Method for testing multi-channel broadband transient electromagnetic radiation emission signals

Abstract

This invention discloses a multi-channel broadband transient electromagnetic radiation emission signal testing method, belonging to the technical field of electromagnetic signal processing methods. The method includes the following steps: First, the signal is subjected to a second-order Gaussian smoothing filter. Then, using the average correlation coefficient (ACC) as an evaluation index, the signals of different channels are decomposed according to the same segmentation boundary. The ACC of the signals at the same segmentation boundary in different channels is calculated, and the optimal decomposition level K and segmentation boundary are determined by continuously searching the ACC. Nonlinear fitting and cancellation are performed on the decomposed multi-components, and the reconstructed signal is the true EUT radiated emission signal. This method integrates signal decomposition and multi-channel adaptive interference cancellation, avoiding the computational complexity and mode aliasing problems of existing decomposition methods, and achieving accurate extraction of EUT signals in complex electromagnetic environments.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic signal processing methods, and in particular to a multi-channel broadband transient electromagnetic radiation emission signal testing method. Background Technology

[0002] To ensure the safe operation of large mobile equipment and keep radiated emissions within acceptable limits, necessary testing and evaluation are required. However, due to the limited space in anechoic chambers, testing the electromagnetic radiation emissions of large mobile equipment presents a significant engineering challenge. Adaptive noise cancellation virtual anechoic chamber technology offers an effective solution to this problem, but suppressing background noise during on-site electromagnetic radiation testing remains crucial.

[0003] Domestic and international scholars have conducted extensive research to explore how to extract the true radiated emission characteristics of Equipment Under Test (EUT) in complex electromagnetic environments. Cakir et al. proposed a background interference cancellation technique for low-frequency electromagnetic radiation testing, which preserves phase information and eliminates background interference by continuously recording and processing oscilloscope data. Awan et al. proposed an open-area electromagnetic radiation time-domain measurement technique and tested it in a laboratory environment. This technique simultaneously captures multi-channel signals from interference sources and the EUT through different antenna ports. It uses an adaptive filter based on stochastic gradients to process the continuous signal stream at a high sampling rate to eliminate interference from electronic sources. It is used to measure radiated emissions and electromagnetic compatibility, conforms to CISPR standards, eliminates the need for an anechoic chamber, shortens measurement time, and allows for rapid pre-compliance electromagnetic compatibility testing. Zhou Zhongyuan et al. proposed a method combining array signal processing and adaptive noise reduction to suppress environmental interference in in-situ measurements. This method extracts signal direction and category through blind source separation and sparse reconstruction to eliminate coherent interference. Zero-well techniques are used to eliminate large-amplitude interference and attenuate and ignore smaller-amplitude interference. Adaptive noise reduction methods further suppress other related interferences. For transient signals, overlapping windowing is used to avoid spectrum loss. Adaptive noise cancellation is achieved by cyclically expanding the signal frame, thus suppressing multi-source interference with fewer array elements.

[0004] Signal decomposition is used to break down complex signals into simpler, more easily analyzed components. The main methods of signal decomposition include variational mode decomposition, empirical mode decomposition, local mean decomposition, and wavelet decomposition. However, these methods have certain drawbacks: computational complexity, mode aliasing, and difficulty in determining the wavelet basis. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method that is simple to calculate and can accurately extract EUT signals in complex electromagnetic environments.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a multi-channel broadband transient electromagnetic radiation emission signal testing method, comprising the following steps:

[0007] Using a multi-electrode antenna as the signal receiving device for electromagnetic radiation emission testing, one antenna serves as the main input antenna, one antenna serves as the main reference antenna, and the remaining antennas serve as auxiliary reference antennas.

[0008] The signal acquired by the main input antenna is used as the main channel signal, the signal acquired by the main reference antenna is used as the main reference channel signal, and the signal acquired by the auxiliary reference antenna is used as the auxiliary reference channel signal. The main reference channel signal and the auxiliary reference channel signal are fused into a single reference channel signal.

[0009] The fused reference channel signal and the main channel signal are subjected to a second Gaussian smoothing filter. The improved empirical Fourier decomposition method is used to decompose the two signals according to the same segmentation boundary. The average correlation coefficient ACC of the signals with the same segmentation boundary in different channels is calculated. The optimal decomposition level K and segmentation boundary {Ω} are determined by continuously searching for the maximum average correlation coefficient ACC.

[0010] The decomposed IMFs are subjected to nonlinear fitting using a BP neural network to make the signal amplitudes of the corresponding IMFs in the two signals tend to be the same. Then, a multimodal signal cancellation system is constructed to obtain the true radiated emission signal of the device under test (EUT).

