CEEMDAN-ICA-based radar time domain signal processing method
By decomposing and separating radar time-domain signals using the CEEMDAN-ICA method, the problem of signal distortion in traditional methods is solved, achieving effective noise suppression and signal separation, and improving the accuracy and signal-to-noise ratio of signal processing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-17
- Publication Date
- 2026-04-14
AI Technical Summary
Traditional radar time-domain signal processing methods are prone to signal distortion after denoising when processing non-stationary discrete radar echo signals, and it is difficult to effectively distinguish target information from clutter. Existing methods cannot effectively preserve signal characteristics during the denoising process.
A signal processing method based on CEEMDAN-ICA is adopted. The radar time-domain signal is decomposed by the CEEMDAN algorithm, the noise component is selected by using the fuzzy entropy coefficient and a virtual noise channel is constructed, and the effective signal is separated by the FastICA algorithm to achieve adaptive mode decomposition and denoising.
It improves the signal-to-noise ratio of radar time-domain signals, effectively separates signal components, preserves signal characteristics, and enhances the accuracy of signal sorting and the support for subsequent analysis, which is superior to traditional methods.
Smart Images
Figure CN121856903A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of radar time-domain signal preprocessing, specifically relating to a radar time-domain signal processing method based on CEEMDAN-ICA. Background Technology
[0002] With the increasing complexity of the electromagnetic environment, the signals received by electronic reconnaissance systems have become a superposition of time-domain signals, with signal radiation sources including radar, communication equipment, and other devices. Radar echo signals, in particular, possess non-stationary and discrete mathematical characteristics, with effective information often hidden in subtle details, making the distinction between target information and clutter quite complex. These characteristics mean that traditional radar time-domain signal processing techniques, such as matched filtering, time-space filtering, and wavelet theory, can lead to some degree of distortion in the denoised radar time-domain signal. Summary of the Invention
[0003] To address the aforementioned problems, the present invention aims to provide a radar time-domain signal processing method based on CEEMDAN-ICA, which can effectively suppress noise while preserving the characteristics of radar time-domain signals as much as possible, and improve the accuracy of radar time-domain signal sorting.
[0004] The specific technical solution for achieving the objective of this invention is as follows:
[0005] A radar time-domain signal processing method based on CEEMDAN-ICA includes the following steps:
[0006] Step 1: Use the CEEMDAN algorithm to decompose the received radar time-domain signal to obtain the IMF components with frequencies from high to low.
[0007] Step 2: Calculate the fuzzy entropy coefficients of each IMF component, find their mean, and select the IMF components with a larger fuzzy entropy mean as noise component layers based on the fuzzy entropy coefficients.
[0008] Step 3: By comparing the fuzzy entropy of each IMF component with its average value, select a portion of the IMF components, add them together to form a virtual noise channel, and then add it to the original signal. It also serves as input to the FastICA algorithm;
[0009] Step 4: Separate the effective signal from the radar time domain signal by using the FastICA algorithm to separate the noise signal and radar time domain signal in the virtual noise channel.
[0010] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0011] The solution of this invention utilizes a mode decomposition method, which does not require prior knowledge or fixed basis functions. It only requires adjusting the parameters to process the original signal and can effectively suppress noise, thereby improving the signal-to-noise ratio of the radar time-domain signal. It can adaptively decompose the signal into modes.
[0012] Meanwhile, this type of method can effectively separate signals of different components during the decomposition process. Therefore, while performing noise reduction, this method can also effectively separate different components of the input signal, preserving the characteristics of the effective signal as much as possible and providing strong support for subsequent signal analysis.
[0013] This paper proposes a threshold-based denoising algorithm combining CEEMDAN-ICA and energy entropy, and compares its filtering performance with that of the aforementioned mode decomposition algorithms. The comparison shows that this proposed algorithm outperforms the aforementioned mode decomposition algorithms, improving the accuracy of radar time-domain signal characteristic analysis.
[0014] The present invention will be further described below with reference to specific embodiments. Attached Figure Description
[0015] Figure 1 This is a schematic diagram of the radar time-domain signal processing method based on CEEMDAN-ICA of the present invention.
[0016] Figure 2 This is a time-domain waveform diagram of the simulated signal in an embodiment of the present invention.
[0017] Figure 3 This is a time-domain waveform diagram obtained by adding noise to the simulated signal in an embodiment of the present invention.
[0018] Figure 4 The image shows the time-domain waveform of the simulated signal obtained after processing by the CEEMDAN method in this embodiment of the invention.
