Frequency band division method and device based on empirical spectrum trend and electronic equipment

Abstract

The invention relates to the technical field of signal processing, and provides a frequency band division method based on an empirical spectrum trend, which is applied to multi-component non-stationary signal processing in a complex noise environment, and comprises the following steps: obtaining an observation signal spectrum trend according to an amplitude spectrum of an input signal, the observation signal spectrum trend comprises a linear combination of a target signal spectrum trend and a background noise spectrum trend; the target signal comprises a mechanical vibration signal, a biomedical signal, an electromagnetic wave signal, a sound wave signal or an artificial signal; calculating the difference between the observation signal spectrum trend and the background noise spectrum trend to obtain an empirical spectrum trend; the empirical spectrum trend indicates overall fluctuation of spectrum energy of the target signal; traversing a first critical point of the empirical spectrum trend to divide a frequency band; on the basis of the frequency band division result, traversing a second critical point of the empirical spectrum trend to determine the center frequency of each signal component; therefore, adaptive frequency band division and signal component center frequency estimation of the multi-component non-stationary signal are realized.

Description

Technical Field

[0001] This invention relates to the field of signal processing, and more particularly to a frequency band division method, apparatus, and electronic device based on empirical spectral trends. Background Technology

[0002] The core task of signal processing is to extract and identify effective components from complex observations, and suppress or remove interference components mixed in the signal, thereby providing reliable input for subsequent applications such as parameter estimation, target recognition, and decision support. With the full arrival of the big data era and the rapid development of intelligent information detection technology, the level of informatization in modern society is constantly improving, which places higher demands on the efficiency, accuracy, security, and adaptability of digital signal processing technologies for various instruments and equipment. To meet these requirements and the various needs for real-time performance and robustness in engineering applications, scholars at home and abroad have researched and developed modern signal processing methods such as time series analysis, spectral kurtosis, envelope modulation, sparse representation, and adaptive signal decomposition. Among them, adaptive signal decomposition methods can decompose arbitrarily complex multi-component signals into a series of single-component signals without considering prior transformations. Currently, commonly used adaptive signal decomposition methods can be roughly divided into four categories from the perspective of the essential principles of signal decomposition: time-domain decomposition methods, iterative filtering decomposition methods, time-frequency decomposition methods, and frequency-domain decomposition methods. Among these, frequency-domain decomposition methods decompose signals from the frequency domain perspective, and have a direct and effective processing effect on signals without time-frequency intersections. However, existing classical frequency domain decomposition methods lack utilization of spectral morphology information: methods such as empirical wavelet transform and adaptive empirical Fourier decomposition divide frequency bands based on the distribution of the amplitude spectrum, and their results are highly susceptible to interference. The performance of variational mode decomposition and its variants is closely related to the selection of model parameters; the number of decomposed modes, the initial center frequency of the modes, and other variational model parameters are crucial to the effectiveness of variational mode decomposition methods and their applications in various fields. Therefore, to improve the signal analysis quality and signal processing capabilities of modern signal processing systems, more effective signal preprocessing methods are needed. Summary of the Invention

[0003] In view of this, this application provides a frequency band division method, apparatus and electronic device based on empirical spectrum trends, which can adaptively achieve robust frequency band division and signal component center frequency estimation in complex noise environments according to the energy distribution characteristics of the input signal in the frequency domain. It can be used to support the feature extraction and parameter estimation of multi-component non-stationary signals in signal processing systems in practical engineering.

[0004] Firstly, this application provides a frequency band division method based on empirical spectral trends, applied to the analysis of multi-component non-stationary signals in complex noise environments. The method includes: extracting the observed signal spectral trend based on the amplitude spectrum of the input signal; the observed signal spectral trend includes a combination of the target signal spectral trend varying with frequency and the background noise spectral trend; the target signal includes mechanical vibration signals, biomedical signals, electromagnetic wave signals, acoustic signals, or artificial signals; estimating the mean background noise based on the amplitude spectrum of the input signal; extracting the background noise spectral trend based on the mean background noise; calculating the difference between the observed signal spectral trend and the background noise spectral trend to obtain the empirical spectral trend; the empirical spectral trend indicates the overall fluctuation of the target signal's spectral energy; and traversing the empirical spectral trend. The first critical point is determined. Sub-bands; the first critical point includes the boundary frequency of the spectrum, the local minimum point of the empirical spectrum trend, or the gradient abrupt change point; For natural numbers greater than or equal to 1; based on the frequency band division results, traverse the empirical spectrum trend. The second critical point is determined. The first center frequency; the second critical point includes the peak point of the spectrum, the peak point of the empirical spectrum trend, or the gradient abrupt change point; thereby realizing adaptive frequency band division and signal component center frequency estimation for multi-component non-stationary signals.