[0011] The beneficial effects of adopting the above technical solution are as follows: the method integrates signal decomposition and multi-channel adaptive interference cancellation, avoiding the problems of computational complexity and mode aliasing in existing decomposition methods, and achieving accurate extraction of EUT signals in complex electromagnetic environments. Simulation and experimental results show that the correlation between the obtained EUT radiated emission signal and the real EUT radiated emission signal can reach 97%, and the power error is stable below 1dB. This method is applicable to broadband transient electromagnetic radiated emission testing and research in complex environments. Attached Figure Description

[0012] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0013] Figure 1 This is the main flowchart of the method described in the embodiments of the present invention;

[0014] Figure 2 This is a flowchart of the SGS-ACCI-EFD iterative process in the method described in the embodiments of the present invention;

[0015] Figure 3 This is a block diagram illustrating the principle of multi-channel background interference cancellation in the method described in this embodiment of the invention;

[0016] Figure 4a This is a simulation test signal diagram of environmental interference in an embodiment of the present invention;

[0017] Figure 4b This is a simulation test signal diagram of EUT+ environmental interference in an embodiment of the present invention;

[0018] Figure 5a This is a T-EFD signal decomposition performance diagram in an embodiment of the present invention;

[0019] Figure 5b This is a diagram illustrating the AEDF signal decomposition performance in an embodiment of the present invention.

[0020] Figure 5c This is a graph showing the ASEI-VMD signal decomposition performance in an embodiment of the present invention;

[0021] Figure 5d This is a performance diagram of GA-VMD-SVD signal decomposition in an embodiment of the present invention;

[0022] Figure 5e This is a performance diagram of EAEFD signal decomposition in an embodiment of the present invention;

[0023] Figure 5f This is a performance diagram of GS-ACCI-EFD and the desired radiated emission signal decomposition in an embodiment of the present invention;

[0024] Figure 6a This is a simulation diagram of the interference signal in an embodiment of the present invention;

[0025] Figure 6b This is a simulation diagram of the desired + interference signal in an embodiment of the present invention;

[0026] Figure 7a This is a simulation diagram of the canceled signal and the actual radiated emission signal in an embodiment of the present invention;

[0027] Figure 7b This is a correlation diagram between the canceled signal and the actual radiated emission signal in an embodiment of the present invention;

[0028] Figure 8 These are photos of the test site;

[0029] Figure 9a This is a diagram of the received signal under environmental interference in an embodiment of the present invention;

[0030] Figure 9b This is a diagram of the received signal of EUT + environmental interference in an embodiment of the present invention;

[0031] Figure 10a This is a comparison diagram of canceled output power in an embodiment of the present invention;

[0032] Figure 10bThis is a comparison chart of the correlation between the canceled outputs in an embodiment of the present invention;

[0033] Figure 11a This is a comparison diagram of received signals under environmental interference in an embodiment of the present invention;

[0034] Figure 11b This is a comparison diagram of the received signals of EUT + environmental interference in an embodiment of the present invention;

[0035] Figure 12a This is a comparison diagram of the canceled output intensity in an embodiment of the present invention;

[0036] Figure 12b This is a comparison diagram of canceled output power in an embodiment of the present invention;

[0037] Figure 13 This is a comparison chart of the correlation of signals after cancellation by different decomposition methods in the embodiments of the present invention. Detailed Implementation

[0038] 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 the embodiments. 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.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0040] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for testing multi-channel broadband transient electromagnetic radiation emission signals, comprising the following steps:

[0041] S1: Use multiple antennas as signal receiving devices for electromagnetic radiation emission testing. One antenna is used as the main input antenna, one antenna is used as the main reference antenna, and the remaining antennas are used as auxiliary reference antennas.

[0042] S2: The signal collected by the main input antenna is used as the main channel signal, the signal collected by the main reference antenna is used as the main reference channel signal, and the signal collected by the auxiliary reference antenna is used as the auxiliary reference channel signal. The main reference channel signal and the auxiliary reference channel signal are fused into a single reference channel signal.

[0043] S3: Perform a second Gaussian smoothing filter on the fused reference channel signal and the main channel signal. Use the improved empirical Fourier decomposition method to decompose the two signals according to the same segmentation boundary. Calculate the average correlation coefficient ACC of the signals with the same segmentation boundary in different channels. Determine the optimal decomposition level K and segmentation boundary {Ω} by continuously searching for the maximum average correlation coefficient ACC.

