A method for individual identification of moving radiation sources
By cleaning and processing data, the problem of unstable features of moving radiation sources was solved, stable features were extracted, and the accuracy of individual radiation source identification was improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NO 8511 RES INST OF CASIC
- Filing Date
- 2024-05-14
- Publication Date
- 2026-06-12
AI Technical Summary
Traditional methods struggle to reliably extract features from moving radiation sources, resulting in low identification accuracy. This is especially true in complex electromagnetic environments where motion introduces Doppler effects and amplitude modulation, distorting the signal and preventing convergence of traditional signal intermediate frequency and envelope features.
By performing data cleaning, sample screening, alignment, removal of carrier frequency and phase noise, normalization, removal of outliers, and equalization, stable radiation source features are extracted, enhancing convergence.
It improves the accuracy of individual radiation source identification, has strong feature stability and convergence, and the algorithm is simple and easy to implement.
Smart Images
Figure CN118535966B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of electronic signal electromagnetic environment monitoring, specifically relating to an individual identification method for moving radiation sources, which is applied to an individual identification system for moving radiation sources. Background Technology
[0002] Currently, the electronic systems of launch platforms are diverse, with multiple types of electronic equipment simultaneously radiating electromagnetic signals, significantly increasing the complexity of electromagnetic environment monitoring. Different types of radiation sources can be identified and sorted using conventional signal parameter estimation. However, for sources of the same type, since the radiated signals are identical, both platforms of the same and different models may be equipped with the same electronic equipment, making traditional parameter estimation insufficient to distinguish between them. Therefore, it is necessary to identify individual radiation sources based on their signal characteristics to differentiate between different sources.
[0003] Traditional methods for identifying individual radiation sources utilize the intermediate frequency (IF) and amplitude envelope of the received radiation source signal as input features. This method is sensitive to the signal-to-noise ratio (SNR) of the received signal, and variations in the signal transmission channel cause amplitude distortion. For moving targets, the distortion is even more severe due to the Doppler effect and amplitude modulation introduced by motion, leading to a lack of convergence and poor stability in the extracted IF and envelope features. Directly using conventional IF and amplitude envelope information as input features results in a significant decrease in recognition accuracy. Therefore, to identify moving targets, it is necessary to extract effective features that can stably characterize moving radiation sources.
[0004] In summary, in order to identify moving radiation source targets, it is urgent to extract a feature that can stably characterize moving radiation sources. Summary of the Invention
[0005] This invention proposes an individual identification method for moving radiation sources, which solves the problems of instability and non-convergence of feature extraction in traditional mid-frequency and envelope characterization of moving radiation sources. This invention utilizes intra-pulse information of the radiation source to extract target features, attempts to remove unstable factors caused by target motion, enhances the convergence of radiation source features, and thus improves the accuracy of target individual identification.
[0006] The technical solution for achieving the present invention is: an individual identification method for moving radiation sources, comprising the following steps:
[0007] Step 1: Clean the input raw signal, filter the samples, remove low-quality samples, and filter signals with a signal-to-noise ratio higher than 20 to improve the sample quality and obtain radiation source signals with interference removed.
[0008] Step 2: Perform sample alignment on the radiation source signals after removing interference to ensure that the start time of the input samples is aligned, improve the convergence of target features, and obtain aligned radiation source signals.
[0009] Step 3: Remove the carrier frequency of the aligned radiation source signal, then perform amplitude normalization, and then extract the phase information of the sample after removing the carrier frequency to obtain the phase information of each sample.
[0010] Step 4: Remove the Doppler displacement component caused by the target motion from the sample phase information to obtain the original characteristic curve of the radiation source.
[0011] Step 5: Remove the constant component of the original characteristic curve of the radiation source, normalize the characteristic amplitude, and obtain the normalized phase.
[0012] Step 6: Remove a large number of outlier samples with phase disorder to further improve sample quality and obtain the radiation source characteristic curve.
[0013] Step 7: Equalize the obtained samples to reduce the influence of phase noise, further improve feature convergence, and obtain the final feature curve.
[0014] Step 8: Perform individual recognition training and identification on the final feature curve to identify individual radiation sources.
[0015] Compared with the prior art, the significant advantages of this invention are:
[0016] (1) The extracted radiation source individual characteristics are stable and have strong convergence.
[0017] (2) The algorithm is simple in principle, has a small amount of computation, and is easy to implement in engineering.
