Dynamic identification detection system of electromagnetic pulse attack based on spectrum analysis
By performing differential and energy mutation analysis on the digital sequence of electromagnetic pulse signals and dynamically adjusting the window length, the resolution problem of the fixed window length STFT method in electromagnetic pulse attack identification is solved, achieving higher identification accuracy and adaptability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- MILITARY SECRECY QUALIFICATION EXAMINATION & CERTIFICATION CENT
- Filing Date
- 2026-02-10
- Publication Date
- 2026-05-29
AI Technical Summary
In existing technologies, the short-time Fourier transform method with a fixed window length is difficult to balance time and frequency resolution when facing complex and ever-changing electromagnetic pulse attacks, resulting in poor recognition performance. In particular, it suffers from insufficient resolution and poor adaptability in the dynamic recognition of electromagnetic pulse signals.
The digital sequence of the electromagnetic pulse signal is acquired through the data acquisition module. The signal segment is segmented by calculating the difference and energy mutation degree. Combined with frequency richness and comprehensive energy mutation degree analysis, the adaptive window length is dynamically adjusted and STFT processing is performed to obtain the time spectrum sequence.
It improves the accuracy of dynamic identification of electromagnetic pulse attacks, takes into account both frequency and time resolution, and achieves more accurate extraction and identification of time and frequency features of electromagnetic pulse signals.
Smart Images
Figure CN121682239B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and more specifically to a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis. Background Technology
[0002] Electromagnetic pulse (EMP) attacks, as a high-intensity, high-frequency, and transient electromagnetic interference method, possess time-domain burst characteristics ranging from nanoseconds to microseconds and a wide spectrum distribution from tens of megahertz to thousands of megahertz. They can easily cause serious interference or even permanent damage to sensitive electronic equipment; their impact on core modules such as communication terminals, computing platforms, sensor nodes, and embedded control devices is particularly significant.
[0003] In existing electromagnetic pulse (EMP) attack identification technologies, a method based on Short Time Fourier Transform (STFT) is typically used to extract time-frequency features from the signal. This algorithm segments the signal using a fixed window length to extract these features. However, EMP signals are characterized by their suddenness, short duration, and complex spectral distribution. They may manifest as narrowband high-frequency interference or broadband transient pulses in different scenarios. When using a fixed window length for STFT processing, the choice of window length directly affects the identification effect. A shorter window length, while improving time resolution, leads to insufficient frequency resolution and fails to accurately reflect the spectral structure of the EMP signal. A longer window length may result in inaccurate timing of the instantaneous signal or even smooth out pulse features, masking abrupt energy changes. For example, in a test of a power communication equipment subjected to a directional microwave attack, the use of a fixed-window-length STFT method failed to pinpoint the timing of the interference pulse, causing system identification delays and ineffective warnings. Similarly, in simulation analysis of nuclear EMP attacks, a fixed window length blurs the boundaries of pulse features in the spectral graph, affecting the classification accuracy of subsequent identification algorithms. Therefore, the existing STFT method with a fixed window length is prone to problems of insufficient resolution and poor adaptability when facing complex and ever-changing EMP attacks, making it difficult to meet the system's requirements for dynamic and accurate identification of electromagnetic pulse attacks. Summary of the Invention
[0004] To address the aforementioned technical problems, the present invention aims to provide a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis. The specific technical solution adopted is as follows:
[0005] The data acquisition module is used to acquire the digital sequence of electromagnetic pulse signals;
[0006] The data processing module is used to segment the digital sequence according to its changing trend characteristics to obtain signal segments of different categories; to obtain energy mutation degree values based on the data difference characteristics and fluctuation characteristics of the signal segments; to obtain feature difference values based on the difference characteristics of energy mutation degree values and data difference characteristics of any two signal segments of the same category; and to obtain a comprehensive difference value of the signal segment based on the time interval characteristics and feature difference values between the signal segment and other signal segments.
[0007] The feature analysis module is used to obtain frequency richness based on the quantity characteristics of corresponding signal segments within a preset sliding window of the digital sequence and the variation characteristics of the comprehensive difference value; and to obtain comprehensive energy mutation degree based on the distribution characteristics of the energy mutation degree values of the signal segments within the preset sliding window.
