Sleep node detection method and system based on high-energy pulse and harmonic analysis

Abstract

The present application belongs to the field of radio detection technology, and particularly relates to a dormant node detection method and system based on high-energy pulse and harmonic analysis, comprising the following steps: obtaining a wideband electromagnetic background sequence and a spatial position grid of a region to be detected, calculating the background fluctuation index, pulse accommodation degree and spectral drift rate of adjacent time periods of each grid point in a preset frequency band; constructing an excitation parameter initial value matrix, determining the initial values of pulse amplitude, time width and pulse interval; applying multiple groups of pulses with different parameters to the region to be detected according to the determined initial values of pulse amplitude, time width and pulse interval, collecting echo signals, harmonic components and electromagnetic disturbance sequences. The present application uses a locking parameter group for retesting, and comprehensively investigates the time delay aggregation degree and the carrier leakage coefficient, thereby rigorously excluding environmental noise and false target interference, and improving the accuracy and reliability of the detection results.

Description

Technical Field

[0001] This invention belongs to the field of radio detection technology, specifically relating to a method and system for detecting dormant nodes based on high-energy pulse and harmonic analysis. Background Technology

[0002] Traditional locator detection relies on basic nonlinear junction detection technology, which involves transmitting continuous microwave signals to the target and receiving harmonics generated by internal semiconductor devices. However, the increasingly dense broadband electromagnetic background interference in modern detection environments, coupled with the robust electromagnetic shielding and low-power sleep architecture commonly used in new GPS locators, results in insufficient penetration and extremely weak echo responses from traditional single-parameter detection signals in complex scenarios. When faced with drastically changing background noise, deeply concealed targets, or dormant nodes, traditional detection methods are prone to high-frequency false alarms and missed detections, failing to meet the practical requirements of screening modern high-precision concealed electronic devices. High-energy pulse excitation can instantly inject high transient energy into the target, exciting the semiconductor nonlinear junction to generate rich harmonic components and broadband echo signals. By adjusting parameters such as pulse amplitude, duration, and pulse interval, and combining this with time-frequency analysis techniques to process the echoes, the unique nonlinear response characteristics of the target device can be extracted to a certain extent, thereby improving the detection success rate in complex environments.

[0003] Chinese patent application CN115561744A discloses a nonlinear node detection method and detector. It uses continuous microwave excitation and receives harmonic echoes generated by semiconductor nonlinear junctions to achieve target detection based on amplitude and phase difference. However, this method uses fixed parameter excitation, does not combine environmental background adaptive optimization, relies only on single harmonic feature interpretation, and lacks high-energy pulse wake-up and retest verification mechanisms. It has insufficient penetration for dormant and strongly shielded targets, is susceptible to metal interference, and has a high false alarm and false negative rate in complex electromagnetic environments, which cannot meet the requirements for high-precision concealed node detection.

[0004] Existing methods largely rely on fixed presets or simple empirical iterations when setting pulse excitation parameters, failing to perceive and optimize the initial parameter matrix for broadband electromagnetic background fluctuations, pulse tolerance, and spectral drift rates between adjacent time periods at the spatial grid points of the test area. This results in pulse energy being easily absorbed by the environment or causing severe electromagnetic background overload interference. In feature extraction and target identification, existing technologies often only involve simple frequency domain amplitude comparisons, failing to fully explore deep edge constraints such as the normalized leading edge slope and normalized trailing edge fall-off time of the echo signal. Furthermore, they neglect the crucial role of electromagnetic disturbance memory length in subsequent pulse interval adjustments. Existing technologies lack the ability to construct nonlinear response spectra by combining the energy ratios of fundamental, harmonic, and fractional frequency orders. They fail to establish a precise mapping relationship between pulse parameters and response characteristics to calculate curvature offset, phase reversal density, and response hysteresis index. Moreover, they lack a comprehensive mechanism for retesting based on locked parameter sets to calculate delay clustering, carrier leakage coefficient, and post-pulse residual oscillation index, thus limiting the positioning and identification accuracy of the detection system in complex backgrounds. Summary of the Invention

[0005] This invention provides a method and system for detecting dormant nodes based on high-energy pulse and harmonic analysis, in order to solve the technical problems of weak penetration, high false alarm and false negative rates, and low detection accuracy of dormant node detection in complex electromagnetic environments in the prior art.

[0006] In a first aspect, the present invention provides a method for detecting dormant nodes based on high-energy pulse and harmonic analysis, comprising the following steps: The process involves acquiring a broadband electromagnetic background sequence and spatial location grid for the area under test, calculating the background fluctuation index, pulse tolerance, and spectral drift rate of each grid point in the preset frequency band; constructing an initial value matrix for excitation parameters, and determining the initial values ​​for pulse amplitude, duration, and pulse interval; applying multiple sets of pulse excitations with different parameters to the area under test according to the determined initial values ​​for pulse amplitude, duration, and pulse interval, acquiring echo signals, harmonic components, and electromagnetic disturbance sequences, extracting the normalized leading edge slope and normalized trailing edge fall time of the echo signals to generate edge constraint quantities; and calculating the disturbance memory length of the electromagnetic disturbance sequence relative to the broadband electromagnetic background sequence to adjust the subsequent pulse interval. The echo signal is segmented for time-frequency analysis to calculate the energy ratios of the fundamental frequency, harmonics, and fractional frequencies. The segment overlap rate is adjusted in conjunction with the spectral drift rate. A nonlinear response spectrum is constructed using the energy ratios, edge constraints, and perturbation memory length. A mapping relationship between pulse parameters and response characteristics is established based on the nonlinear response spectrum. The curvature offset, phase reversal density, and response hysteresis exponent are calculated. A pulse parameter set that meets preset conditions is selected as the locked parameter set. The test was repeated using the locked parameter set; the delay concentration, carrier leakage coefficient and post-pulse residual oscillation index were calculated; and the test results were output by combining the nonlinear response spectrum and the response hysteresis index.

[0007] Its effects are as follows: by acquiring a broadband electromagnetic background and constructing a multidimensional nonlinear response spectrum, it solves the problems of insufficient penetration of traditional continuous microwave detection in complex electromagnetic environments and susceptibility to environmental noise interference. It achieves accurate extraction of the physical characteristics of target electronic devices under strong background interference, and significantly improves the recognition success rate and detection distance of dormant targets.