[0018] (3) It has a high accuracy rate in individual identification after being used for individual identification. Attached Figure Description
[0019] Figure 1 This is the data processing flow of the present invention.
[0020] Figure 2 This is an example of the amplitude spectrum of nine samples of the same signal in this invention.
[0021] Figure 3 Example of the effect after removing samples with low signal-to-noise ratio.
[0022] Figure 4 This is an example of the sample effect after removing superimposed interference signals according to the present invention.
[0023] Figure 5 Example of amplitude spectrum of original sample data.
[0024] Figure 6This is an example of the amplitude spectrum of sample data after removing the carrier frequency according to the present invention.
[0025] Figure 7 This is an example of the phase spectrum of the sample signal after removing the carrier frequency according to the present invention.
[0026] Figure 8 This is the radiation source signal receiving model of the present invention.
[0027] Figure 9 This is an example of the phase spectrum of sample data after delinearization according to the present invention.
[0028] Figure 10 This is an example of the phase spectrum of sample data after removing constant terms and normalizing according to the present invention.
[0029] Figure 11 The effect of superimposing the characteristic curves of 225 samples from the same radiation source.
[0030] Figure 12 This invention demonstrates the effect of removing outlier characteristic curves from a large number of samples from the same radiation source.
[0031] Figure 13 This is the effect of superimposing multiple feature curves after the equalization processing of the present invention.
[0032] Figure 14 This is a comparison of the superimposed feature curves of two different individuals in this invention. Detailed Implementation
[0033] 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 a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0034] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.
[0035] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly and specifically defined.
[0036] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixing," etc., should be interpreted broadly. For example, "fixing" can mean a fixed connection, a detachable connection, or an integral part; "connection" can mean a mechanical connection or an electrical connection. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.
[0037] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible to those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.
[0038] The following section will further introduce the specific implementation method, as well as the technical difficulties and inventive points of this invention, using this design example as an example.
[0039] Combination Figure 1 The present invention provides a method for individual identification of moving radiation sources, comprising the following steps:
[0040] Step 1: Clean the input raw signal, filter the samples, remove low-quality samples, and filter signals with a signal-to-noise ratio higher than 20 to improve the sample quality and obtain radiation source signals with interference removed.
[0041] In step 1, the specific steps are as follows:
[0042] First, due to differences in the transmission distance and angle of the target signal, the amplitude of signals from different batches of the same target varies significantly, resulting in fluctuations in the signal-to-noise ratio (SNR). Consequently, the SNR of the received signal samples varies considerably. Figure 2 As shown, taking a small number of samples as an example, the amplitude fluctuations of the 9 samples are relatively large, with the signal-to-noise ratio (SNR) of sample 9 being significantly lower than that of the other samples. Therefore, it is necessary to perform SNR judgment on the input signal samples, discarding signals with an SNR less than 20 and retaining signals with high SNR, such as... Figure 3 As shown.
[0043] The radiation source signal s1(t) received by the system can be expressed as:
[0044]
[0045] Where a(t) is the envelope of the received signal including noise, f0 is the carrier frequency, φ(t) is the phase including noise, and s j (t) represents the externally superimposed interference signal, j represents the imaginary part, T represents the signal duration, and t represents the sampling time.
[0046] Then, due to the presence of a complex electromagnetic environment, a large number of received samples are often superimposed with electromagnetic signals from other targets. j (t), such as Figure 2 As shown, sample 7 is superimposed with other pulse signals. This signal superposition disrupts the stability of the target sample and causes distortion in feature extraction. Therefore, it is necessary to judge the superposition of the input signal and remove samples with superimposed interference signals. The effect after removal is as follows. Figure 4 As shown.
[0047] The interference-free radiation source signal s2(t) can be expressed as:
[0048]
[0049] Step 2: Perform sample alignment on the radiation source signals after removing interference to ensure that the start time of the input samples is aligned, improve the convergence of target features, and obtain aligned radiation source signals.
[0050] The start time of the input samples (i.e., the radiation source signals after interference has been removed) is aligned. In order to achieve feature superposition of different batches of sample signals, the start and end times of different samples need to be kept consistent, so as to lay the foundation for the accumulation of features of subsequent samples. Figure 4 This is for aligning the starting position.