[0019] Figure 5 This is a time-domain waveform diagram obtained after processing the simulated signal using the method of this invention in an embodiment of the invention. Detailed Implementation Example
[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only some embodiments of the present invention, and not all embodiments. 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.
[0021] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.
[0022] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.
[0023] Combination Figure 1 A radar time-domain signal processing method based on CEEMDAN-ICA includes the following steps:
[0024] Step 1: Use the CEEMDAN algorithm to decompose the received radar time-domain signal to obtain the IMF components from high to low frequency:
[0025] Step 1-1: Decompose each noisy sequence using EMD. ,in For the i-th Gaussian white noise, The number of noises. It is the noise figure;
[0026] Obtain the first modal component and define the first CEEMDAN modal component as:
[0027]
[0028] in, The mode is the i-th noisy sequence after EMD decomposition;
[0029] Steps 1-2: Extract the first IMF component from the original signal. Subtracting from the middle yields the first residual component. for:
[0030]
[0031] Steps 1-3: Decompose each residual value using EMD This continues until its first modal component is calculated, and the second CEEMDAN modal component is defined as:
[0032]
[0033] This indicates the remaining parameters added to the Gaussian white noise. Indicates the noise figure;
[0034] Steps 1-4, for Calculate the first Individual residuals:
[0035]
[0036] For each Each This continues until its first modal component is calculated, and the (j+1)th CEEMDAN modal component is defined as:
[0037]
[0038] Steps 1-5: j = j + 1, then proceed to step 1-4;
[0039] Steps 1-6, and steps 1-4 and 1-5, are repeated until the obtained residuals cannot be further decomposed by EMD because they satisfy the IMF criteria or are less than three local extrema. The final residuals satisfy:
[0040]
[0041] Where N is the total number of modal components, the input signal is represented as:
[0042]
[0043] Step 2: Calculate the fuzzy entropy coefficients of each IMF component, find their mean, and select the IMF components with fuzzy entropy coefficients larger than the mean fuzzy entropy value as noise component layers.
[0044] For the time series of each IMF component L is the data length, m is the initial pattern dimension, and an m-dimensional vector is constructed:
[0045]
[0046] in, The reconstructed time series, ;
[0047] Define two m-dimensional vectors and The distance between them is the maximum absolute value of the differences between their corresponding elements, that is:
[0048]
[0049] in, ;
[0050] Introducing fuzzy membership functions to define vectors and Similarity between them:
[0051]
[0052] Where r is the similarity tolerance parameter, which is defined as R times the standard deviation of the original one-dimensional time series, i.e. , R is the standard deviation of the original data; typically, the value of R ranges from 0.1 to 0.2.
[0053] Define a function:
[0054]
[0055] Then you can get Relationship dimension under dimension:
[0056]
[0057] Increasing the pattern dimension m by 1, i.e., repeating the above steps for the m+1 dimension vector, yields the relation dimension in the m+1 dimension, as shown in the formula:
[0058]
[0059] Based on this, the fuzzy entropy coefficient of the IMF component of this order is determined:
[0060]
[0061] By calculating the fuzzy entropy coefficients of each order of IMF components and the mean of the fuzzy entropy coefficients, the first k IMFs with fuzzy entropy greater than the mean are added together as noise components to form a virtual noise channel.
[0062] Step 3: By comparing the fuzzy entropy of each IMF component with its average value, select a portion of the IMF components, add them together to form a virtual noise channel, and then add it to the original signal. It also serves as input to the FastICA algorithm;
[0063] Step 4: Separate the effective signal from the radar time-domain signal using the FastICA algorithm, combining the noise signal in the virtual noise channel and the radar time-domain signal.
[0064] In the standard ICA algorithm, it is assumed that the signal is an m-dimensional signal. It is formed by n independent source signals passing through an unknown signal. This is obtained through linear aliasing in the transmission system.
[0065]
[0066] It can be rewritten in matrix form
[0067]
[0068] In the formula, A is the unknown channel coefficient. The goal of the ICA algorithm is to calculate a matrix W from the received signal X, resulting in the output matrix Y = WX, where X is the source signal matrix. The optimal approximate estimator. However, considering the complexity of actual situations, the original signal model will also be affected by noise. Therefore, the actual received signal can be expressed as:
[0069]
[0070] in, This is a virtual noise channel formed by the noise components determined in step 3. Original signal Where A is an unknown channel coefficient The mixture matrix consists of s, where s is the effective signal to be separated.