[0008] Furthermore, the broadband electromagnetic background sequence and spatial location grid of the area to be measured are obtained, and the background fluctuation index, pulse tolerance, and spectral drift rate of each grid point in the preset frequency band are calculated, including: The broadband electromagnetic background sequence was divided into sub-bands with a bandwidth of 500kHz, and the amplitude standard deviation of each sub-band was extracted as the background fluctuation index. The first derivative of the broadband electromagnetic background sequence is calculated to generate a derivative sequence. The absolute amplitude of the broadband electromagnetic background sequence between the zero crossover points where the signs of the derivatives reverse in two consecutive times is extracted and integrated. The resulting integrated area is used as the pulse tolerance. A fast Fourier transform is performed on the broadband electromagnetic background sequence every 10ms. The center frequency shift value is extracted from the spectrum generated by two consecutive fast Fourier transforms. The center frequency shift value is divided by 10ms to obtain the spectral drift rate of adjacent time periods.

[0009] Its effect is that by dividing the background sequence into sub-bands and performing integration operations, it can adaptively sense the background noise level of different frequency bands in the detection area, solve the technical difficulty that the detection pulse energy is easily swallowed by the environmental background or causes overload, and ensure the accurate matching of excitation parameters with the current radio environment.

[0010] Furthermore, an initial value matrix for the excitation parameters is constructed to determine the initial values ​​for the pulse amplitude, pulse width, and pulse interval, including: The result of multiplying the background undulation index by a constant term is taken as the first element of the main diagonal; The value obtained by dividing the pulse tolerance by the total duration of the broadband electromagnetic background sequence is taken as the second element of the main diagonal; The reciprocal of the spectral drift rate is taken as the third element of the main diagonal; All matrix elements except the main diagonal are set to 0 to construct a 3×3 third-order diagonal matrix as the initial value matrix for the excitation parameters. Multiply the first element of the main diagonal by the first dimension conversion factor to map it as the initial value of the pulse amplitude; multiply the second element of the main diagonal by the second dimension conversion factor to map it as the initial value of the pulse width; and multiply the third element of the main diagonal by the third dimension conversion factor to map it as the initial value of the pulse interval.

[0011] Its effects are as follows: by using a diagonal matrix to map background features and determine the initial values ​​of pulse amplitude and duration, it avoids the problem of low detection efficiency caused by blindly traversing parameters in traditional methods. Through mathematical modeling, it achieves the optimal allocation of detection resources and significantly shortens the search cycle for dormant nodes in a large area.

[0012] Furthermore, the perturbation memory length of the electromagnetic perturbation sequence relative to the broadband electromagnetic background sequence is calculated to adjust the subsequent pulse interval, including: Perform discrete cross-correlation operations on the electromagnetic disturbance sequence and the broadband electromagnetic background sequence to generate a one-dimensional array of cross-correlation functions; Extract the position in the one-dimensional array of cross-correlation functions where the absolute value of the amplitude first drops to 10% of the absolute value of the amplitude of the maximum peak, and use the corresponding position as the decay coordinate point; Calculate the index distance between the location coordinates corresponding to the maximum peak and the attenuation coordinates, and multiply the index distance by the sampling period to obtain the perturbation memory length; The subsequent pulse interval is adjusted to be the sum of the perturbation memory length and the preset safety time margin.

[0013] Its effect is that by dynamically adjusting the pulse interval by calculating the perturbation memory length determined by the cross-correlation function, the energy superposition and aliasing and intermodulation interference that may be generated by the excitation of adjacent strong pulses in the time domain are eliminated, ensuring the purity and timing determinism of echo signal acquisition in complex application scenarios.

[0014] Furthermore, a nonlinear response spectrum is constructed using the energy ratio, edge constraint, and perturbation memory length, including: A three-dimensional Cartesian coordinate system is established using the energy ratio as the X-axis coordinate data, the edge constraint quantity formed by multiplying the normalized front slope by the preset dimension conversion coefficient and adding the normalized back edge fall time as the Y-axis coordinate data, and the perturbation memory length as the Z-axis coordinate data. The feature data of different grid points in the area to be tested are mapped one by one to a three-dimensional Cartesian coordinate system. The mean value of the Z-axis coordinate data with the same X-axis and Y-axis coordinate data is calculated to generate a single-value spatial three-dimensional scatter set. Thin plate spline interpolation algorithm is used to perform meshed surface fitting on a three-dimensional scatter point set in space to generate a continuous three-dimensional spatial surface, which is then used as a nonlinear response spectrum.

[0015] Its effect is that by constructing a three-dimensional nonlinear response spectrum that includes energy ratio, edge constraint and memory length and performing surface fitting, the physical response characteristics of the hidden target are transformed into an intuitive feature fingerprint spectrum, which solves the problem that existing technologies rely solely on frequency domain amplitude interpretation, making it difficult to distinguish between real semiconductor junctions and linear metal reflectors.

[0016] Furthermore, the curvature offset, phase reversal density, and response hysteresis index are calculated, including: Calculate the principal curvature values ​​corresponding to each coordinate point on the nonlinear response spectrum surface, subtract the principal curvature values ​​from the fixed curvature 0 of the reference plane, and extract the absolute value of the subtraction result as the curvature offset. The instantaneous phase of the echo signal within adjacent 1μs window widths is extracted and subtracted to obtain the initial phase difference. The number of transitions with an absolute value greater than π radians in the initial phase difference is counted as the phase reversal count. The initial phase difference is processed by principal value mapping with a period of 2π to eliminate spurious 2π truncation transitions caused by range limitation. The phase reversal count is divided by the total duration of the echo signal to obtain the phase reversal density. The peak moment of the echo signal pulse is extracted as the start time, and the moment when the amplitude of the harmonic component decays to 5% of the peak amplitude is extracted as the end time. The end time is subtracted from the start time, and the difference is multiplied by the damping penalty factor to generate the response hysteresis index.