[0005] In some possible implementations, the empirical spectral trend is obtained by calculating the difference between the estimated spectral trend of the observed signal and the estimated spectral trend of the background noise, including: using... Gaussian smoothing was used to extract the spectral trend of the observed signal; Gaussian smoothing is used to extract the background noise spectrum trend; based on the difference between the observed signal spectrum trend and the background noise spectrum trend, an empirical spectrum trend is obtained.

[0006] In some possible implementations, calculating the difference between the observed signal spectrum trend and the background noise spectrum trend to obtain the empirical spectrum trend includes: performing a centrally symmetric mirror extension on the discrete signal sampled by the digital system; performing a discrete Fourier transform on the extended time series, retaining the non-negative frequency components to obtain a one-sided spectrum; and based on the one-sided spectrum, solving for the difference between the estimated value of the observed signal spectrum trend and the estimated value of the background noise spectrum trend in discrete form to obtain the empirical spectrum trend.

[0007] In some possible implementations, the method further includes estimating the mean background noise of the amplitude spectrum of the observed signal: performing a centrally symmetric mirror extension and discrete Fourier transform on the real-valued digital signal of the input signal, retaining non-negative frequency components to obtain a one-sided spectrum; performing mirror extension on both sides of the amplitude spectrum corresponding to the one-sided spectrum of the input signal to obtain a first frequency sequence; wherein the length of each side of the mirror extension is... ;pass Centered on, with a length of sliding window Calculate the local sample mean of the first frequency sequence. :

[0008]

[0009] in, For window within The second frequency sequence after sorting the data in ascending order; remove the sliding window. Internal value greater than the first threshold The data is used to calculate the mean background noise as follows:

[0010]

[0011] in, For the truncation length, satisfying ; The set experience threshold factor.

[0012] In some possible implementations, the traversal of the empirical spectrum trend The first critical point is determined. Sub-bands, including: those that determine empirical spectral trends A first critical point, including local minima; the local minima indicate the frequency points at which one energy peak ends and another energy peak begins in the amplitude spectrum of the observed signal; according to the... The first critical point is determined. Sub-band.

[0013] In some possible implementations, the empirical spectrum trend is traversed. The first critical point is determined. Each sub-band includes: the frequency corresponding to each local minimum point that determines the empirical spectral trend. , According to the first The local minimum point and the th local minimum point The frequency corresponding to the local minimum point is determined by the first... Sub-bands of each frequency band for:

[0014]

[0015] Among them, the The frequency corresponding to each local minimum point and the The frequency corresponding to each local minimum point Sub-band The two boundary frequencies.

[0016] In some possible implementations, the boundary frequency also includes the zero frequency and the Nyquist frequency.

[0017] In some possible implementations, the empirical spectrum trend is traversed. The second critical point is determined. The center frequencies include: determining the empirical spectral trend in the 1st... Sub-bands of each frequency band The peak point within corresponds to the frequency Based on peak points Determine the first Sub-bands of each frequency band The center frequency.

[0018] The adaptive frequency band division method provided in this application uses the mean of background noise to calculate the empirical spectrum trend, which is an estimate of the target signal spectrum trend. By using the empirical spectrum trend as the basis for frequency band division, it can adaptively identify different signal components and robustly determine the frequency band boundary under different noise environments, thereby realizing effective frequency band division and signal component center frequency estimation for multi-component non-stationary signals under complex noise backgrounds.

[0019] The adaptive frequency band division method provided in this application fully utilizes the characteristic of spectral trend intrinsically reflecting the spectral energy distribution to estimate the center frequency of signal components. Based on the energy distribution within the divided frequency band, the center frequency of the signal components is estimated, which significantly improves the estimation accuracy.