[0044] S4: Perform nonlinear fitting of each decomposed IMFs using a BP neural network to make the signal amplitudes of the corresponding IMFs in the two signals tend to be the same. Then, construct a multimodal signal cancellation system to obtain the true radiated emission signal of the device under test (EUT).

[0045] The above steps will be explained in detail below with reference to specific content.

[0046] Empirical Fourier decomposition

[0047] The EFD (Empirical Fourier Decomposition) method combines EWT and Fourier decomposition to segment the Fourier spectrum and construct corresponding zero-phase filter banks. Then, the segmented frequency bands are reconstructed using inverse FFT. In this way, a complex, nonlinear, and non-stationary signal x(t) is decomposed into several single-component signals (intrinsic mode functions, IMFs) y. i The sum of (t) and the physically meaningful instantaneous frequency, i.e.:

[0048]

[0049] Where n(t) represents noise or residual signal.

[0050] The implementation steps of the EFD method are as follows:

[0051] Step 1: The signal x(t) is processed by Fast FFT to obtain F(ω). Its spectrum f(ω) = |F(ω)| is calculated and then normalized to [0, π].

[0052]

[0053] Step 2: Adaptively segment the spectrum f(ω) to obtain... Where K is the factorization number; f s For sampling frequencies f0 = 0 and f K =Fs / 2.

[0054] To obtain more accurate spectrum segmentation, local maxima and minima are used to divide the spectrum into N consecutive intervals, arranged in descending order. Thus, the N spectral intervals are obtained as Ω = [Ω1, Ω2, ..., LΩ]. N [By defining Ω0 = 0 and Ω in EFD] N =π, the spectrum segmentation boundary is as follows:

[0055]

[0056] Among them, X n Represents Ω n and Ω n+1 The Fourier spectrum amplitude between them.

[0057] Step 3: With ω n-1 and ω n A zero-phase filter was established with the cutoff frequency as the reference frequency. A zero-phase filter can preserve the main Fourier spectrum components and eliminate other components that were not obtained through segmentation. After filtering, the desired result is obtained.

[0058]

[0059] Step 4: Use interval B k =[f k-1 ,f k Perform inverse FFT processing on the signal to obtain a single decomposed component x. k (t)

[0060]

[0061] Step 5: Finally, sum all the decomposed components to obtain the reconstructed signal x. re (t):

[0062]

[0063] SGS-ACCI Process

[0064] Gaussian smoothing filters use a Gaussian function as the weighting function, selecting weights based on the size of the Gaussian window function for linear smoothing. The filtered signal is obtained by convolving the Gaussian kernel function with the original signal. Gaussian smoothing filters possess the properties of low-pass filters, effectively removing residual high-frequency random noise within a relatively small weight range, while having minimal impact on the high-frequency effective signal. Gaussian smoothing can be expressed as:

[0065]

[0066] In the formula: τ is the center point of the Gaussian window function on the time axis; h g (t) is the time-domain electromagnetic signal after Gaussian smoothing; g(t-τ) is the Gaussian window function, and the one-dimensional zero-mean Gaussian function is:

[0067]

[0068] In the formula: d is the Gaussian smoothing filter coefficient, which is used to control the time window width of the Gaussian window function.

[0069] By controlling the parameter d in equation (9), the bandwidth of the smoothing signal is adjusted, thereby achieving the purpose of adaptively determining the optimal segmentation boundary.

[0070] To adaptively determine the smoothing filter coefficients in each signal decomposition process, this application introduces the concept of envelope entropy (EE). The specific formula for EE is as follows:

[0071]

[0072] Where α(j) represents the envelope signal extracted after performing Hilbert demodulation on a specific IMF signal, P j This represents the normalized form of α(j), which essentially corresponds to a probability distribution sequence.

[0073] To adaptively determine the smoothing filter coefficients during each signal decomposition, this application introduces envelope entropy. The degree of signal disorder varies at each level of signal decomposition. Initially, the smoothing filter coefficients are set to 10, and the number of decomposition levels is K. In each iteration, d increases by 10. After i iterations, EE occurs. i With EE i-1 When the ratio is less than or equal to 1, the iteration ends, and the filtering coefficients corresponding to the (i-1)th iteration are the optimal smoothing filtering coefficients.