[0017] Furthermore, the delay clustering, carrier leakage coefficient, and post-pulse residual oscillation index are calculated, including: Extract the time delay value of the peak value of the echo signal envelope corresponding to each pulse excitation during the retest period to form a time series, calculate the statistical variance of the time series, and divide 1 by the sum of the statistical variance and the preset positive constant as the time delay clustering degree. The echo baseband signal within 200μs after the retest pulse is turned off is intercepted, and the maximum residual level amplitude of this echo baseband signal is extracted. The maximum residual level amplitude is divided by the full-scale transmission level amplitude under pulse excitation state as the carrier leakage coefficient. The electromagnetic disturbance sequence after the retest pulse cutoff time is smoothed by a sliding window with a step size of 0.1μs to obtain the base signal. The number of fluctuation extreme points in the electromagnetic disturbance sequence whose amplitude exceeds the 10% amplitude boundary of the base signal is counted. The number of fluctuation extreme points is divided by the total number of retest samplings to obtain the residual oscillation index after the pulse.

[0018] Furthermore, a bandpass filter bank is used to divide the broadband electromagnetic background sequence into multiple sub-bands.

[0019] Furthermore, a derivative sequence is generated by calculating the first derivative of the broadband electromagnetic background sequence using the forward difference method.

[0020] Secondly, the present invention provides a dormant node detection system based on high-energy pulse and harmonic analysis, including a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned dormant node detection method based on high-energy pulse and harmonic analysis is implemented.

[0021] The beneficial effects are as follows: This invention acquires a broadband electromagnetic background sequence and spatial location grid of the area under test, and comprehensively evaluates the background fluctuation index and pulse tolerance, enabling accurate perception of complex electromagnetic environments. During the pulse excitation stage, the edge constraint and perturbation memory length of the echo signal are extracted to specifically adjust the subsequent pulse interval, improving the matching degree and detection efficiency of pulse excitation. Segmented time-frequency analysis is used to calculate the energy ratio of multiple frequencies, and a nonlinear response spectrum is constructed by combining phase reversal density and response hysteresis index, deeply exploring the unique electromagnetic response law of the target device and enhancing the nonlinear feature detection capability of hidden GPS locators. Retesting is performed using a locked parameter set, and the time delay clustering and carrier leakage coefficient are comprehensively examined, rigorously eliminating environmental noise and false target interference, improving the accuracy and reliability of the detection results. Attached Figure Description

[0022] Figure 1 This is a flowchart of a dormant node detection method based on high-energy pulse and harmonic analysis.

[0023] Figure 2 This is a schematic diagram of broadband electromagnetic background and pulse tolerance.

[0024] Figure 3 This is a schematic diagram of pulse harmonic envelope and response hysteresis.

[0025] Figure 4 This is a diagram illustrating the comparison of detection results. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] An embodiment of the dormant node detection method based on high-energy pulse and harmonic analysis provided by this invention: like Figure 1 As shown, the method for detecting dormant nodes based on high-energy pulse and harmonic analysis includes the following steps: S1: Obtain background and mesh, calculate parameters and excite, extract features and adjust pulse interval.

[0028] The process involves acquiring a broadband electromagnetic background sequence and spatial location grid for the area under test, calculating the background fluctuation index, pulse tolerance, and spectral drift rate of each grid point in the preset frequency band; constructing an initial value matrix for excitation parameters, and determining the initial values ​​for pulse amplitude, duration, and pulse interval; applying multiple sets of pulse excitations with different parameters to the area under test according to the determined initial values ​​for pulse amplitude, duration, and pulse interval, acquiring echo signals, harmonic components, and electromagnetic disturbance sequences, extracting the normalized leading edge slope and normalized trailing edge fall time of the echo signals to generate edge constraint quantities; and calculating the disturbance memory length of the electromagnetic disturbance sequence relative to the broadband electromagnetic background sequence to adjust the subsequent pulse interval.

[0029] An omnidirectional broadband antenna is connected via a general-purpose software radio peripheral to scan the area under test and receive radio frequency signals. These signals are then sampled by an analog-to-digital converter to obtain a broadband electromagnetic background sequence. Simultaneously, a differential GPS receiver is used to acquire latitude and longitude coordinates to construct a spatial location grid. The Welch method from the NumPy library is used to calculate the power spectral density of a preset frequency band, and the ratio of the standard deviation to the mean of the power spectral density is used as the background fluctuation index. The peak-finding algorithm from the SciPy library is used to find the width of the trough region in the background sequence as the pulse tolerance. The center frequency of adjacent time periods is extracted using short-time Fourier transform, and the spectral drift rate is calculated using a first-order forward difference algorithm. A multidimensional array is constructed using matrix generation functions from the NumPy library as the initial value matrix for the excitation parameters. The weighted sum of the device safety threshold and the ambient noise floor is set as the initial value for the pulse amplitude, combined with the pulse tolerance to set the initial value for the pulse width, and the initial value for the pulse interval is set according to the reciprocal of the spectral drift rate. Multiple sets of excitation signals with different parameters are generated by an arbitrary waveform generator, amplified by an RF power amplifier, and then radiated to the area under test by a directional antenna. The data acquisition card is then activated to synchronously acquire the echo signal after RF front-end mixing and filtering, as well as harmonic components and electromagnetic disturbance sequences; see reference. Figure 2 As shown in the figure, this figure is the time-domain echo response waveform of the region under test under pulse excitation. The characteristics of the pulse rising edge and falling edge can be seen from the figure.

[0030] The echo signal undergoes extreme value normalization. The gradient calculation function from the NumPy library is used to calculate the maximum derivative of the rising edge of the signal as the slope of the normalization front. The number of sample points required for the signal to fall from the peak to 10% of the baseline level is multiplied by the sampling period to obtain the normalized fallback time. These two values ​​are multiplied together and a constant bias is added to generate the edge constraint. The cross-correlation function from the SciPy library is used to calculate the cross-correlation coefficient sequence between the electromagnetic disturbance sequence and the broadband electromagnetic background sequence. The time delay corresponding to the cross-correlation coefficient decaying to a threshold of 0.1 is found as the disturbance memory length. This disturbance memory length is multiplied by a safety factor within 1 and added to the subsequent pulse interval.