[0020] Secondly, this application provides a frequency band division device based on empirical spectral trends, applied to multi-component non-stationary signal processing in complex noise environments. The device includes: an amplitude spectrum analysis module for extracting the observed signal spectral trend based on the amplitude spectrum of the input signal; the observed signal spectral trend includes a combination of the target signal spectral trend varying with frequency and the background noise spectral trend; the target signal includes mechanical vibration signals, biomedical signals, electromagnetic wave signals, sound wave signals, or artificial signals; an empirical spectrum extraction module for calculating the difference between the observed signal spectral trend and the background noise spectral trend to obtain the empirical spectral trend; the empirical spectral trend indicates the overall fluctuation of the target signal spectral energy; and a frequency band division module for traversing the empirical spectral trend. The first critical point is determined. Each sub-band; the first critical point includes the boundary frequency of the spectrum, the local minimum point of the empirical spectrum trend, or the gradient abrupt change point; A natural number greater than or equal to 1; a boundary center determination module, used to traverse the empirical spectrum trend based on the frequency band division results. The second critical point is determined. A center frequency; the second critical point includes the peak point of the spectrum, the peak point of the empirical spectrum trend, or the gradient abrupt change point; to achieve adaptive frequency band division.

[0021] Thirdly, this application provides a computing device, comprising: at least one memory for storing a program; and at least one processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to execute a method as described in any of the first aspects.

[0022] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to perform the method as described in any of the first aspects. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the various embodiments disclosed in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only a few embodiments disclosed in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] The accompanying drawings used in the description of the embodiments or prior art are briefly introduced below.

[0025] Figure 1 A flowchart of a frequency band division method based on empirical spectral trends provided in this application embodiment;

[0026] Figure 2 This is a schematic diagram of the observed signal spectrum trend model provided in the embodiments of this application;

[0027] Figure 3 The diagram shows the time-domain waveform and amplitude spectrum of the target signal and the time-domain waveform and amplitude spectrum of the noisy signal provided in Embodiment 1 of this application.

[0028] Figure 4 A flowchart of the frequency band division method based on empirical spectrum trends provided in Embodiment 1 of this application;

[0029] Figure 5 This is a schematic diagram of a sliding window for estimating the mean background noise provided in Embodiment 1 of this application;

[0030] Figure 6 This is a schematic diagram of the frequency band division result provided in Embodiment 1 of this application;

[0031] Figure 7 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0032] Figure 8 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0033] Figure 9 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0034] Figure 10 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0035] Figure 11 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0036] Figure 12 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0037] Figure 13 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results;

[0038] Figure 14 This application provides a schematic diagram of a frequency band division device based on empirical spectrum trends.

[0039] Figure 15 This application provides a computing device in its embodiments. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be described below with reference to the accompanying drawings.

[0041] In the description of the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.

[0042] In the description of the embodiments in this application, the term "and / or" is merely a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, B existing alone, and A and B existing simultaneously. Furthermore, unless otherwise stated, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple terminals refer to two or more terminals.

[0043] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and their variations all mean "including but not limited to," unless otherwise specifically emphasized.

[0044] In the description of the embodiments in this application, "some embodiments" are mentioned, which describe a subset of all possible embodiments. However, it is understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0045] In the description of the embodiments of this application, the terms "first, second, third, etc." or module A, module B, module C, etc. are used only to distinguish similar objects and do not represent a specific ordering of objects. It is understood that, where permitted, a specific order or sequence can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0046] In the description of the embodiments of this application, the reference numerals for the steps, such as S110, S120, etc., do not necessarily indicate that the steps will be executed in this manner. Where permissible, the order of the steps can be interchanged or executed simultaneously.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0048] In recent years, much research has focused on preprocessing techniques applicable to adaptive signal decomposition methods, thereby improving signal decomposition performance in complex situations such as low signal-to-noise ratios and mixed multi-source signals. Preprocessing techniques for frequency domain decomposition methods mainly include frequency band partitioning based on sub-band boundary detection and frequency band partitioning based on spectral trends. Noise can easily cause significant interference to traditional frequency band partitioning methods. Especially in real-world scenarios, noise is usually not ideal white noise, but rather frequency-dependent colored noise; methods based solely on white noise modeling are unlikely to guarantee robustness in complex noise environments.