[0074] For broadband transient electromagnetic radiation emission testing in complex electromagnetic environments, the traditional empirical Fourier decomposition method is used to decompose the signal. This method determines the segmentation boundary based on local minima, which is prone to getting trapped in local optima. It performs well for narrowband signals but suffers from information loss for broadband signals. Therefore, this application first employs adaptive Gaussian smoothing filtering, resulting in a relatively smooth signal surface and avoiding local optima. The filtered signal is then segmented according to the same segmentation boundary to ensure that the frequency information contained in each decomposed IMF is approximately the same. Using the average correlation coefficient (ACC) as the adaptation index, through continuous optimization and iteration, the decomposition level with the largest ACC is selected, and the segmentation boundary at this point is determined as the optimal segmentation boundary. The ACC calculation formula is Equation (11).

[0075]

[0076] like Figure 2 The following are the specific decomposition steps of SGS-ACCI-EFD:

[0077] (1) The input signal EUT+ and the environmental interference are denoted as x1 and x2, respectively. The initial decomposition layer number K = 2 and the same segmentation boundary Ω are set.

[0078] (2) Calculate the frequency domain envelope entropy EE of each IMF signal and calculate the maximum envelope entropy when decomposing K layers.

[0079] (3) When MAX{EE appears K} / MAX{EE K-1 When} < 1, the corresponding d value is the optimal smoothing filter coefficient; otherwise, the iteration continues by increasing d by 10 until the condition is met.

[0080] (4) Perform smoothing filtering on x1 and x2, and the processed signals are denoted as X1 and X2.

[0081] (5) Perform improved empirical Fourier decomposition on X1 and X2, and redefine the dividing boundary as {Ω} = [Ω1, Ω2, ..., Ω2]. K Apply {Ω} to the input signals x1 and x2; decompose the input signals while maintaining the same center frequency and dividing boundaries, and decompose the modal signals as follows: IMF 1i IMF 2i These represent the modal signals after the channel signal is decomposed.

[0082] (6) Find the IMF 1i and IMF 2i The average correlation coefficient (ACC) of the corresponding modal signals. When the average correlation coefficient of the decomposition level K+1 is greater than the average correlation coefficient of K, it indicates that the IMF... 1i and IMF 2i The correlation is constantly increasing, and the decomposition level K is increased by 1, prompting the iterative optimization search process to continue. When ACC is satisfied... K / ACC K-1 When the value is less than 1, the iteration ends, and the optimal decomposition level K-1 and the partition boundary {Ω} are determined. ACC K ACC represents the average correlation coefficient when the number of decomposition layers is K. K-1 This represents the average correlation coefficient when the number of decomposition layers is K-1.

[0083] Multi-channel background interference cancellation method:

[0084] like Figure 3As shown, a multi-antenna array is used as the signal receiving device for electromagnetic radiation emission testing. An auxiliary reference channel is added to the existing main input channel and reference input channel to ensure the integrity of the acquired environmental interference and achieve better interference cancellation. The signals from the main reference channel and the auxiliary reference channel are fused into a single channel signal, resulting in more complete acquisition of environmental interference. Then, a second Gaussian smoothing filter is applied to the signal. Using the improved Empirical Fourier Decomposition (EFD) method, the two signals are decomposed along the same dividing boundary. The average correlation coefficient ACC of the signals with the same dividing boundary in different channels is calculated. The optimal decomposition level K and the dividing boundary {Ω} are determined by continuously searching for the maximum average correlation coefficient. A BP neural network is used for nonlinear fitting of each decomposed IMF to make the signal amplitudes of the corresponding IMFs in the two signals tend to be the same. Finally, a multi-mode signal cancellation system is constructed to obtain the true EUT radiated emission signal.

[0085] Antennas 1, 2, ..., n are signal receiving devices. The signals collected by n antennas can be represented as:

[0086]

[0087] Antenna 1 is designated as the main input channel, antenna 2 as the main reference input channel, and antennas 3 to n as auxiliary reference input channels, collectively referred to as x. f (t), the purpose is to ensure the integrity of the acquired background noise and remove some interference from the main input channel and the main reference channel. x f (t) represents x TD3 (t), x TD4 (t), ..., x TDn The sum of noise components in (t), x TDn (t) represents the signal acquired by the nth antenna, x nn (t) represents the nth element of the signal acquired by the nth antenna. The main input channel signal and the reference channel signal, after adaptive Gaussian smoothing filtering, are represented as:

[0088]

[0089] Where (x) TD1 -x f (t)).h g When (t)>0, |(x) TD1 -x f (t)).hg(t)|=(x TD1 -x f (t)).h g (t), when (x) TD1 -x f (t)).h gWhen (t) < 0, |(x TD1 -x f (t)).h g (t)|=0.