[0051] Step 3: Due to frequency conversion during transmission modulation and reception processing, the original sample data contains fixed carrier frequency information. This fixed carrier frequency, related to the transmitter and receiver frequency conversion processing, causes changes in signal amplitude and phase, preventing the accumulation of sample features. Figure 5 As shown, therefore, it is necessary to remove the sample carrier frequency of the aligned radiation source signal, followed by amplitude normalization. The result after processing is as follows. Figure 6 As shown, the phase information of the samples after removing the carrier frequency is extracted to obtain the phase information of each sample.
[0052] The radiation source signal s3(t) after removing the carrier frequency and normalizing the amplitude can be expressed as:
[0053]
[0054] In the formula, A0 is the mean of the envelope.
[0055] from Figure 6 It can be seen that, despite the removal of the carrier frequency, the amplitude variation trends of different samples differ greatly due to factors such as noise, exhibiting poor regularity and failing to converge.
[0056] Extract the phase information of the samples after removing the carrier frequency to obtain the phase information of each sample, such as... Figure 7 As shown.
[0057] Step 4: Remove the Doppler displacement component caused by the target motion from the phase information of the sample (i.e., remove the carrier frequency and normalized amplitude signal) to obtain the original characteristic curve of the radiation source.
[0058] exist Figure 8 In the model, the phase φ1(t) of the radiation source received by the moving target can be expressed as:
[0059]
[0060] λ is the signal wavelength, r(t) is the instantaneous distance between the radiation source and the receiving device, φ0(t) is the phase modulation phase within the signal pulse, and n φ (t) represents phase noise, φ f (t) represents the phase characteristic information of the individual radiation source, φ a (t) represents the initial phase, for Figure 8 In the geometric model, this distance can be expressed as:
[0061]
[0062] The target's flight altitude is h, and its flight speed is V. a The azimuth coordinates of the receiving station are y0, the shortest distance from the receiving station to the aircraft's trajectory is R0, and the instantaneous slant range from the aircraft to the target is r(t), where t is the sampling time. Since R0 >> (V a Applying Fresnel approximation to the above equation, we obtain t-y0),
[0063]
[0064] Then, the instantaneous phase φ2(t) of the received signal can be expressed as:
[0065]
[0066] because The effect of the quadratic term on the phase can be ignored, and the instantaneous phase can be approximated as φ3(t):
[0067]
[0068] As can be seen from the above formula, apart from the internal modulation phase and phase noise, the signal phase is mainly related to the position and velocity of the moving target, and mainly consists of a constant term and a first-order term in the phase. For a moving target, its position and velocity are constantly changing, and this change is detrimental to the accumulation of phase features and needs to be removed.
[0069] The instantaneous phase after removing the first term, i.e., the original characteristic curve φ4(t) of the radiation source, can be expressed as:
[0070]
[0071] In the formula, φ0'(t) represents the intra-pulse modulation information after removing the first-order term, and n' φ (t) represents the phase noise of the first-order term, φ' f (t) represents the individual phase characteristic information after removing the first-order terms. Here, φ0'(t) is related to the intra-pulse modulation phase of the emitted signal from the radiation source and is the same for all individuals. φ (t) represents phase random noise, which is related to the radiation source's transmission channel, electromagnetic wave transmission path, and receiver channel, causing random fluctuations in the extracted characteristic curve φ(t) and broadening the characteristic curve's dispersion range. This is due to the individual phase characteristics φ of the radiation source. f (t) typically represents high-frequency information, hidden within phase noise; its first-order term is negligible. Therefore, φ' f (t) can still characterize the individual features of the radiation source.
[0072] from Figure 7 It can be seen that the original phase exhibits a clear linear variation trend. This fixed linear phase is mainly caused by the Doppler frequency shift resulting from the target's motion. It changes with the target's velocity, position, and relative motion direction, which is detrimental to the accumulation of target signal features and needs to be removed. The individual features of the target signal are mainly found in the higher-order terms of the phase; removing the lower-order terms will not erase the feature information of the sample.
[0073] The characteristic curve of the sample signal is obtained after linear phase removal, and the effect is as follows: Figure 9 As shown in the figure, the characteristic trends of each sample are the same and have convergence characteristics. The messy parts at both ends of the characteristic curve in the figure are the phase of noise, which are randomly distributed and do not have convergence characteristics.
[0074] Step 5: Initial values φ for different samples of the original characteristic curve of the radiation source a (t) are different, and φ4(t) is a constant term. Since the amplitude range of φ4(t) is related to the location of the radiation source, it is necessary to remove the constant term of φ4(t) and normalize the feature curves with different ranges of variation to obtain the normalized phase.