[0049] For example, adaptive frequency band division results based on scale space representation often have serious redundancy problems, and the method cannot obtain ideal results under high signal-to-noise ratio conditions. Low-pass filtering and Fourier model fitting methods are greatly affected by background noise. Overly smooth spectral trend extraction will cause different signal components to be classified into the same frequency band, while overly coarse spectral trends will cause excessive decomposition of the spectrum. For multi-component non-stationary signals in complex noise environments, it is difficult to find a suitable parameter to achieve effective frequency band division.

[0050] In view of this, this application proposes an adaptive frequency band division method based on empirical spectral trends. This method is based on modeling the observed signal spectral trend as a linear combination of the target signal spectral trend and the background noise spectral trend. First, the observed signal spectral trend is extracted based on the input signal amplitude spectrum. Then, the mean value of the background noise is estimated based on the input signal amplitude spectrum, and the background noise spectral trend is further extracted based on the mean value of the noise. As an estimate of the target signal spectral trend, the empirical spectral trend is obtained by calculating the difference between the estimated observed signal spectral trend and the estimated background noise spectral trend. Therefore, the empirical spectral trend indicates the overall fluctuation of the spectral energy of the target signal component in the observed signal. Finally, multiple sub-bands are divided based on the local minima of the empirical spectral trend and the boundary frequencies of the spectrum. The center frequency of the signal component is determined based on the peak point of the empirical spectral trend and the peak point of the spectrum in each sub-band, thereby realizing adaptive frequency band division and signal component center frequency estimation.

[0051] Since the empirical spectrum trend reflects the overall fluctuation of the target signal's spectral energy and has good noise robustness, the boundary frequency and center frequency of each sub-band component can be determined based on the local minimum value of the empirical spectrum trend, thus achieving adaptive frequency band division.

[0052] The frequency band division method based on empirical spectrum trends proposed in this application can be widely applied to technical scenarios such as mechanical vibration signal analysis, biomedical signal analysis, underwater acoustic signal feature extraction, weak signal detection and noise suppression, and can improve the robustness of the algorithm in complex noise environments.

[0053] The following description, in conjunction with the accompanying drawings and specific embodiments, further elaborates on a frequency band division method based on empirical spectrum trends proposed in this application.

[0054] Figure 1 A flowchart illustrating the frequency band division method based on empirical spectral trends provided in this application embodiment. Figure 1 As shown, it includes the following steps.

[0055] S11, extract the observation signal spectrum trend based on the amplitude spectrum of the input signal. The observation signal spectrum trend includes a linear combination of the target signal spectrum trend of frequency change and the background noise spectrum trend. The target signal includes mechanical vibration signal, biomedical signal, electromagnetic wave signal, sound wave signal or artificial signal. The target signal spectrum trend includes mechanical vibration signal spectrum trend, biomedical signal spectrum trend, electromagnetic wave signal spectrum trend, sound wave signal spectrum trend or artificial signal spectrum trend.

[0056] For example, Figure 2 This is a schematic diagram illustrating the trend of the observed signal spectrum provided in an embodiment of this application. Figure 2 As shown, the observed signal spectrum trend is a mixed signal that varies with frequency. The observed signal spectrum trend can be viewed as a linear combination of the target signal spectrum trend and the background noise spectrum trend, as shown in the formula. As shown:

[0057] (1)

[0058] in, To observe the trend of the signal spectrum, For the target signal spectrum trend, Background noise spectrum trend, For frequency variables.

[0059] For example, for the input signal Its amplitude spectrum The Fourier transform results can be obtained from it. Obtained by taking the modulus.

[0060] In some possible implementations, by analyzing the amplitude spectrum of the input signal Spectral trend extraction is performed to obtain the spectral trend of the observed signal. Spectral trend extraction methods include extracting spectral trends using mathematical morphology, extracting spectral trends using smoothing filtering, extracting spectral trends using linear predictive coding, or extracting spectral trends using low-rank sparse decomposition.