[0090] Perform empirical Fourier decomposition on the smoothed and filtered signal and determine the segmentation boundary.

[0091] φ k =[{f1,f2},{f3,f4},L,{f n-1 ,f n}]=[φ1,φ2,...,φ K (14)

[0092] Where f1, f2, K f n To divide the frequency points, φ1, φ2, ..., φ K It is a set of segmented frequency bands.

[0093] Using the dividing boundary φ k Frequency domain decomposition is performed on the signals from channels 1, 2, ..., n to ensure that the broadband EUT radiated emission signal is decomposed into a single sub-signal, and that each sub-signal contains the same frequency information. The average correlation coefficient (ACC) is used as the evaluation metric. With K=2 as the initial decomposition layer, the signal is adaptively iterated to determine the optimal number of decomposition layers and the segmentation boundary. The decomposed sub-band signals are denoted as:

[0094]

[0095] In the formula, S0(t)=x TD1 (t),

[0096] Performing a Fast Fourier Transform (FFT) on S0(t) and S1(t) in the frequency domain yields the following signals.

[0097]

[0098] Since there is an amplitude difference between the noise mode signals of the main input channel and the reference channel, a BP neural network is used to perform nonlinear fitting on the decomposed two-channel mode signals, and the resulting fitting coefficient vector is denoted as m:

[0099]

[0100] The component of the desired radiated emission signal contained in the main input channel signal is denoted as the sum of the noise signal and the desired radiated emission signal (the useful signal is denoted as d(ω), and the noise signal is denoted as v0(ω)), as shown below.

[0101] F0(ω)=d(ω)+mv0(ω) (18)

[0102] With F0(ω) as the objective function and F1(ω) as the fitted function, the expression after multiplying by the weight coefficient vector m is:

[0103] F 11 (ω)=mF1(ω)=[m1F 21 (ω),m1F 22 (ω),...,m K F 2K (ω)] (19)

[0104] After performing nonlinear fitting on the two-channel signals, multi-component cancellation is performed. For noise signals, F... 1K (ω)-m K F 1K (ω)→0, for signal components containing EUT, we have v0(ω)-m K F 1K (ω)→0, from which the EUT radiation emission signal d(ω) can be obtained.

[0105] Simulation test results and analysis

[0106] SGS-ACCI-EFD Method:

[0107]

[0108] In equation (20), j = 1, 2, 3; A 1j =0.002, 0.001, 0.001; A 2j =0.002, 0.002, 0.0015; and the signal-to-noise ratio (SNR) is -5dB. The background interference in acquisition channels A and B are different. N is the sampling point, T is the pulse period (T = 0.25), σ is the pulse width (σ = 0.1), and "freqs" is the sampling frequency. The simulated EUT radiated emission signal is S0, and the simulated electromagnetic interference signals are f1 = 100MHz, f2 = 500MHz, f3 = 900MHz, S1, and t = max(t) / 2. The standard deviation is expressed as "sigma" and is 8 × 10⁻⁶. -15 .

[0109] like Figures 4a-4b The input signal power waveform is used for simulation testing. Due to the close testing distance, the background interference of the two channels is highly correlated. The expected radiated emission signal to be tested is an amplitude-frequency modulated signal with a starting frequency of 600MHz and a cutoff frequency of 800MHz, with a bandwidth of 200MHz. How to decompose the expected radiated emission signal into only one sub-signal without containing noise signals from other frequency bands under the condition of no prior knowledge has become a key research problem.

[0110] Figures 5a-5fThe performance of different signal decomposition methods for broadband electromagnetic radiation emission signals was compared. As shown in the figure, the T-EFD, AEFD, and EAEFD methods essentially select local extrema in the test frequency band as the segmentation boundary to decompose the signal. Therefore, if the desired test frequency band contains multiple extrema, over-decomposition occurs due to local optima, preventing the desired radiated emission signal from being accurately decomposed into a single sub-signal, thus affecting the algorithm's performance. The ASEI-VMD and GA-VMD-SVD methods also exhibit over-decomposition and unstable decomposition, making them primarily suitable for narrowband electromagnetic radiation emission testing. Simulation analysis verifies that the proposed SGS-ACCI-EFD method, by redefining the segmentation boundary, can decompose transient electromagnetic radiation emission signals into a single sub-signal with stable performance and short running time, solving a key step in broadband electromagnetic radiation emission testing.