[0008] The signal recognition module is used to obtain a relative scaling factor based on the frequency richness and the comprehensive energy mutation degree of the same preset sliding window; to obtain an adaptive window length at any time based on the relative scaling factor of different preset sliding windows; and to perform STFT processing on the electromagnetic pulse signal within the adaptive window length to obtain a time-spectrum sequence.
[0009] Furthermore, the step of segmenting the digital sequence according to its changing trend characteristics to obtain different categories of signal segments includes:
[0010] The digital sequence is differentially divided to obtain a differential sequence; the digital sequence segments corresponding to different consecutive time periods exceeding the constant 0 in the differential sequence are classified as signal segments of one category; the digital sequence segments corresponding to different consecutive time periods equal to the constant 0 in the differential sequence are classified as signal segments of one category; and the digital sequence segments corresponding to different consecutive time periods below the constant 0 in the differential sequence are classified as signal segments of one category.
[0011] Furthermore, the step of obtaining the energy mutation degree value based on the data difference characteristics and fluctuation characteristics of the signal segment includes:
[0012] The degree of change is obtained by calculating the ratio of the absolute value of the difference between the first and last amplitudes of the signal segment to the number of amplitudes of the signal segment; the degree of change is obtained by multiplying the coefficient of variation of the differential sequence segment corresponding to the signal segment by the degree of change.
[0013] Furthermore, the step of obtaining the feature difference value based on the difference characteristics of the energy mutation degree values and the data difference characteristics of any two signal segments under the same category includes:
[0014] Calculate the absolute value of the difference between the energy change degree values of any two signal segments and normalize it to obtain a first difference value; calculate the absolute value of the difference between the median amplitude values of any two signal segments and normalize it to obtain a second difference value; calculate the sum of the first difference value and the second difference value to obtain the characteristic difference value of any two signal segments.
[0015] Further, the step of obtaining the comprehensive difference value of the signal segment based on the time interval characteristics and characteristic difference values between the signal segment and other signal segments includes:
[0016] The reciprocal of the time interval between the signal segment and other signal segments of the same category is used as the weight of the feature difference value; the weighted average of the feature difference values between the signal segment and all other signal segments is calculated to obtain the comprehensive difference value of the signal segment.
[0017] Furthermore, the step of obtaining the frequency richness based on the quantitative characteristics of the corresponding signal segments within a preset sliding window of the digital sequence and the variation characteristics of the comprehensive difference value includes:
[0018] The product of the coefficient of variation of the comprehensive difference value of the signal segments within the preset sliding window and the number of signal segments is calculated and positively correlated to obtain the frequency richness corresponding to the preset sliding window.
[0019] Further, the step of obtaining the comprehensive energy mutation degree based on the distribution characteristics of the energy mutation degree values of the signal segments within the preset sliding window includes:
[0020] Calculate the difference between the maximum and minimum values of the energy mutation degree of the signal segment within the preset sliding window to obtain a first difference; calculate the product of the first difference and the maximum value of the energy mutation degree within the preset sliding window and perform a positive correlation mapping to obtain the comprehensive energy mutation degree corresponding to the preset sliding window.
[0021] Further, the step of obtaining the relative scaling factor based on the frequency richness and the comprehensive energy abrupt change degree of the same preset sliding window includes:
[0022] The ratio of the frequency richness to the comprehensive energy mutation degree for the same preset sliding window is calculated to obtain the relative proportionality coefficient.
[0023] Furthermore, the step of obtaining the adaptive window length at any given time based on the relative scaling coefficients of different preset sliding windows includes:
[0024] All relative scaling coefficients are sorted according to the time order of different preset sliding windows. The sorting results are interpolated and smoothed to obtain a relative scaling coefficient sequence of the same length as the number sequence. The product of the relative scaling coefficient at any time in the number sequence and the preset window length is calculated to obtain the adaptive window length at that time.