[0061] In some possible implementations, utilizing Gaussian smoothing for estimating the trend of the observed signal spectrum:

[0062] (2)

[0063] in, For penalty parameters; For estimating the trend of the observed signal spectrum; express Norm; This represents the input variable value that minimizes the output of the objective function; Used to limit the trend of observed signal spectrum Smoothness; Used to ensure the trend of the observed signal spectrum Indicates the spectral energy fluctuations caused by the main signal components.

[0064] S12, estimate the mean background noise based on the amplitude spectrum of the input signal.

[0065] In some possible implementations, for the input signal amplitude spectrum Methods for estimating the mean of background noise include sorting statistics, threshold segmentation, local smoothing, or iterative estimation.

[0066] In some possible implementations, threshold segmentation and median filtering are used to estimate the mean of the background noise. .

[0067] S13, extract the background noise spectrum trend based on the mean background noise.

[0068] In some possible implementations, by averaging the background noise Trend extraction is performed to obtain the background noise spectrum trend. Trend extraction methods include trend extraction using mathematical morphology, trend extraction using smoothing filtering, trend extraction using linear predictive coding, or trend extraction using low-rank sparse decomposition.

[0069] In some possible implementations, utilizing Gaussian smoothing for estimating the background noise spectrum trend:

[0070]

[0071] in, For estimating the trend of the observed signal spectrum; Used to limit the trend of background noise spectrum Smoothness; Used to ensure background noise spectrum trend The trend change is dominated by the frequency band where the indicator noise energy is concentrated.

[0072] S14, calculate the difference between the observed signal spectrum trend and the background noise spectrum trend to obtain the empirical spectrum trend.

[0073] For example, based on the above model, an empirical spectral trend is constructed as an estimate of the target signal spectral trend. This trend is calculated by subtracting the background noise spectral trend from the observed signal spectral trend, as shown in the formula. As shown:

[0074]

[0075] In some possible implementations, utilizing Gaussian smoothing computational empirical spectral trends:

[0076]

[0077] It can be based on the one-sided amplitude spectrum of the input signal , and based on the single-sided amplitude spectrum of the input signal The estimated mean background noise is The optimization problem of extracting empirical spectral trends is transformed into a discrete form, and the empirical spectral trends are obtained by solving the problem. .

[0078] S15 uses the first critical point of the empirical spectrum trend to divide the frequency band.

[0079] In some possible implementations, traversing the empirical spectrum trend The first critical point is determined. Each sub-band; the first critical point includes the boundary frequency of the spectrum, the local minimum point of the empirical spectrum trend, or the gradient abrupt change point; It is a natural number greater than or equal to 1.

[0080] For example, the empirical spectrum trend of The first critical point is determined as the sub-band boundary. , Using this boundary, the spectrum of the input signal is... Divided into Sub-band; the first critical point includes the spectral boundary frequency and empirical spectral trend. The local minimum point or gradient abrupt change point. The boundary frequencies include the zero frequency and the Nyquist sampling frequency.

[0081] In some possible implementations, it can be based on empirical spectrum trends. The local minimum points determine the sub-band boundaries. .

[0082] S16, using the second critical point of the empirical spectrum trend within the sub-band to determine the center frequency of the signal component.

[0083] In some possible implementations, based on the frequency band division results, the empirical spectrum trend is traversed. The second critical point is determined. The first center frequency; the second critical point includes the peak point of the spectrum, the peak point of the empirical spectrum trend, or the gradient abrupt change point.

[0084] For example, traversal Sub-band, for the first The second critical point of each sub-band is determined as the first... The center frequency of each signal component , .

[0085] In some possible implementations, it can be based on empirical spectrum trends. In the Peak points within each sub-band Determine the first The center frequency of each signal component .

[0086] The adaptive frequency band division method provided in this application embodiment is based on the observed signal spectrum trend model. It uses the mean of background noise to calculate the empirical spectrum trend and uses the empirical spectrum trend as the basis for frequency band division. It can adaptively achieve efficient frequency band division and signal component center frequency estimation for multi-component non-stationary signals under complex noise background.

[0087] The adaptive frequency band allocation method provided in this application embodiment can process the redundant estimation components by using nearest matching or other strategies.