[0031] In one possible implementation, a broadband electromagnetic background sequence and spatial location grid of the area to be measured are acquired, and the background fluctuation index, pulse tolerance, and spectral drift rate of each grid point in a preset frequency band are calculated, including: The broadband electromagnetic background sequence was divided into sub-bands with a bandwidth of 500kHz, and the amplitude standard deviation of each sub-band was extracted as the background fluctuation index. The first derivative of the broadband electromagnetic background sequence is calculated to generate a derivative sequence. The absolute amplitude of the broadband electromagnetic background sequence between the zero crossover points where the signs of the derivatives reverse in two consecutive times is extracted and integrated. The resulting integrated area is used as the pulse tolerance. A fast Fourier transform is performed on the broadband electromagnetic background sequence every 10ms. The center frequency shift value is extracted from the spectrum generated by two consecutive fast Fourier transforms. The center frequency shift value is divided by 10ms to obtain the spectral drift rate of adjacent time periods.

[0032] A wideband antenna and a high-speed RF digitizer with a sampling rate of up to 200MHz were used to acquire a wideband electromagnetic background sequence ranging from 0.1GHz to 3GHz with a sampling window of 50ms. A spatial location grid with sides of 1 meter was then created using high-precision RTK technology. Based on this, a bandpass filter bank or digital down-conversion technology was used to precisely divide the acquired wideband electromagnetic background sequence into multiple sub-bands with a bandwidth of 500kHz. The amplitude standard deviation of each sub-band sequence within the current sampling window was calculated. For example, the calculated amplitude standard deviation of a sub-band with a center frequency of 1.5GHz was 0.015mV, which was used to evaluate the background fluctuation index of each frequency band. The broadband electromagnetic background sequence is processed by performing first-order discrete derivative operations using the forward difference method to generate the corresponding derivative sequence. The zero-crossing point where the positive and negative signs are reversed is accurately located in the derivative sequence. The time interval between two adjacent zero-crossing points is extracted, and the absolute level amplitude within the interval is integrated in the time domain using Simpson's integral rule. For example, the absolute amplitude of a signal interval of 0.5 μs is integrated to obtain an area value of 0.25 mV·μs. This physical quantity accurately represents the pulse capacity of the grid point.

[0033] The timing trigger is activated, and a sample frame of 1024 points is extracted from the background sequence at precise 10ms intervals. A fast Fourier transform is then performed, and the centroid method is used to locate the center frequency shift of two consecutive spectra. If the center frequency is measured to drift from 1.5001GHz to 1.5003GHz within two consecutive 10ms periods, i.e., the shift value is 200kHz, then the shift value is divided by the time interval of 10ms to obtain the spectral drift rate as 20MHz / s.

[0034] In one possible implementation, an initial value matrix for the excitation parameters is constructed, and initial values ​​for the pulse amplitude, duration, and pulse interval are determined, including: The result of multiplying the background undulation index by a constant term is taken as the first element of the main diagonal; The value obtained by dividing the pulse tolerance by the total duration of the broadband electromagnetic background sequence is taken as the second element of the main diagonal; The reciprocal of the spectral drift rate is taken as the third element of the main diagonal; All matrix elements except the main diagonal are set to 0 to construct a 3×3 third-order diagonal matrix as the initial value matrix for the excitation parameters. Multiply the first element of the main diagonal by the first dimension conversion factor to map it as the initial value of the pulse amplitude; multiply the second element of the main diagonal by the second dimension conversion factor to map it as the initial value of the pulse width; and multiply the third element of the main diagonal by the third dimension conversion factor to map it as the initial value of the pulse interval.

[0035] After acquiring the aforementioned feature data, the microprocessor invokes a matrix generation algorithm to multiply the measured background fluctuation index of 0.015mV by the engineering constant 1.5 to obtain 0.0225. This value is used as the first element at position (1,1) on the main diagonal of a 3×3 diagonal matrix. The calculated pulse tolerance of 0.25mV·μs is divided by the total sampling duration of the current broadband electromagnetic background sequence of 50000μs, yielding a ratio of 0.000005, which is used as the second element at position (2,2) on the main diagonal. The reciprocal of the obtained spectral drift rate of 20MHz / s is taken to obtain 0.05 seconds, which is used as the third element at position (3,3) on the main diagonal. During this calculation, the system memory forcibly assigns the remaining six elements in the matrix, excluding positions (1,1), (2,2), and (3,3), to floating-point numbers of 0.0, constructing a third-order diagonal initial value matrix representing the initial electromagnetic state of the environment.

[0036] The controller performs a matrix-to-physical control parameter mapping operation: multiplying the first element 0.0225 by a preset first dimension conversion coefficient of 2000V / mV, the initial value of the pulse amplitude of the solid-state pulse generator is calculated and set to 45V, with the preferred amplitude range set between 30V and 100V to adapt to attenuation environments at different distances; multiplying the second element 0.000005 by a second dimension conversion coefficient. The unit is ns / s. A wideband pulse with an initial pulse width of 1ns is obtained and transmitted. The pulse width parameter is usually finely adjusted within the range of 0.5ns to 5ns. The third element 0.05 is multiplied by the third dimension conversion factor of 1000ms / s to determine the initial pulse interval as 50ms.

[0037] In one possible implementation, calculating the perturbation memory length of the electromagnetic perturbation sequence relative to the broadband electromagnetic background sequence to adjust the subsequent pulse interval includes: Perform discrete cross-correlation operations on the electromagnetic disturbance sequence and the broadband electromagnetic background sequence to generate a one-dimensional array of cross-correlation functions; Extract the position in the one-dimensional array of cross-correlation functions where the absolute value of the amplitude first drops to 10% of the absolute value of the amplitude of the maximum peak, and use the corresponding position as the decay coordinate point; Calculate the index distance between the location coordinates corresponding to the maximum peak and the attenuation coordinates, and multiply the index distance by the sampling period to obtain the perturbation memory length; The subsequent pulse interval is adjusted to be the sum of the perturbation memory length and the preset safety time margin.

[0038] After each excitation shutdown, the high-speed data acquisition card synchronously acquires an electromagnetic disturbance sequence of 10,000 sample points at a fixed sampling rate of 500 MSPS. This disturbance sequence, along with a broadband electromagnetic background reference sequence of equal length under pulseless excitation conditions, is then fed into a DSP processor to perform discrete cross-correlation calculations, generating a one-dimensional array of cross-correlation functions with 19,999 points. The program iterates through this one-dimensional array to search for the global maximum peak value, assuming this peak value is 15.0V. 2 It appears at index 10500; starting from this point, the absolute value of the amplitude of the scan sequence points is extended to the right. When the scan reaches point 10850, the absolute value of the amplitude first decays to 1.5V. 2 That is, the global maximum peak value is 15.0V. 2 If 10% is used, then the 10850th point is precisely marked as the attenuation coordinate point.