[0025] The present invention has the following beneficial effects:
[0026] In this invention, acquiring different categories of signal segments allows for segmentation of the digital sequence, facilitating more accurate analysis of frequency and energy mutation characteristics. The energy mutation degree value reflects the energy mutation characteristics of the signal segment. Due to the significant non-stationarity of electromagnetic pulse signals, complex features such as energy spikes, frequency jumps, and amplitude jitter may occur over time. Acquiring feature difference values characterizes the differences in energy characteristics between different signal segments. Since electromagnetic pulse signals with more recent times should have more similar characteristics, acquiring comprehensive difference values can more accurately characterize the degree of feature difference between a signal segment and other signal segments based on time interval characteristics. Acquiring frequency richness characterizes the component distribution of the signal in the frequency domain, while acquiring comprehensive energy mutation degree reflects the instantaneous change intensity of the signal in the time domain, thus reflecting abnormal signal behavior from different dimensions. Acquiring the relative proportionality coefficient characterizes the relationship between frequency richness and comprehensive energy mutation degree, thereby improving the accuracy of determining the adaptive window length. Acquiring the adaptive window length ensures that the frequency resolution and time resolution better match the actual signal characteristics during time-frequency feature extraction, taking into account both time and frequency characteristics. Finally, acquiring the time-spectrum sequence based on the adaptive window improves the accuracy of dynamic identification of electromagnetic pulse attacks. Attached Figure Description
[0027] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0028] Figure 1 This is a block diagram of a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis, provided as an embodiment of the present invention. Detailed Implementation
[0029] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0031] The following description, in conjunction with the accompanying drawings, details a specific scheme for a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis provided by this invention.
[0032] Please see Figure 1 The diagram illustrates a block diagram of a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis, according to an embodiment of the present invention. The system includes the following modules:
[0033] The data acquisition module S1 is used to acquire the digital sequence of electromagnetic pulse signals.
[0034] In this embodiment of the invention, the implementation scenario is the dynamic identification of electromagnetic pulse signals to improve identification accuracy. During the dynamic identification and monitoring of electromagnetic pulse signal attacks, the electromagnetic environment of the target area is first monitored in real time using a high-frequency broadband receiving antenna or electromagnetic detection sensor. The received analog electromagnetic signals are then highly sampled using a high-speed analog-to-digital converter and converted into a digital sequence, thereby obtaining the digital sequence of the electromagnetic pulse signal.
[0035] The data processing module S2 is used to segment the digital sequence according to the changing trend characteristics to obtain signal segments of different categories; to obtain the energy mutation degree value according to the data difference characteristics and fluctuation characteristics of the signal segments; to obtain the feature difference value according to the difference characteristics of the energy mutation degree values and data difference characteristics of any two signal segments in the same category; and to obtain the comprehensive difference value of the signal segment according to the time interval characteristics and feature difference values of the signal segment and other signal segments.
[0036] Electromagnetic pulse (EMP) attacks refer to attacks that interfere with, damage, or paralyze electronic devices and systems by releasing intense, transient electromagnetic energy. They are typically characterized by high energy, extremely short duration, rapid propagation, and a wide impact range, covering frequencies from tens of megahertz to thousands of megahertz. Because EMP signals differ from ordinary signals, existing methods typically utilize spectral analysis for EMP identification and detection, with Short-Time Fourier Transform (STFT) widely used to extract time-frequency features. However, in practical applications, EMP signals often exhibit strong transients, sudden bursts, and large frequency spans, making it difficult to balance time and frequency resolution when using the STFT algorithm for real-time signal time-frequency feature extraction. For example, a short window length, while improving time resolution, leads to insufficient frequency resolution, failing to accurately reflect the spectral structure of the EMP signal; a long window length may result in inaccurate timing of the transient signal, or even smooth out the pulse characteristics, masking its abrupt energy changes and causing missed identification opportunities.
