 and S5 until the change in the position of the most recently updated cluster center is less than a preset distance, and take each sub-band feature data in all data tables in each cluster as data in the same similarity classification.

9. The real-time processing and intelligent analysis system for ground-space time-frequency electromagnetic data according to claim 7 is characterized in that: Through deep learning algorithms, models for anomaly detection and prediction of different similarity classifications are constructed, including the following steps: Acquire historical data of sub-band feature data of different similarity classifications, and image data of amplitude spectrum, phase spectrum and energy spectrum of corresponding sub-bands in the sub-band feature data of different similarity classifications; The historical data is divided into data when anomalies occur, data in a preset time period before anomalies, and normal data, and data labels are added to the corresponding data according to the classification; The neural network model is trained by adding labeled amplitude spectrum, phase spectrum and energy spectrum image data of different similarity classifications to build anomaly detection and prediction models of different similarity classifications.

Images