[0111] Broadband electromagnetic radiation emission test simulation

[0112] Combining the SGS-ACCI-EFD method, this application proposes a method for testing multi-channel broadband transient electromagnetic radiation emission signals. The use of multiple channels to collect background interference is to ensure the integrity of background noise and improve the accuracy of background interference cancellation. Figures 6a-6b This is the simulated input signal for broadband electromagnetic radiation emission testing.

[0113] Figures 7a-7b To compare the canceled radiated emission signal with the expected radiated emission signal, the figure shows that the canceled radiated emission signal and the expected radiated emission signal have very small differences in amplitude, and the noise outside the 600-800MHz test bandwidth is relatively small. Figures 7a-7b The correlation between the canceled signal and the actual radiated emission signal shows that the proposed algorithm achieves a correlation of over 97.5% between the canceled radiated emission signal and the expected radiated emission signal, meeting the requirements of actual testing and verifying the effectiveness of the proposed algorithm.

[0114] Table 1 compares the performance of cancellation based on different decomposition methods, focusing on three metrics: runtime, signal correlation, and power error. As shown in the table, the proposed SGS-ACCI-EFD method, compared to T-EFD, AEDF, EAEFD, GA-VMD-SVD, and ASEI-VMD methods, achieves a runtime of 0.2 s, a signal correlation of 0.975, and an accuracy improvement of over 17%. Furthermore, the power error between the canceled radiated emission signal and the true radiated emission signal is less than 1 dB, representing a 3 dB improvement. From the perspective of noise cancellation performance, the effectiveness of the proposed algorithm is verified.

[0115] Table 1. Performance comparison of elimination based on different decomposition methods

[0116]

[0117] Experimental test results and analysis

[0118] Experimental setup as follows Figure 8 As shown. Antennas 1, 2, 3, and 4 are self-developed Vivaldi miniaturized antennas with identical specifications, operating in the frequency range of 0.47–12 GHz and a gain range of 1–11 dBi. All four antennas are positioned L away from the EUT transmitting antenna, but their test orientations differ: Antenna 1 faces the EUT transmitting antenna to collect the radiated emission signal and environmental interference; antennas 2, 3, and 4 face away from the EUT transmitting antenna to collect more complete environmental interference data. The EUT transmitting antenna is a 3142E hybrid log-periodic antenna with a test frequency range of 30 MHz–6 GHz. A signal generator, model 1435D, simulates the actual radiated emission signal from the EUT, with a frequency range of 9 kHz–6 GHz and a maximum output power of 20 dBm. Data acquisition is achieved using a Tektronix MSO64B four-channel oscilloscope with a test bandwidth of 4 GHz and a signal sampling frequency of 25 GHz / s. All four antennas are on the same horizontal plane, and the electromagnetic interference is primarily from the surrounding environment. like Figure 8 This is the test site.

[0119] Multi-frequency narrowband electromagnetic radiation emission test:

[0120] The selected test site was an outdoor playground, where the background noise was mainly environmental noise. The actual radiated emission signal from the EUT was an amplitude-frequency modulated (AM) signal with a frequency of 800 MHz, a power of 0 dB, a sine wave, an amplitude modulation rate of 1 MHz, a modulation depth of 30%, a frequency modulation rate of 7 MHz, and a modulation frequency deviation of 4 MHz. The experimental setup and testing site layout were as follows: Figure 8 Same. For example... Figures 9a-9b For receiving signal comparison ( Figure 9a These are environmental interference signals collected by antennas 2, 3, and 4. Figure 9b (This refers to the EUT + environmental interference signal collected by antenna 1). During the experiment, the four antennas were 1m away from the EUT's transmitting antenna, creating a 2m × 2m test field. The electromagnetic environment in this test field was considered relatively constant. The figure shows that the environmental noise collected by the two antennas is strongly correlated.

[0121] Figures 10a-10bThe output power and signal correlation of the electromagnetic radiation emission signal obtained by this method after cancellation were compared with those of the actual EUT radiation emission signal. As shown in the figure, compared with ASEI-VMD, GA-VMD-SVD, EAEFD, AEFD, and T-EFD, the proposed algorithm has the closest signal power to the actual EUT radiation emission signal, and the power error can be stabilized within 1dB; its correlation is above 97%, which is better than other algorithms. Therefore, the effectiveness of the SGS-ACCI-EFD algorithm proposed in this application is proved.

[0122] Broadband electromagnetic radiation emission test:

[0123] Figures 11a-11b This is a comparison of the two received signals. Figure 12a It can be seen that the intensity of the EUT radiated emission signal obtained after cancellation by this method is very small compared with the actual radiated emission signal, such as... Figure 12b To compare the output signal power after cancellation with the actual transmitted signal power, it can be seen that the signal power difference is within 1dB in the bandwidth range of 400-750MHz, indicating that the method has a significant effect on noise cancellation in a wide bandwidth and meets the experimental test requirements.