[0037] Furthermore, in this embodiment of the invention, the analysis is based on a digital sequence acquired in real time in the environment of a power grid device. To more accurately identify potential electromagnetic pulse (EMP) attacks in the acquired signals, it is necessary to initialize and analyze the frequency and other characteristics of the acquired signals. This allows for the acquisition of a window length that conforms to the signal transformation characteristics when the signal undergoes a short-time Fourier transform (STFT). This facilitates the use of the STFT algorithm to more accurately acquire the time-frequency characteristics of the monitored signal, thereby achieving more effective and accurate dynamic identification of EMP signal attacks. First, the digital sequence is segmented according to its changing trend characteristics to obtain different categories of signal segments. Preferably, in this embodiment, the step of obtaining different categories of signal segments includes: differentiating the digital sequence to obtain a differential sequence; the differential sequence reflects the amplitude change characteristics of adjacent moments in the digital sequence. The digital sequence segments corresponding to different consecutive time periods exceeding a constant 0 in the differential sequence are classified as a type of signal segment; consecutive time periods exceeding a constant 0 in the differential sequence indicate an upward trend in the amplitude of the digital sequence. The digital sequence segments corresponding to different consecutive time periods equal to a constant 0 in the differential sequence are classified as a type of signal segment; consecutive time periods equal to a constant 0 in the differential sequence indicate a stable and unchanging trend in the amplitude of the digital sequence. The digital sequence segments corresponding to different consecutive time periods below the constant 0 in the difference sequence are classified as signal segments. Consecutive time periods below the constant 0 in the difference sequence indicate a decreasing trend in the amplitude of the digital sequence. The difference sequence allows for the division and classification of consecutive time periods with different amplitude trends in the digital sequence. Signal segments within the same category exhibit the same amplitude change trend, facilitating a more accurate analysis of the frequency and energy abrupt change characteristics of the signal segments.
[0038] For any signal segment within any category, the energy mutation degree value is first obtained to obtain a more accurate and suitable window based on the energy mutation characteristics and frequency characteristics of the signal. Therefore, the energy mutation degree value is obtained based on the data difference characteristics and fluctuation characteristics of the signal segment. Preferably, in this embodiment, the step of obtaining the energy mutation degree value includes: calculating the ratio of the absolute value of the difference between the first and last amplitudes of the signal segment to the number of amplitudes in the signal segment to obtain the degree of change; the greater the degree of change, the greater the amplitude change of the signal segment and the stronger the energy mutation degree. The product of the coefficient of variation of the differential sequence segment corresponding to the signal segment and the degree of change is calculated to obtain the energy mutation degree value of the signal segment. It should be noted that the coefficient of variation is existing technology; the greater the coefficient of variation, the worse the data stability, the more obvious the fluctuation, and the stronger the energy mutation degree in the signal segment. When the coefficient of variation is zero, a preset minimum positive number is used to replace the coefficient of variation to avoid affecting the calculation result of the energy mutation degree value. In this embodiment, the preset minimum positive number is 0.01. The greater the energy mutation degree value of the signal segment, the more obvious the overall mutation degree of the signal in that segment.
[0039] Furthermore, since electromagnetic pulse signals typically exhibit significant non-stationarity, they may display complex characteristics over time, such as sudden energy increases, frequency jumps, and amplitude jitter. To enable STFT (Single-TFT) to dynamically sense the degree of local abrupt changes or frequency richness features of electromagnetic pulse signals during feature extraction, thereby obtaining a more suitable window length, it is necessary to compare and analyze the degree and magnitude of energy abrupt changes between any given signal segment and other signal segments. For example, if a signal segment shows a significant difference in energy compared to other signal segments, it usually indicates that the segment may experience abnormal energy fluctuations or sudden interference, requiring focused analysis. Subsequently, feature difference values are obtained based on the differences in the degree of energy abrupt changes and data differences between any two signal segments of the same category.
[0040] Preferably, in this embodiment of the invention, the step of obtaining the feature difference value includes: calculating and normalizing the absolute value of the difference between the energy change degree values of any two signal segments to obtain a first difference value; the normalization method is linear normalization. The larger the difference in the energy change degree values of the two signal segments, the larger the first difference value, meaning the difference between the two signal segments is greater. The absolute value of the difference between the median amplitudes of the two signal segments is calculated and normalized to obtain a second difference value; the larger the difference in the median amplitudes of the two signal segments, the larger the second difference value, meaning the difference in the local amplitude characteristics of the two signal segments is greater. The sum of the first difference value and the second difference value is calculated to obtain the feature difference value of the two signal segments; the larger the feature difference value, the more obvious the signal characteristic difference between the two signal segments.
[0041] Furthermore, after obtaining the feature difference values of any two signal segments within the same category, the characteristics of electromagnetic pulse signals with closer timing should be more similar. A larger feature difference value between a signal segment and its nearest neighboring signal segments indicates a more significant difference in signal characteristics. Therefore, a comprehensive difference value for the signal segment is obtained based on the time interval characteristics and feature difference values between the signal segment and other signal segments. Preferably, in this embodiment, the step of obtaining the comprehensive difference value includes: using the reciprocal of the time interval between the signal segment and other signal segments of the same category as the weight of the feature difference value; the closer the time interval, the greater the weight, and the greater the contribution of the corresponding feature difference value. The weighted average of the feature difference values between the signal segment and all other signal segments is calculated to obtain the comprehensive difference value of the signal segment; a larger comprehensive difference value indicates a greater feature difference between the signal segment and other signal segments of the same category.