[0039] Based on this coordinate information, the microprocessor subtracts the attenuation coordinate point index 10850 from the maximum peak index 10500 to obtain an index distance of 350 sample points. Multiplying these 350 points by the period of 2 ns corresponding to a sampling rate of 500 MSPS, the current perturbation memory length is calculated to be 700 ns. To prevent temporal aliasing and spatial nonlinear intermodulation interference caused by two consecutive strong electromagnetic pulse excitations, the timing control module simply sums the calculated perturbation memory length of 700 ns with a fixed preset safety time margin of 800 ns. The optimal range of this safety time margin is strictly limited to between 500 ns and 1 μs, resulting in an adjusted subsequent pulse interval of 1.5 μs. This backoff mechanism ensures that the residual energy of the previous excitation has been fully extinguished before subsequent pulses are issued in complex electromagnetic environments.

[0040] S2, time-frequency analysis to construct response spectrum, calculate features to filter and lock parameter groups.

[0041] The echo signal is segmented for time-frequency analysis to calculate the energy ratios of the fundamental frequency, harmonics, and fractional frequencies. The segment overlap rate is adjusted in conjunction with the spectral drift rate. A nonlinear response spectrum is constructed using the energy ratios, edge constraints, and perturbation memory length. A mapping relationship between pulse parameters and response characteristics is established based on the nonlinear response spectrum. The curvature offset, phase reversal density, and response hysteresis exponent are calculated. A pulse parameter set that meets the preset conditions is selected as the locked parameter set.

[0042] The spectral drift rate is input to the activation function and mapped to the interval between 0 and 1. The overlap rate is calculated using a base overlap rate of 50%. The short-time Fourier transform function from the SciPy library is then used to perform a piecewise time-frequency transform on the echo signal according to the overlap rate and a Hamming window to obtain the spectral matrix. (See also...) Figure 3 As shown, the fundamental and harmonic energy distribution of the echo signal can intuitively represent the nonlinear response characteristics of the target, clearly demonstrating the energy differences and distribution characteristics of the harmonic and fractional frequency components. The energy value at the fundamental frequency of the excitation signal is extracted from the spectrum matrix as the fundamental frequency energy; the sum of the energy at the second and third harmonics is extracted as the harmonic energy; and the sum of the energy at the 0.5 and 1.5 harmonics is extracted as the fractional frequency energy. The energy ratio is obtained by dividing the sum of the harmonic and fractional frequency energies by the fundamental frequency energy. The energy ratio, along with the edge constraint and perturbation memory length, is concatenated column-wise and input into the tensor processing framework PyTorch. Principal component analysis is used to reduce the dimensionality and extract the first three principal components to construct the nonlinear response spectrum in a three-dimensional coordinate system. A nonlinear mapping model is established using support vector regression, with the impulse parameters as input and the nonlinear response spectrum characteristics as output. The second derivative of the response spectrum curve is calculated using the second-order central difference algorithm from the NumPy library as the curvature offset.

[0043] The instantaneous phase sequence of the echo signal is extracted by calling the Hilbert transform function of the SciPy library. The initial phase difference sequence is obtained by differential calculation with adjacent 1μs window widths. The initial phase difference sequence is subjected to principal value mapping processing with a period of 2π to eliminate spurious 2π truncation jumps caused by range limitation. The number of jumps with an absolute value greater than a preset inversion threshold (such as 0.8π radians) in the statistically processed phase difference sequence is taken as the phase inversion number, and the ratio of the phase inversion number to the total duration of the signal is taken as the phase inversion density.

[0044] The integral area of ​​the response envelope between the end of the pulse excitation and the complete dissipation of the echo energy is calculated as the response hysteresis index. Logical judgment conditions are set: curvature offset greater than a first empirical threshold, phase reversal density higher than a second empirical threshold, and response hysteresis index within a specified interval. Boolean indexes are used to traverse the parameter library, and excitation parameters that satisfy all the above logical judgment conditions are extracted and stored in an independent array as a locked parameter group.

[0045] In one possible implementation, a nonlinear response spectrum is constructed using the energy ratio, edge constraint, and perturbation memory length, including: A three-dimensional Cartesian coordinate system is established using the energy ratio as the X-axis coordinate data, the edge constraint quantity formed by multiplying the normalized front slope by the preset dimension conversion coefficient and adding the normalized back edge fall time as the Y-axis coordinate data, and the perturbation memory length as the Z-axis coordinate data. The feature data of different grid points in the area to be tested are mapped one by one to a three-dimensional Cartesian coordinate system. The mean value of the Z-axis coordinate data with the same X-axis and Y-axis coordinate data is calculated to generate a single-value spatial three-dimensional scatter set. Thin plate spline interpolation algorithm is used to perform meshed surface fitting on a three-dimensional scatter point set in space to generate a continuous three-dimensional spatial surface, which is then used as a nonlinear response spectrum.

[0046] The host computer analysis software extracts the energy ratio of the fundamental frequency to the second harmonic of the echo signal as the X-axis coordinate data; it extracts the previously calculated normalized leading edge slope of 0.8 / ns, multiplies the slope by the set dimensionless conversion coefficient of 1.0, and adds it to the measured normalized trailing edge fall-back time of 2.2ns to obtain the edge constraint value of 3.0 as the Y-axis coordinate data; simultaneously, it maps the determined perturbation memory length of 700ns to the Z-axis coordinate data, thus constructing a three-dimensional Cartesian coordinate system representing multidimensional physical parameters in memory. During the area scanning process, as the acquisition array moves, the feature data pairs (X,Y,Z) generated by thousands of detection grid points are entered into this three-dimensional space one by one; for overlapping singular points with multiple different Z-axis values ​​on the same (X,Y) bottom coordinate due to measurement noise, an arithmetic mean operation is forcibly performed to output a single-valued Z-axis height, thereby generating a high-purity and conflict-free three-dimensional spatial scatter set.