[0124] like Figure 13 For the comparison of signal correlation after cancellation by different decomposition methods, the analysis of the figure shows that, for transient electromagnetic radiation emission testing, the proposed SGS-ACCI-EFD multi-channel cancellation method has a higher signal correlation than ASEI-VMD, EAEFD, GA-VMD-SVD, AEFD, and T-EFD, reaching over 97%. The correlation of the other methods is between 50% and 90%, indicating that the method of this application has smaller signal error after cancellation and can more accurately obtain the desired EUT radiation emission signal.

[0125] Table 2 Performance Comparison of Different Decomposition Methods

[0126]

[0127] Table 2 above compares the performance of different decomposition methods. As shown in the table, this method redefines the segmentation boundary by introducing Gaussian smoothing filtering, and determines the optimal number of decomposition layers and segmentation boundary through iterative analysis of the average correlation coefficient. It is suitable for broadband transient electromagnetic radiation emission testing. Compared to T-EFD, AEFD, and EAEFD methods, the algorithm proposed in this application avoids the problem of broadband failure and has a relatively short running time. Compared to ASEI-VMD and GA-VMD-SVD methods, it not only reduces the time by an order of magnitude but also avoids over-decomposition and EUT information loss. Compared to the VMD method, the algorithm in this paper is more stable.

[0128] In summary, the method described in this application has the following advantages:

[0129] (1) Compared with EAEFD, GA-VMD-SVD, AEFD, and T-EFD algorithms, the method described in this application determines the optimal number of decomposition layers by redefining the signal segmentation boundary and using the average correlation coefficient for adaptive iteration. This method is applicable to both narrowband signal decomposition and transient signal decomposition.

[0130] (2) Compared with the ASEI-VMD method, the method described in this application does not have under-decomposition, over-decomposition and mode mixing phenomena, and the running time is reduced from more than 5 minutes to less than 10 seconds, which greatly improves the running speed.

[0131] (3) This method can effectively obtain the real EUT radiated emission signal without prior knowledge. The correlation between the obtained EUT radiated emission signal and the real EUT radiated emission signal is greater than 97%, and the error of the obtained signal power is less than 1 dB. Simulation and experimental test results show that this method is suitable for transient electromagnetic radiated emission test research in complex electromagnetic environments.

Claims

1. A method for testing multi-channel broadband transient electromagnetic radiation emission signals, characterized in that... Includes the following steps: Using a multi-electrode antenna as the signal receiving device for electromagnetic radiation emission testing, one antenna serves as the main input antenna, one antenna serves as the main reference antenna, and the remaining antennas serve as auxiliary reference antennas. The signal acquired by the main input antenna is used as the main channel signal, the signal acquired by the main reference antenna is used as the main reference channel signal, and the signal acquired by the auxiliary reference antenna is used as the auxiliary reference channel signal. The main reference channel signal and the auxiliary reference channel signal are fused into a single reference channel signal. The fused reference channel signal and the main channel signal are subjected to a second Gaussian smoothing filter. The improved empirical Fourier decomposition method is used to decompose the two signals according to the same segmentation boundary. The average correlation coefficient ACC of the signals with the same segmentation boundary in different channels is calculated. The optimal decomposition level K and segmentation boundary {Ω} are determined by continuously searching for the maximum average correlation coefficient ACC. The decomposed IMFs are subjected to nonlinear fitting using a BP neural network to make the signal amplitudes of the corresponding IMFs in the two signals tend to be the same. Then, a multimodal signal cancellation system is constructed to obtain the true radiated emission signal of the device under test (EUT).

2. The multi-channel broadband transient electromagnetic radiation emission signal testing method as described in claim 1, characterized in that: Antennas 1, 2, ..., n are signal receiving devices. The signals collected by n antennas are represented as follows: Antenna 1 is designated as the main input channel, antenna 2 as the main reference input channel, and antennas 3 to n as auxiliary reference input channels, collectively referred to as x. f (t), x f (t) represents x TD3 (t), x TD4 (t), ..., x TDn The sum of noise components in (t), x TDn (t) represents the signal acquired by the nth antenna, x nn (t) represents the nth element of the signal collected by the nth antenna.