[0042] The feature analysis module S3 is used to obtain frequency richness based on the quantity characteristics and the variation characteristics of the comprehensive difference value of the corresponding signal segments within the preset sliding window of the digital sequence; and to obtain the comprehensive energy mutation degree based on the distribution characteristics of the energy mutation degree value of the signal segments within the preset sliding window.
[0043] After obtaining the comprehensive difference value of different signal segments, the frequency richness of the digital sequence can be analyzed. Therefore, the frequency richness is obtained based on the quantity characteristics of the corresponding signal segments and the change characteristics of the comprehensive difference value within a preset sliding window of the digital sequence. Preferably, in this embodiment of the invention, the step of obtaining the frequency richness includes: calculating the product of the coefficient of variation of the comprehensive difference value of the signal segments within the preset sliding window and the number of signal segments, and mapping them positively to obtain the frequency richness corresponding to the preset sliding window. In this embodiment of the invention, the length of the preset sliding window is the length of a sequence segment of 5 microseconds, and the sliding step size of the preset sliding window in the digital sequence is half of the window. The implementer can determine this according to the implementation scenario. The more signal segments within the preset sliding window, the faster the frequency of the signal change within the preset sliding window may be, and the greater its frequency richness may be. The larger the coefficient of variation of the comprehensive difference value of the signal segments within the preset sliding window, the more complex the frequency of the signal segments in that time period may be, and the greater its frequency richness may be. It should be noted that when the coefficient of variation is zero, a preset minimum positive number is used as a substitute to avoid affecting the calculation result of the frequency richness. The purpose of positive correlation mapping is to ensure that different data are analyzed on the same scale. In this embodiment of the invention, the formula for positive correlation mapping is: In the formula, X represents the mapping result. This represents the maximum value of the mapped dataset. This represents the minimum value of the mapped dataset. Indicates the required mapping value. This represents the maximum value of the mapped interval. This represents the minimum value of the mapping interval.
[0044] Furthermore, since electromagnetic pulse signals typically exhibit both complex spectral structures and dramatic energy abrupt changes, it is necessary to simultaneously extract the degree of local abrupt changes within a preset sliding window. Frequency richness reflects the distribution of signal components in the frequency domain, while the degree of abrupt change reveals the intensity of instantaneous changes in the signal in the time domain. Both reflect abnormal signal behavior from different dimensions. By jointly analyzing these two characteristics, a comprehensive perception of signal characteristics within the preset sliding window can be achieved. This allows for dynamic adjustment of the STFT window length, ensuring a larger window is used when the frequency is relatively complex, while a smaller window is used when the degree of abrupt change is large, thereby improving time positioning accuracy and ultimately enhancing the dynamic recognition effect of non-stationary electromagnetic pulse signals. Therefore, the comprehensive energy abrupt change degree is obtained based on the distribution characteristics of the energy abrupt change degree values of the signal segment within the preset sliding window.
[0045] Preferably, in this embodiment of the invention, the step of obtaining the comprehensive energy mutation degree includes: calculating the difference between the maximum and minimum values of the energy mutation degree of the signal segment within a preset sliding window to obtain a first difference; the larger the difference between the maximum and minimum values, the larger the first difference, which means that the drop in the energy mutation degree of the signal segment within that time period is greater, and the more obvious the energy mutation degree is. The product of the first difference and the maximum value of the energy mutation degree within the preset sliding window is calculated and positively correlated to obtain the comprehensive energy mutation degree corresponding to the preset sliding window; the larger the first difference and the larger the maximum value, the larger the comprehensive energy mutation degree, which means that the energy mutation degree of the signal segment within that time period is greater.
[0046] The signal recognition module S4 is used to obtain a relative scaling factor based on the frequency richness and comprehensive energy change rate of the same preset sliding window; to obtain the adaptive window length at any time based on the relative scaling factor of different preset sliding windows; and to perform STFT processing on the electromagnetic pulse signal within the adaptive window length to obtain a time spectrum sequence.