[0075] The instantaneous phase after removing the constant term and normalizing, i.e., the normalized phase φ5(t), can be expressed as:
[0076]
[0077] In the formula, A φ This represents the absolute value of the variation range of the phase characteristic curve.
[0078] The effects of removing constant terms and phase normalization are as follows: Figure 10 As shown in the figure, the characteristic trends of each sample are the same, and the sample characteristics converge significantly.
[0079] Step 6: Remove the outlier samples with disordered phases from a large number of normalized phase φ5(t) samples to further improve the sample quality and obtain the radiation source characteristic curve.
[0080] After removing the constant term and normalizing the k-th sample, the radiation source characteristic curve can be expressed as follows:
[0081]
[0082] In the formula, "—" indicates the phase components after removing the constant term and the first-order term and normalizing. This indicates the interference phase caused by the superposition of some samples.
[0083] For a large number of samples, some sample pulses contain superimposed signals with amplitudes comparable to the original signals. These signals are difficult to remove based solely on time-domain amplitude variations, leading to distorted extracted sample features. It is necessary to remove outlier samples with disordered feature curves to further improve the quality of sample features.
[0084] The original characteristic curves of a large number of samples from the same radiation source are as follows: Figure 11 As shown in the figure, some feature curves are obviously distorted. This phenomenon is caused by the superposition of other pulse signals, which has a significant impact on feature extraction.
[0085] After removing singular characteristic curves, as shown Figure 12 As shown in the figure, the characteristic curves of the remaining radiation sources after removal are smooth and the curve trends are basically the same.
[0086] Step 7: Reduce phase noise Characteristic curve of radiation source The impact. This results in a wide amplitude distribution range for the characteristic curve. To reduce its impact, it is necessary to perform equalization processing on the obtained samples to further improve feature convergence and obtain the final characteristic curve. During the equalization process, due to the different amplitude distributions of samples from the same radiation source... Since the value is fixed, it inherently possesses convergence properties; therefore, equilibrium processing further improves its convergence. The final characteristic curve is as follows: Figure 13 As shown in the figure, after sample equalization, the convergence of the final feature curve is significantly improved, and the distribution is more compact.
[0087] Step 8: Train and infer the final feature curves. For example, compare the feature curves of two individuals from different radiation sources. Figure 14As shown in the figure, the two colored curves are the characteristic curves of two targets. It can be seen that there is a difference between the two characteristic curves. This difference is the key to distinguishing different individual radiation sources. It can characterize different radiation source individuals and can be identified by individual identification software, thereby identifying individual radiation sources.
[0088] Comparative results of the extracted features from different radiation sources from measured data demonstrate that the proposed method for identifying individual moving radiation sources is feasible. It effectively removes the influence of motion on radiation source features, and the proposed features exhibit convergence characteristics, enabling the characterization of individual moving radiation sources and enhancing individual radiation source identification capabilities. This technology can be applied to electromagnetic environment monitoring systems for identifying individual radiation sources, showing broad application prospects.
Claims
1. A method for individual identification of moving radiation sources, characterized in that, Includes the following steps: Step 1: Clean the input raw signal, filter the samples, remove low-quality samples, and filter signals with a signal-to-noise ratio higher than 20 to improve the sample quality and obtain the radiation source signal with interference removed. Step 2: Perform sample alignment on the interference-removed radiation source signals to ensure that the start time of the input samples is aligned, improve the convergence of target features, and obtain aligned radiation source signals; Step 3: Remove the carrier frequency of the aligned radiation source signal, then perform amplitude normalization, and then extract the phase information of the sample after removing the carrier frequency to obtain the phase information of each sample. Step 4: Remove the Doppler displacement component caused by the target motion from the sample phase information to obtain the original characteristic curve of the radiation source; Step 5: Remove the constant component of the original characteristic curve of the radiation source, normalize the characteristic amplitude, and obtain the normalized phase; Step 6: Remove a large number of outlier samples with phase disorder to further improve sample quality and obtain the radiation source characteristic curve; Step 7: Equalize the obtained samples to reduce the impact of phase noise, further improve feature convergence, and obtain the final feature curve; Step 8: Perform individual recognition training and identification on the final feature curve to identify individual radiation sources.