[0088] The adaptive frequency band division method provided in this application estimates the center frequency of signal components based on the energy distribution within the divided frequency band. It makes full use of the characteristic that the intrinsic spectral trend reflects the spectral energy distribution to estimate the center frequency of signal components, thus significantly improving the estimation accuracy.

[0089] Example 1

[0090] The frequency band division method based on empirical spectrum trends proposed in Embodiment 1 of this application is widely used in underwater acoustic signal feature extraction technology and can improve the robustness of signal processing algorithms in complex noise environments.

[0091] Assumption Formula The given signal is a multi-component non-stationary underwater acoustic signal. The underwater acoustic signal to be processed From a linear frequency modulated signal Five cosine wave signals ~ and a periodic repetitive pulse signal The mixture consists of seven components:

[0092]

[0093] in, The number of repetitive pulse signals; For the first The time it takes for each pulse to be generated is determined by the characteristic frequency, which is 40 Hz in this case. It is a unit step function; the signal sampling rate is 8 kHz; the signal duration is 2 s.

[0094] For the formula , Figure 3Figure (a) shows the target signal. A schematic diagram of the time-domain waveform; Figure 3 (b) The amplitude spectrum of the target signal is given. Schematic diagram; amplitude spectrum Through the Fourier transform results Obtained by taking the modulus.

[0095] Figure 3 (b) shows that the main frequency components of the repetitive pulse signal spectrum are 1800~2200 Hz, and the signal components... , and The frequency domain distribution overlaps with the local energy concentration region of the background noise.

[0096] The input signal is the target signal The noisy mixed signal after adding background noise. Figure 3 (c) shows the time-domain waveform of the observed signal; Figure 3 (d) shows the amplitude spectrum of the input signal. Figure 3 (d) is Figure 3 (c) Obtained by taking the modulus after performing a Fourier transform.

[0097] Figure 3 (d) shows that the background noise has obvious frequency correlation and non-flat spectral distribution characteristics, with high low-frequency energy and local energy enhancement at about 1500 Hz, which is used to simulate a colored noise environment with frequency correlation and local energy concentration characteristics.

[0098] For example, Figure 4 This is a flowchart of the frequency band division method based on empirical spectral trends provided in Embodiment 1 of this application. Before executing this method, the input data, initialization parameters, and output target of the algorithm are determined; the input data is the discrete signal sampled by the digital system. ,in, The number of sampling points; initialization parameters include penalty parameters. Frequency sequence mirror extension length Experience threshold factor and experience threshold factor Output targets include underwater acoustic signal frequency band delineation boundaries. , Number of signal components and the center frequencies of each component , .like Figure 4 As shown, the frequency band division method based on empirical spectral trends includes the following steps:

[0099] S41, To overcome the boundary effect, the time series of the input signal is subjected to a centrally symmetric mirror extension to obtain the extended time series. For this extended time series Perform a Discrete Fourier Transform and retain the spectral components corresponding to the non-negative frequencies to obtain the one-sided spectrum. .

[0100] S42, based on the frequency sequence of the single-sided amplitude spectrum Calculate the mean background noise Specifically, in this embodiment, the average background noise is determined through the following steps S421-S425:

[0101] S421, in the amplitude spectrum Both sides are respectively length of The mirror continuation yields the first frequency sequence after continuation:

[0102]

[0103] S422, constructed with Centered on, with a length of sliding window This is used to calculate the local sample mean of the first frequency sequence.

[0104] For example, Figure 5 This is a schematic diagram of a sliding window for estimating the mean background noise provided in an embodiment of this application.

[0105] S423, for example Figure 5 The window shown within The data are sorted in ascending order to obtain the second frequency sequence:

[0106]

[0107] in, For the window The minimum value among the data. For the window The maximum value among the data.

[0108] S424, according to the second frequency sequence The local sample mean was calculated to be:

[0109]

[0110] S425, Values ​​within the window greater than the first threshold The data is excluded, according to the formula. Calculate the mean background noise:

[0111]

[0112] in, The truncation length is defined to satisfy: , ; This is an empirical threshold factor, typically ranging from 2 to 10. However, in practical engineering applications, it needs to be adjusted based on the noise level and signal characteristics. It should be adjusted appropriately. The mean of the second frequency sequence; This is an empirical threshold used to determine the cutoff length when estimating the global noise mean.