[0071]
[0072]
[0073] Assume the added virtual noise channel contains m types of noise signals. The received signal is then represented as:
[0074]
[0075] in, and Let s be the source signal s and the i-th noise component, respectively. Weights in the received signal;
[0076] Right now:
[0077]
[0078] The FastICA algorithm is used to obtain the estimated separation matrix by calculating the inverse of matrix A, such that... The effective signal to be separated The optimal approximate estimator is obtained by extracting the source signal from the output signal S since the noise u is artificially introduced, thereby achieving effective signal separation in the radar time domain signal and thus achieving the purpose of noise reduction.
[0079] Figure 2 This is the time-domain waveform of the simulated signal in this embodiment. Figure 3 This is the time-domain waveform obtained by adding noise to the simulated signal. Figure 4 This is the time-domain waveform obtained after processing the simulated signal using the CEEMDAN method. Figure 5 It is a time-domain waveform diagram obtained after the simulated signal is processed by the method of this invention.
[0080] The sampling frequency of the simulated signal was 2000Hz, the sampling time was 1s, and the center frequencies of the simulated signal were 100Hz, 80Hz, and 70Hz, with maximum amplitudes of 2, 5, and 3, respectively. The added background noise signal-to-noise ratio was 0dB. As can be seen from the figure, the time-domain waveform of the signal processed by this invention is significantly better than the result processed by the CEEMDAN method. Therefore, it can be demonstrated that this invention has a good effect on the preprocessing of the original radar time-domain signal.
[0081] This solution also provides a radar time-domain signal processing system based on CEEMDAN-ICA, including the following modules:
[0082] Decomposition module: The CEEMDAN algorithm is used to decompose the received radar time-domain signal to obtain the IMF components with frequencies from high to low.
[0083] Noise component module: used to calculate the fuzzy entropy coefficients of each order of IMF components, calculate their mean, and select IMF components with a larger fuzzy entropy mean value as noise component layers based on the fuzzy entropy coefficients.
[0084] FastICA separation module: Used to compare the fuzzy entropy of each IMF component with the average value, where the top k IMFs with fuzzy entropy greater than the average value are treated as noise components, and these are added together to form a virtual noise channel, which is then compared with the original signal. Simultaneously, it serves as the input to FastICA; the noise signal in the virtual noise channel and the radar time-domain signal are separated into effective signals in the radar time-domain signal using the FastICA algorithm.
[0085] This solution also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:
[0086] Step 1: Use the CEEMDAN algorithm to decompose the received radar time-domain signal to obtain the IMF components with frequencies from high to low.
[0087] Step 2: Calculate the fuzzy entropy coefficients of each IMF component, find their mean, and select the IMF components with a larger fuzzy entropy mean as noise component layers based on the fuzzy entropy coefficients.
[0088] Step 3: By comparing the fuzzy entropy of each IMF component with its average value, select a portion of the IMF components, add them together to form a virtual noise channel, and then add it to the original signal. It also serves as input to the FastICA algorithm;
[0089] Step 4: Separate the effective signal from the radar time domain signal by using the FastICA algorithm to separate the noise signal and radar time domain signal in the virtual noise channel.
[0090] This solution also provides a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, performs the following steps:
[0091] Step 1: Use the CEEMDAN algorithm to decompose the received radar time-domain signal to obtain the IMF components with frequencies from high to low.
[0092] Step 2: Calculate the fuzzy entropy coefficients of each IMF component, find their mean, and select the IMF components with a larger fuzzy entropy mean as noise component layers based on the fuzzy entropy coefficients.
[0093] Step 3: By comparing the fuzzy entropy of each IMF component with its average value, select a portion of the IMF components, add them together to form a virtual noise channel, and then add it to the original signal. It also serves as input to the FastICA algorithm;
[0094] Step 4: Separate the effective signal from the radar time domain signal by using the FastICA algorithm to separate the noise signal and radar time domain signal in the virtual noise channel.
[0095] The embodiments described above are merely one implementation method of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A radar time-domain signal processing method based on CEEMDAN-ICA, characterized in that, Includes the following steps: Step 1: Use the CEEMDAN algorithm to decompose the received radar time-domain signal to obtain the IMF components with frequencies from high to low. Step 2: Calculate the fuzzy entropy coefficients of each IMF component, find their mean, and select the IMF components with a larger fuzzy entropy mean as noise component layers based on the fuzzy entropy coefficients. Step 3: By comparing the fuzzy entropy of each IMF component with its average value, select a portion of the IMF components, add them together to form a virtual noise channel, and then add it to the original signal. It also serves as input to the FastICA algorithm; Step 4: Separate the effective signal from the radar time domain signal by using the FastICA algorithm to separate the noise signal and radar time domain signal in the virtual noise channel.