2. The method for individual identification of moving radiation sources according to claim 1, characterized in that: In step 1, the input raw signal is cleaned, samples are screened to remove low-quality samples, and signals with a signal-to-noise ratio higher than 20 are selected to improve sample quality, resulting in interference-free radiation source signals. The specific steps are as follows: After sample screening and data cleaning of the input signal, the radiation source signal s1(t) received by the system is represented as: Where a(t) is the envelope of the received signal including noise, f0 is the carrier frequency, φ(t) is the phase including noise, and s j (t) represents the externally superimposed interference signal, j represents the imaginary part, T represents the signal duration, and t represents the sampling time; The interference-free radiation source signal sample is represented as follows:
3. The method for individual identification of moving radiation sources according to claim 2, characterized in that: In step 3, the carrier frequency of the aligned radiation source signal is removed, followed by amplitude normalization, and then the phase information of the sample after removing the carrier frequency is extracted to obtain the phase information of each sample, as follows: The radiation source signal s3(t) after removing the carrier frequency and normalizing the amplitude is expressed as: In the formula, A0 is the mean of the envelope.
4. The method for individual identification of moving radiation sources according to claim 3, characterized in that: In step 4, the Doppler displacement component caused by the target motion is removed from the sample phase information to obtain the original characteristic curve of the radiation source, as follows: The phase φ1(t) of the radiation source received by the moving target is expressed as: λ is the signal wavelength, r(t) is the instantaneous distance between the radiation source and the receiving device, φ0(t) is the phase modulation phase within the signal pulse, and n φ (t) represents phase noise, φ f (t) represents the phase characteristic information of the individual radiation source, φ a (t) represents the initial phase; r(t) is represented as: The target's flight altitude is h, and its flight speed is V. a The azimuth coordinates of the receiving station are y0, and the shortest distance from the receiving station to the aircraft's flight path is R0. Since R0 >> (V a Applying Fresnel approximation to the above equation, we obtain t-y0), Then, the instantaneous phase φ2(t) of the received signal is expressed as: because The effect of the quadratic term on the phase is negligible, and the instantaneous phase is approximated as φ3(t): The instantaneous phase after removing the first term, i.e., the original characteristic curve φ4(t) of the radiation source, can be expressed as: In the formula, φ0'(t) represents the intra-pulse modulation information after removing the first-order term, and n' φ (t) represents the phase noise of the first-order term, φ' f (t) represents the individual phase characteristic information after removing the first term; where φ0'(t) is related to the intra-pulse modulation phase of the emitted signal from the radiation source and is the same for all individuals; n' φ (t) represents phase random noise, which is related to the radiation source's transmission channel, electromagnetic wave transmission path, and receiver channel, causing random fluctuations in the extracted characteristic curve φ(t) and broadening the characteristic curve's dispersion range; due to the individual phase characteristics φ of the radiation source... f (t) typically represents high-frequency information, hidden within phase noise; its first-order term is negligible. Therefore, φ' f (t) can still characterize the individual features of the radiation source.
5. The method for individual identification of moving radiation sources according to claim 4, characterized in that: In step 5, the constant component of the original characteristic curve of the radiation source is removed, and the characteristic amplitude is normalized to obtain the normalized phase, as follows: The instantaneous phase φ5(t) after removing the constant term and normalizing is expressed as: In the formula, A φ This represents the absolute value of the variation range of the phase characteristic curve.
6. The method for individual identification of moving radiation sources according to claim 5, characterized in that: In step 6, outlier samples with phase errors are removed from a large number of samples to further improve sample quality and obtain the radiation source characteristic curve, as follows: The k-th sample, after removing the constant term and normalizing, is represented as follows: In the formula, "—" indicates the phase components after removing the constant term and the first-order term and normalizing. This indicates the interference phase caused by the superposition of some samples.
7. The method for individual identification of moving radiation sources according to claim 6, characterized in that: In step 7, the obtained multiple samples are equalized to reduce the influence of phase noise and further improve feature convergence, resulting in the final feature curve, as follows: Reduce phase noise Effect on characteristic curve; This results in a wide amplitude distribution range for the characteristic curve. To reduce its impact, it is necessary to perform equalization processing on the obtained multiple samples to further improve the feature convergence. During the equalization process, due to the different samples from the same radiation source... Since it is a fixed value, it has inherent convergence properties. Therefore, the equalization process will further improve its convergence, thus obtaining the final characteristic curve.