[0047] The data rendering engine calls a thin-plate spline interpolation algorithm based on radial basis functions and sets smoothing parameters. To balance surface stiffness and approximation error, a smooth surface fitting was performed on the spatial scatter set under a high-resolution 100×100 grid, with a value of 0.01. This fitting operation spanned the sparse unsampled region and output a globally continuous and second-differentiable three-dimensional spatial surface map. This surface, with its spatial local bulges and geometric distortions, intuitively and accurately represents the characteristic fingerprint spectrum of the nonlinear response of the RF diode inside the antenna of an environmentally concealed GPS terminal.

[0048] In one possible implementation, the calculation of curvature offset, phase reversal density, and response hysteresis index includes: Calculate the principal curvature values ​​corresponding to each coordinate point on the nonlinear response spectrum surface, subtract the principal curvature values ​​from the fixed curvature 0 of the reference plane, and extract the absolute value of the subtraction result as the curvature offset. The instantaneous phase of the echo signal within adjacent 1μs window widths is extracted and subtracted to obtain the initial phase difference. The number of transitions with an absolute value greater than π radians in the initial phase difference is counted as the phase reversal count. The initial phase difference is processed by principal value mapping with a period of 2π to eliminate spurious 2π truncation transitions caused by range limitation. The phase reversal count is divided by the total duration of the echo signal to obtain the phase reversal density. The peak moment of the echo signal pulse is extracted as the start time, and the moment when the amplitude of the harmonic component decays to 5% of the peak amplitude is extracted as the end time. The end time is subtracted from the start time, and the difference is multiplied by the damping penalty factor to generate the response hysteresis index.

[0049] For the three-dimensional nonlinear response spectrum generated by the aforementioned fitting, the main control computing unit calculates the eigenvalues ​​of the Hessian matrix at discrete coordinate points on the surface one by one, and then obtains the first principal curvature and the second principal curvature; extracts the principal curvature with the larger absolute value, subtracts the principal curvature from the fixed curvature 0 of the reference plane in the ideal targetless state, and extracts the absolute value 0.45 as the curvature offset. The level of this curvature offset is strongly positively correlated with the nonlinear scattering cross-sectional area. In the phase domain feature extraction stage, Hilbert transform is used to extract the instantaneous phase sequence of the baseband echo signal. Adjacent data are extracted with a fixed window width of 1μs and subtracted point by point to obtain the initial phase difference. The principal value mapping algorithm with a period of 2π is called to replace the traditional dewinding algorithm, which strictly limits the phase difference to ±π range, thereby automatically eliminating 2π false jumps caused by boundary truncation. Subsequently, the microprocessor counts the number of real abnormal jumps whose absolute value of the mapped phase difference exceeds the preset inversion threshold. Assuming that 25 such phase inversions are detected in the 50μs echo record, the phase inversion density is calculated to be 0.5 times / microsecond by dividing 25 times by the total duration of 50μs.

[0050] A high-precision clock system records the instant when the leading edge of the echo pulse rises to its highest peak value. =2.0μs as the start time, and simultaneously activate the envelope detector circuit to monitor the second harmonic component of the echo pulse; when the harmonic envelope level first declines and decays to 5% of its own global peak amplitude, this moment is immediately locked as the end time. =6.5μs; the duration difference of 6.5μs-2.0μs is obtained as 4.5μs. Then, this time difference is multiplied by the dimensionless damping penalty factor of 0.8 set by the system to obtain a response hysteresis index of 3.6μs. This index eliminates the instantaneous reflection interference of passive metal objects and exposes the existence of GPS low noise amplifier with capacitive reverse recovery characteristics.

[0051] S3 is the retest calculation index, which is used to output the detection results by fusing features.

[0052] The test was repeated using the locked parameter set; the delay concentration, carrier leakage coefficient and post-pulse residual oscillation index were calculated; and the test results were output by combining the nonlinear response spectrum and the response hysteresis index.

[0053] The locked parameter group in the independent array is reloaded into the arbitrary waveform generator and RF transmission link. A second fixed-point pulse excitation is performed on the same test area, and the re-measured echo signal is recorded synchronously. The arrival time of multiple suspected reflection peaks in the re-measured echo signal is extracted using a threshold method. The variance calculation function of the NumPy library is used to calculate the variance of all arrival times, and the reciprocal of the variance is used as the delay clustering degree. A bandpass filter is constructed using the SciPy library to extract the leakage signal within the transmission frequency band. The ratio of the root mean square value of the leakage signal to the overall root mean square value of the re-measured echo signal is calculated as the carrier leakage coefficient. A small segment of the re-measured echo signal is intercepted after the pulse excitation turn-off moment. The state-space modeling algorithm is used to extract the autoregressive model poles of this segment of signal. The average distance from all poles to the unit circle of the complex plane is calculated as the post-pulse residual oscillation index. The three-dimensional feature coordinates of the nonlinear response spectrum, the response hysteresis exponent, the delay clustering degree, the carrier leakage coefficient, and the post-pulse residual oscillation exponent are concatenated into a seven-dimensional feature vector. This seven-dimensional feature vector is then input into a random forest classifier that has been pre-trained using normal background data and data from GPS locators. The classifier outputs a Boolean value indicating whether a locator exists and a probability value indicating its existence as the detection result.

[0054] In one possible implementation, the calculation of delay clustering, carrier leakage coefficient, and post-pulse residual oscillation index includes: Extract the time delay value of the peak value of the echo signal envelope corresponding to each pulse excitation during the retest period to form a time series, calculate the statistical variance of the time series, and divide 1 by the sum of the statistical variance and the preset positive constant as the time delay clustering degree. The echo baseband signal within 200μs after the retest pulse is turned off is intercepted, and the maximum residual level amplitude of this echo baseband signal is extracted. The maximum residual level amplitude is divided by the full-scale transmission level amplitude under pulse excitation state as the carrier leakage coefficient. The electromagnetic disturbance sequence after the retest pulse cutoff time is smoothed by a sliding window with a step size of 0.1μs to obtain the base signal. The number of fluctuation extreme points in the electromagnetic disturbance sequence whose amplitude exceeds the 10% amplitude boundary of the base signal is counted. The number of fluctuation extreme points is divided by the total number of retest samplings to obtain the residual oscillation index after the pulse.