3. The multi-channel broadband transient electromagnetic radiation emission signal testing method as described in claim 2, characterized in that: The main input channel signal and the reference channel signal, after adaptive Gaussian smoothing filtering, are represented as follows: Where (x) TD1 -x f (t)).h g When (t)>0, |(x) TD1 -x f (t)).hg(t)|=(x TD1 -x f (t)).h g (t), when (x) TD1 -x f (t)).h g When (t) < 0, |(x TD1 -x f (t)).h g (t)|=0; The smoothed and filtered signal is subjected to empirical Fourier decomposition, and the segmentation boundary is determined as follows: f k =[{f1,f2},{f3,f4},…,{f n-1 ,f n }]=[φ1,φ2,...,φ K ] Where f1, f2, ... f n To divide the frequency points, φ1, φ2, ..., φ K It is a set of segmented frequency bands.

4. The multi-channel broadband transient electromagnetic radiation emission signal testing method as described in claim 3, characterized in that: Using the dividing boundary φ k Frequency domain decomposition is performed on the signals from channels 1, 2, ..., n to ensure that the radiated emission signal of the broadband EUT is decomposed into a single sub-signal, and that each sub-signal contains the same frequency information. The average correlation coefficient (ACC) is used as the evaluation metric, with K=2 as the initial decomposition level. The signal is then adaptively iterated to determine the optimal number of decomposition levels and the segmentation boundary. The decomposed sub-band signals are denoted as... In the formula, S0(t)=x TD1 (t), 5. The multi-channel broadband transient electromagnetic radiation emission signal testing method as described in claim 4, characterized in that: Performing a Fast Fourier Transform (FFT) on S0(t) and S1(t) in the frequency domain yields the following signals: Since there is an amplitude difference between the noise mode signals of the main input channel and the reference channel, a BP neural network is used to perform nonlinear fitting on the decomposed two-channel mode signals, and the resulting fitting coefficient vector is denoted as m: The component of the desired radiated emission signal contained in the main input channel signal is denoted as the sum of the noise signal and the desired radiated emission signal. The useful signal is denoted as d(ω), and the noise signal is denoted as v0(ω), as shown below: F0(ω)=d(ω)+mv0(ω) With F0(ω) as the objective function and F1(ω) as the fitted function, the expression after multiplying by the weight coefficient vector m is: F 11 (ω)=mF1(ω)=[m1F 21 (ω),m1F 22 (ω),...,m K F 2K (oh)] After performing nonlinear fitting on the two-channel signals, multi-component cancellation is performed. For noise signals, F... 1K (ω)-m K F 1K (ω)→0, for signal components containing EUT, we have v0(ω)-m K F 1K (ω)→0, from which the EUT radiation emission signal d(ω) can be obtained.

6. The multi-channel broadband transient electromagnetic radiation emission signal testing method as described in claim 4, characterized in that: The method for determining the optimal number of decomposition levels and the partition boundary includes the following steps: 1) Let the input signal EUT + environmental interference be denoted as x1 and the environmental interference as x2. Set the initial decomposition level K = 2 and the same segmentation boundary Ω. 2) Calculate the frequency domain envelope entropy EE for each IMF signal and calculate the maximum envelope entropy MAX{EE} when decomposing K layers. K }; 3) When MAX{EE} appears K } / MAX{EE K-1 When} < 1, the corresponding d value is the optimal smoothing filter coefficient, where d is the smoothing filter coefficient; otherwise, the iteration continues by increasing d by 10 until the condition is met. 4) Perform smoothing filtering on x1 and x2, and denote the processed signals as X1 and X2; 5) Perform improved empirical Fourier decomposition on X1 and X2, and redefine the dividing boundary as {Ω} = [Ω1, Ω2, ..., Ω2]. K Apply {Ω} to the input signals x1 and x2; decompose the input signals while maintaining the same center frequency and dividing boundaries, and decompose the modal signals as follows: IMF 1i IMF 2i These represent the modal signals after channel signal decomposition; 6) Calculate the IMF 1i and IMF 2i The average correlation coefficient (ACC) of the corresponding modal signals; when the average correlation coefficient of the decomposition level K+1 is greater than the average correlation coefficient of K, it indicates that the IMF... 1i and IMF 2i The correlation is constantly increasing, the decomposition level K is increased by 1, prompting the iterative optimization search process to continue; when ACC is satisfied... K / ACC K-1 When the value is less than 1, the iteration ends, and the optimal decomposition level K-1 and the partition boundary {Ω} are determined. ACC K ACC represents the average correlation coefficient when the number of decomposition layers is K. K-1 This represents the average correlation coefficient when the number of decomposition layers is K-1.

Images