[0047] After obtaining the frequency richness and comprehensive energy mutation degree corresponding to the preset sliding window, a relative scaling factor can be obtained based on the frequency richness and comprehensive energy mutation degree of the same preset sliding window. Preferably, in this embodiment of the invention, the step of obtaining the relative scaling factor includes: calculating the ratio of the frequency richness to the comprehensive energy mutation degree of the same preset sliding window to obtain the relative scaling factor. The larger the relative scaling factor, the greater the frequency richness within the preset sliding window; conversely, the smaller the relative scaling factor, the greater the degree of energy mutation. Further, the adaptive window length at any time can be obtained based on the relative scaling factors of different preset sliding windows, specifically including: sorting all relative scaling factors according to the time order of different preset sliding windows, interpolating and smoothing the sorting results to obtain a relative scaling factor sequence of the same length as the digital sequence. In this embodiment of the invention, the existing spline interpolation method is used for interpolation, and Gaussian filtering is used to smooth the interpolated sequence. The window adjustment coefficient when using STFT for time-frequency feature extraction at any time in the digital sequence can be obtained through the relative scaling factor sequence. The product of the relative scaling factor corresponding to any time in the digital sequence and the preset window length is calculated to obtain the adaptive window length at that time. A common electromagnetic measurement window length is 5.12 microseconds, which is used as the preset window length in this embodiment of the invention. The implementer can determine the length according to the implementation scenario. A larger relative scaling factor indicates a greater frequency richness within the sliding window, requiring a larger window for time-frequency feature extraction to achieve higher frequency resolution and accurately characterize its spectral structure. Conversely, a smaller scaling factor indicates a greater energy abrupt change within the sliding window, requiring a smaller window for time-frequency feature extraction to improve time resolution and facilitate precise capture of instantaneous energy changes. The relationship between these two factors allows for dynamic adjustment of the window size to balance time-frequency characteristics.
[0048] Furthermore, STFT processing can be performed on the electromagnetic pulse signal within the adaptive window length to obtain a time-spectrum sequence. The digital sequence is traversed, and STFT processing is performed step-by-step at each time point according to the corresponding adaptive window length to generate a high-resolution time-spectrum sequence. It should be noted that Short-Time Fourier Transform (STFT) is an existing technology, and the specific steps will not be elaborated further. After obtaining the time-spectrum sequence, time-frequency attack features can be extracted from each time-spectrum sequence using feature extraction algorithms, such as those based on spectral entropy, frequency centroid, instantaneous bandwidth, and energy focusing. Then, existing anomaly detection models, such as isolated forests and single-class support vector machines, are used to identify, classify, and monitor the attack features. Finally, early warning conditions are set for early warning, such as issuing an alert when three consecutive anomalies are detected, thereby achieving more accurate dynamic identification and monitoring of electromagnetic pulse attacks. Implementers can customize the identification and monitoring method based on the time-spectrum sequence according to the implementation scenario.
[0049] In summary, this invention provides a dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis. The system segments the signal sequence according to its changing trends, obtains energy mutation degree values based on data difference and fluctuation characteristics of the signal segments, obtains feature difference values based on the energy mutation degree values and data difference characteristics of two signal segments of the same category, obtains comprehensive difference values of the signal segments based on time interval characteristics and feature difference values, obtains frequency richness based on the number of signal segments within a preset sliding window and comprehensive difference values, obtains comprehensive energy mutation degree based on the energy mutation degree values of the signal segments within the preset sliding window, and obtains a relative proportionality coefficient based on frequency richness and comprehensive energy mutation degree. This invention obtains an adaptive window length based on the relative proportionality coefficient; and performs STFT processing on the electromagnetic pulse signal within the adaptive window length, improving the accuracy of dynamic identification.