[0055] Once the detection device locks onto a suspected grid, it rapidly sends out up to 100 consecutive retest pulses. The time difference measurement module extracts the absolute time delay from the rising edge of each of the 100 trigger commands to the highest peak of the corresponding echo envelope with picosecond precision, thus constructing a time series containing 100 floating-point numbers. The values ​​in the time series are typically distributed between 45.2 ns and 45.8 ns. The statistical variance of this series is calculated to be 0.04 ns. 2 Dividing the number 1 by the sum of the variance 0.04 and the preset minimum normal number 0.01 to prevent overflow (i.e., 1 divided by 0.05), the highly sensitive time delay clustering parameter is 20. This extremely high clustering clearly indicates that the target is a spatially fixed electronic device rather than transient RF noise. Locating the moment when the excitation pulse drops to the 0-point of the turn-off threshold, a high-bandwidth oscilloscope is used to capture the pure echo baseband signal within a 200μs time window after this turn-off moment. The maximum residual level amplitude of the baseband signal detected across the entire window is retrieved. This 35mV is directly divided by the full-scale emission peak value of 100V applied to the antenna at the moment of pulse excitation (i.e., 100000mV), resulting in a carrier leakage coefficient of 0.00035.

[0056] In processing the digital sequence of electromagnetic disturbances following the same shutdown moment, a sliding window with a width of 1 μs is set and shifted to the right on the sequence in steps of 0.1 μs. A moving average algorithm is used to generate a base signal representing the low-frequency trend. The original broadband disturbance points are compared one by one with the corresponding base level to accurately count the number of abnormal fluctuation extreme points whose absolute amplitude exceeds the ±10% threshold boundary of the base level. Assuming that 150 exceeding the limit points are detected in a total of 10,000 samples, the residual oscillation index after the pulse is calculated to be 0.015 by dividing 150 by 10,000. The controller integrates the time delay clustering, leakage coefficient, and this oscillation index into the decision matrix and outputs a deterministic detection conclusion.

[0057] This experiment was conducted in a complex broadband electromagnetic environment containing concealed GPS micro-terminals and various passive metallic reflectors. The detection system hardware consisted of a high-speed RF digitizer and a broadband antenna, with a sampling rate set to 500 MSPS and a full-scale emission level of 100V for the pulse generator. The control group used a traditional fixed 50ms pulse interval transmission mechanism and a one-dimensional energy threshold detection algorithm for target identification. The experimental group employed a complete detection scheme, extracting attenuation coordinates through discrete cross-correlation calculations and calculating the perturbation memory length to adjust the pulse interval. Simultaneously, a three-dimensional nonlinear response spectrum was constructed using a thin-plate spline interpolation algorithm, combined with the response hysteresis index, for multi-dimensional feature joint decision-making. (See reference...) Figure 4 As shown in the figure, this figure compares the performance of the traditional solution and the solution of the present invention. The results show that the present invention reduces interference, improves the signal-to-noise ratio and detection accuracy, and suppresses false alarms.

[0058] In a test involving 10,000 consecutive detection cycles, the control group recorded 452 instances of temporal aliasing and spatial nonlinear intermodulation interference, with an average overall echo signal-to-noise ratio of 12.4 dB, a detection accuracy of 76.5% for concealed GPS terminals, and a false alarm rate of 17.8% for passive metal reflections. The experimental group employing the complete scheme, under the same number of detection cycles, reduced the number of temporal aliasing and nonlinear intermodulation interference instances to 0, increased the average overall echo signal-to-noise ratio to 21.6 dB, improved the detection accuracy for nonlinear semiconductor targets to 97.2%, and reduced the false alarm rate to 1.5%.

[0059] Based on the backoff mechanism of adjusting the subsequent pulse interval according to the perturbation memory length, the attenuation characteristics of residual electromagnetic energy in the environment are precisely matched, eliminating the energy superposition and mixing phenomenon of adjacent strong pulses at the physical emission level. In addition, based on the continuous spatial surface constructed by the energy ratio and edge constraint, as well as the fusion calculation of multiple features such as response hysteresis index and time delay clustering, the linear reflection interference of passive linear metal objects is completely eliminated, and the capacitive reverse recovery physical fingerprint unique to the RF amplifier circuit is accurately locked, realizing a leap in the recognition rate of concealed targets in complex backgrounds and comprehensive suppression of false alarm interference.

[0060] An embodiment of the dormant node detection system based on high-energy pulse and harmonic analysis provided by this invention: The dormant node detection system based on high-energy pulse and harmonic analysis includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the aforementioned dormant node detection method based on high-energy pulse and harmonic analysis.

[0061] The dormant node detection system based on high-energy pulse and harmonic analysis also includes other components well known to those skilled in the art, such as communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0062] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

[0063] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for detecting dormant nodes based on high-energy pulse and harmonic analysis, characterized in that, Includes the following steps: Obtain the broadband electromagnetic background sequence and spatial location grid of the area to be tested, calculate the background fluctuation index, pulse tolerance, and spectral drift rate of each grid point in the preset frequency band; construct the initial value matrix of excitation parameters, and determine the initial values ​​of pulse amplitude, pulse width, and pulse interval. Multiple sets of pulses with different parameters are applied to the area to be measured according to the determined pulse amplitude, duration and initial pulse interval. Echo signals, harmonic components and electromagnetic disturbance sequences are collected. The normalized leading edge slope and normalized trailing edge fall time of the echo signals are extracted to generate edge constraint quantities. The disturbance memory length of the electromagnetic disturbance sequence relative to the broadband electromagnetic background sequence is calculated to adjust the subsequent pulse interval. The echo signal is segmented for time-frequency analysis to calculate the energy ratios of the fundamental frequency, harmonics, and fractional frequencies. The segment overlap rate is adjusted in conjunction with the spectral drift rate. The nonlinear response spectrum is constructed using the energy ratios, edge constraints, and perturbation memory length. A mapping relationship between impulse parameters and response characteristics is constructed based on the nonlinear response spectrum; Calculate curvature offset, phase reversal density, and response hysteresis index; select pulse parameter sets that meet preset conditions as locked parameter sets; Retesting was performed using the locked parameter set; Calculate the delay clustering, carrier leakage coefficient, and post-pulse residual oscillation index; The detection results are output by combining the nonlinear response spectrum and the response hysteresis index.

2. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 1, characterized in that, Acquire the broadband electromagnetic background sequence and spatial location grid of the area to be measured, and calculate the background fluctuation index, pulse tolerance, and spectral drift rate of each grid point in the preset frequency band, including: The broadband electromagnetic background sequence was divided into sub-bands with a bandwidth of 500kHz, and the amplitude standard deviation of each sub-band was extracted as the background fluctuation index. The first derivative of the broadband electromagnetic background sequence is calculated to generate a derivative sequence. The absolute amplitude of the broadband electromagnetic background sequence between the zero crossover points where the signs of the derivatives reverse in two consecutive times is extracted and integrated. The resulting integrated area is used as the pulse tolerance. A fast Fourier transform is performed on the broadband electromagnetic background sequence every 10ms. The center frequency shift value is extracted from the spectrum generated by two consecutive fast Fourier transforms. The center frequency shift value is divided by 10ms to obtain the spectral drift rate of adjacent time periods.

3. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 1, characterized in that, Construct an initial value matrix for the excitation parameters, and determine the initial values ​​for the pulse amplitude, pulse width, and pulse interval, including: The result of multiplying the background undulation index by a constant term is taken as the first element of the main diagonal; The value obtained by dividing the pulse tolerance by the total duration of the broadband electromagnetic background sequence is taken as the second element of the main diagonal; The reciprocal of the spectral drift rate is taken as the third element of the main diagonal; All matrix elements except the main diagonal are set to 0 to construct a 3×3 third-order diagonal matrix as the initial value matrix for the excitation parameters. Multiply the first element of the main diagonal by the first dimension conversion factor to map it as the initial value of the pulse amplitude; multiply the second element of the main diagonal by the second dimension conversion factor to map it as the initial value of the pulse width; and multiply the third element of the main diagonal by the third dimension conversion factor to map it as the initial value of the pulse interval.

4. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 1, characterized in that, Calculating the perturbation memory length of the electromagnetic perturbation sequence relative to the broadband electromagnetic background sequence to adjust the subsequent pulse interval includes: Perform discrete cross-correlation operations on the electromagnetic disturbance sequence and the broadband electromagnetic background sequence to generate a one-dimensional array of cross-correlation functions; Extract the position in the one-dimensional array of cross-correlation functions where the absolute value of the amplitude first drops to 10% of the absolute value of the amplitude of the maximum peak, and use the corresponding position as the decay coordinate point; Calculate the index distance between the location coordinates corresponding to the maximum peak and the attenuation coordinates, and multiply the index distance by the sampling period to obtain the perturbation memory length; The subsequent pulse interval is adjusted to be the sum of the perturbation memory length and the preset safety time margin.

5. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 1, characterized in that, The nonlinear response spectrum is constructed using energy ratios, edge constraints, and perturbation memory lengths, including: A three-dimensional Cartesian coordinate system is established using the energy ratio as the X-axis coordinate data, the edge constraint quantity formed by multiplying the normalized front slope by the preset dimension conversion coefficient and adding the normalized back edge fall time as the Y-axis coordinate data, and the perturbation memory length as the Z-axis coordinate data. The feature data of different grid points in the area to be tested are mapped one by one to a three-dimensional Cartesian coordinate system. The mean value of the Z-axis coordinate data with the same X-axis and Y-axis coordinate data is calculated to generate a single-value spatial three-dimensional scatter set. Thin plate spline interpolation algorithm is used to perform meshed surface fitting on a three-dimensional scatter point set in space to generate a continuous three-dimensional spatial surface, which is then used as a nonlinear response spectrum.

6. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 1, characterized in that, Calculate curvature offset, phase reversal density, and response hysteresis index, including: Calculate the principal curvature values ​​corresponding to each coordinate point on the nonlinear response spectrum surface, subtract the principal curvature values ​​from the fixed curvature 0 of the reference plane, and extract the absolute value of the subtraction result as the curvature offset. The instantaneous phase of the echo signal within adjacent 1μs window widths is extracted and subtracted to obtain the initial phase difference. The number of transitions with an absolute value greater than π radians in the initial phase difference is counted as the phase reversal count. The initial phase difference is processed by principal value mapping with a period of 2π to eliminate spurious 2π truncation transitions caused by range limitation. The phase reversal count is divided by the total duration of the echo signal to obtain the phase reversal density. The peak moment of the echo signal pulse is extracted as the start time, and the moment when the amplitude of the harmonic component decays to 5% of the peak amplitude is extracted as the end time. The end time is subtracted from the start time, and the difference is multiplied by the damping penalty factor to generate the response hysteresis index.

7. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 1, characterized in that, Calculate the delay clustering, carrier leakage coefficient, and post-pulse residual oscillation index, including: Extract the time delay value of the peak value of the echo signal envelope corresponding to each pulse excitation during the retest period to form a time series, calculate the statistical variance of the time series, and divide 1 by the sum of the statistical variance and the preset positive constant as the time delay clustering degree. The echo baseband signal within 200μs after the retest pulse is turned off is intercepted, and the maximum residual level amplitude of this echo baseband signal is extracted. The maximum residual level amplitude is divided by the full-scale transmission level amplitude under pulse excitation state as the carrier leakage coefficient. The electromagnetic disturbance sequence after the retest pulse cutoff time is smoothed by a sliding window with a step size of 0.1μs to obtain the base signal. The number of fluctuation extreme points in the electromagnetic disturbance sequence whose amplitude exceeds the 10% amplitude boundary of the base signal is counted. The number of fluctuation extreme points is divided by the total number of retest samplings to obtain the residual oscillation index after the pulse.

8. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 2, characterized in that, A bandpass filter bank is used to divide the broadband electromagnetic background sequence into multiple sub-bands.

9. The method for detecting dormant nodes based on high-energy pulse and harmonic analysis according to claim 2, characterized in that, A derivative sequence is generated by calculating the first derivative of a broadband electromagnetic background sequence using the forward difference method.

10. A dormant node detection system based on high-energy pulse and harmonic analysis, characterized in that, It includes a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the dormant node detection method based on high-energy pulse and harmonic analysis as described in any one of claims 1-9 is implemented.

Images