2. The radar time-domain signal processing method based on CEEMDAN-ICA according to claim 1, characterized in that, The radar time-domain signal decomposition in step 1 is specifically as follows: Step 1-1: Decompose each noisy sequence using EMD. ,in For the i-th Gaussian white noise, The number of noises. It is the noise figure; Obtain the first modal component and define the first CEEMDAN modal component as: ； in, The mode is the i-th noisy sequence after EMD decomposition; Steps 1-2: Extract the first IMF component from the original signal. Subtracting from the middle yields the first residual component. for: ； Steps 1-3: Decompose each residual value using EMD This continues until its first modal component is calculated, and the second CEEMDAN modal component is defined as: ； This indicates the remaining parameters added to the Gaussian white noise. Indicates the noise figure; Steps 1-4, for Calculate the first Individual residuals: ； For each Each is decomposed through EMD This continues until its first modal component is calculated, and the (j+1)th CEEMDAN modal component is defined as: ； Steps 1-5: j = j + 1, then proceed to step 1-4; Steps 1-6, and steps 1-4 and 1-5, are repeated until the obtained residuals cannot be further decomposed by EMD because they satisfy the IMF criteria or are less than three local extrema. The final residuals satisfy: ； Where N is the total number of modal components, the input signal is represented as: 。 3. The radar time-domain signal processing method based on CEEMDAN-ICA according to claim 1, characterized in that, The fuzzy entropy coefficient in step 2 is: For the time series of each IMF component L is the data length, m is the initial pattern dimension, and an m-dimensional vector is constructed: ； in, The reconstructed time series, ; Define two m-dimensional vectors and The distance between them is the maximum absolute value of the differences between their corresponding elements, that is: ； in, ; Introducing fuzzy membership functions to define vectors and Similarity between them: ； Where r is the similarity tolerance parameter, which is defined as R times the standard deviation of the original one-dimensional time series, i.e. , The standard deviation of the original data; Define a function: ； Then you can get Relationship dimension under dimension: ； Increasing the pattern dimension m by 1, i.e., repeating the above steps for the m+1 dimension vector, yields the relation dimension in the m+1 dimension, as shown in the formula: ； Based on this, the fuzzy entropy coefficient of the IMF component of this order is determined: 。 4. The radar time-domain signal processing method based on CEEMDAN-ICA according to claim 3, characterized in that, By calculating the fuzzy entropy coefficients of each order of IMF components and the mean of the fuzzy entropy coefficients, the first k IMFs with fuzzy entropy greater than the mean are added together as noise components to form a virtual noise channel.
5. The radar time-domain signal processing method based on CEEMDAN-ICA according to claim 1, characterized in that, Step 4, which involves separating the effective signal from the radar time-domain signal using the FastICA algorithm, specifically involves: The input signal for the FastICA algorithm is: ； in, This is a virtual noise channel formed by the noise components determined in step 3. Original signal Where A is an unknown channel coefficient The mixture matrix consists of s, where s is the effective signal to be separated. ； ； Assume the added virtual noise channel contains m types of noise signals. The received signal is represented as: ； in, and Let s be the source signal s and the i-th noise component, respectively. Weights in the received signal; Right now: ； The FastICA algorithm is used to obtain the estimated separation matrix by calculating the inverse of matrix A, such that... The effective signal to be separated The optimal approximate estimator is obtained, thereby achieving effective signal separation in the radar time domain signal.
6. A radar time-domain signal processing system based on CEEMDAN-ICA, characterized in that, Includes the following modules: Decomposition module: The CEEMDAN algorithm is used to decompose the received radar time-domain signal to obtain the IMF components with frequencies from high to low. Noise component module: used to calculate the fuzzy entropy coefficients of each order of IMF components, calculate their mean, and select IMF components with a larger fuzzy entropy mean value as noise component layers based on the fuzzy entropy coefficients. FastICA separation module: Used to compare the fuzzy entropy of each IMF component with the average value, where the top k IMFs with fuzzy entropy greater than the average value are treated as noise components, and these are added together to form a virtual noise channel, which is then compared with the original signal. Simultaneously, it serves as the input to FastICA; the noise signal in the virtual noise channel and the radar time-domain signal are separated into effective signals in the radar time-domain signal using the FastICA algorithm.
7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-5.
8. A computer-storable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-5.