[0050] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0051] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. A dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis, characterized in that, The system includes the following modules: The data acquisition module is used to acquire the digital sequence of electromagnetic pulse signals; The data processing module is used to segment the digital sequence according to its changing trend characteristics to obtain signal segments of different categories; to obtain energy mutation degree values based on the data difference characteristics and fluctuation characteristics of the signal segments; to obtain feature difference values based on the difference characteristics of energy mutation degree values and data difference characteristics of any two signal segments of the same category; and to obtain a comprehensive difference value of the signal segment based on the time interval characteristics and feature difference values between the signal segment and other signal segments. The feature analysis module is used to obtain frequency richness based on the quantity characteristics of corresponding signal segments within a preset sliding window of the digital sequence and the variation characteristics of the comprehensive difference value; and to obtain comprehensive energy mutation degree based on the distribution characteristics of the energy mutation degree values of the signal segments within the preset sliding window. The signal recognition module is used to obtain a relative scaling factor based on the frequency richness and the comprehensive energy mutation degree of the same preset sliding window; to obtain an adaptive window length at any time based on the relative scaling factor of different preset sliding windows; and to perform STFT processing on the electromagnetic pulse signal within the adaptive window length to obtain a time-spectrum sequence. The step of obtaining the energy mutation degree value based on the data difference characteristics and fluctuation characteristics of the signal segment includes: The degree of change is obtained by calculating the ratio of the absolute value of the difference between the first and last amplitudes of the signal segment to the number of amplitudes of the signal segment; the degree of change is obtained by multiplying the coefficient of variation of the differential sequence segment corresponding to the signal segment by the degree of change. The step of obtaining the adaptive window length at any given time based on the relative scaling coefficients of different preset sliding windows includes: All relative scaling coefficients are sorted according to the time order of different preset sliding windows. The sorting results are interpolated and smoothed to obtain a relative scaling coefficient sequence of the same length as the number sequence. The product of the relative scaling coefficient at any time in the number sequence and the preset window length is calculated to obtain the adaptive window length at that time.
2. The dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis according to claim 1, characterized in that, The step of segmenting the digital sequence according to its changing trend characteristics to obtain different categories of signal segments includes: The digital sequence is differentially divided to obtain a differential sequence; the digital sequence segments corresponding to different consecutive time periods exceeding the constant 0 in the differential sequence are classified as signal segments of one category; the digital sequence segments corresponding to different consecutive time periods equal to the constant 0 in the differential sequence are classified as signal segments of one category; and the digital sequence segments corresponding to different consecutive time periods below the constant 0 in the differential sequence are classified as signal segments of one category.
3. The dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis according to claim 1, characterized in that, The step of obtaining the feature difference value based on the difference characteristics of energy mutation degree values and data difference characteristics of any two signal segments under the same category includes: Calculate the absolute value of the difference between the energy change degree values of any two signal segments and normalize it to obtain a first difference value; calculate the absolute value of the difference between the median amplitude values of any two signal segments and normalize it to obtain a second difference value; calculate the sum of the first difference value and the second difference value to obtain the characteristic difference value of any two signal segments.
4. The dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis according to claim 1, characterized in that, The step of obtaining the comprehensive difference value of the signal segment based on the time interval characteristics and feature difference values between the signal segment and other signal segments includes: The reciprocal of the time interval between the signal segment and other signal segments of the same category is used as the weight of the feature difference value; the weighted average of the feature difference values between the signal segment and all other signal segments is calculated to obtain the comprehensive difference value of the signal segment.
5. The dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis according to claim 1, characterized in that, The step of obtaining frequency richness based on the quantitative characteristics of the corresponding signal segments within a preset sliding window of the digital sequence and the variation characteristics of the comprehensive difference value includes: The product of the coefficient of variation of the comprehensive difference value of the signal segments within the preset sliding window and the number of signal segments is calculated and positively correlated to obtain the frequency richness corresponding to the preset sliding window.
6. The dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis according to claim 1, characterized in that, The step of obtaining the comprehensive energy mutation degree based on the distribution characteristics of the energy mutation degree values of the signal segments within the preset sliding window includes: Calculate the difference between the maximum and minimum values of the energy mutation degree of the signal segment within the preset sliding window to obtain a first difference; calculate the product of the first difference and the maximum value of the energy mutation degree within the preset sliding window and perform a positive correlation mapping to obtain the comprehensive energy mutation degree corresponding to the preset sliding window.
7. The dynamic identification and detection system for electromagnetic pulse attacks based on spectrum analysis according to claim 1, characterized in that, The step of obtaining the relative scaling factor based on the frequency richness and the comprehensive energy abrupt change degree of the same preset sliding window includes: The ratio of the frequency richness to the comprehensive energy mutation degree for the same preset sliding window is calculated to obtain the relative proportionality coefficient.