[0113] S43, utilizing Gaussian smoothing method, from the single-sided amplitude spectrum and background noise mean Extract the observed signal spectrum trend and the background noise spectrum trend, and calculate the difference between the observed signal spectrum trend and the background noise spectrum trend as the empirical spectrum trend. Specifically, the formula The optimization problem shown is transformed into solving the empirical spectral trend in the following discrete form:

[0114]

[0115] in, Denotes the Euclidean norm; It is the identity matrix; It is a second-order difference matrix:

[0116]

[0117] S44, using the local minima of the empirical spectral trend and the boundary frequencies of the spectrum, determines the sub-band boundaries, dividing the frequency band into K sub-bands:

[0118]

[0119] Among them, the sub-band boundary frequency The sub-band boundary includes the zero frequency and the Nyquist sampling frequency. The zero frequency and the Nyquist frequency are considered as the first and second boundary frequencies of the entire band, i.e., the first boundary... Second boundary ; It is a natural number greater than or equal to 1.

[0120] S45, Traversal Each sub-band is used to determine the center frequency of the signal component by utilizing the empirical spectral trend peak points and spectral peak points within the sub-band.

[0121] Specifically, for the first Sub-band peak point of empirical spectrum trend Corresponding experience threshold The second threshold, when the second threshold Greater than the peak value of the amplitude spectrum within the sub-band At that time, the first The center frequency of each signal component is determined as the frequency corresponding to the peak point of the empirical spectrum trend. Conversely, when the second threshold Less than or equal to the peak value of the amplitude spectrum within the sub-band At that time, the first The center frequency of each signal component is determined as the frequency corresponding to the peak point of the spectrum. .

[0122] in, This is an empirical threshold factor, typically ranging from 2 to 10, and its value can be the same as the empirical threshold factor mentioned in the previous steps. While maintaining consistency is important, in practical engineering applications, adjustments need to be made based on noise levels and signal characteristics. It should be adjusted appropriately; This represents the input variable value that maximizes the output of the objective function.

[0123] The final embodiment of this application is determined. The boundary frequency, number of signal components, and center frequency of each sub-band are used to achieve adaptive frequency band division and center frequency estimation of signal components.

[0124] For example, Figure 6 This diagram illustrates the bandwidth allocation result of the adaptive bandwidth allocation method based on empirical spectrum trends provided in this embodiment of the application. Figure 6 As shown, the extracted empirical spectrum trend accurately reflects the energy distribution of the signal, effectively overcoming the interference caused by the non-uniform distribution of the power spectral density of colored noise, thus achieving ideal adaptive frequency band division results and signal component center frequency estimation results.

[0125] Finally, the results were analyzed and the performance was verified.

[0126] Numerical simulation experiments were conducted to evaluate multi-component non-stationary signals under colored noise background, different signal-to-noise ratio conditions, and different parameter settings to verify the effectiveness and robustness of the adaptive frequency band division method provided in the embodiments of this application.

[0127] To further verify the effectiveness of the frequency band division method based on empirical spectral trends, numerical simulation experiments were used to test the center frequency estimation results of each signal component of the noisy signal (1) under different signal-to-noise ratio (SNR) conditions. At the same time, the influence of different parameter settings on the estimation performance was examined. The estimation performance of the frequency band division method based on empirical spectral trends in this application was compared and analyzed with the adaptive frequency band division method based on scale space representation, the adaptive frequency band division method based on low-pass filtering to extract trends, and the adaptive frequency band division method based on Fourier model fitting to extract spectral trends.

[0128] Assuming noisy underwater acoustic signal Total Each signal component, in Embodiment 1 of this application Including a linear frequency modulated signal Five cosine wave signals ~ and periodic repetitive pulse signals Seven components. The experiment was conducted using the Monte Carlo method. To account for the negative impact of over-division of frequency bands, the algorithm's performance was analyzed based on the relative root mean-square error (RRMSE).

[0129]

[0130] in, Number of Monte Carlo simulations; For the first The actual carrier frequency of each signal component; The total number of signal components obtained by dividing the frequency bands in the algorithm. Among them, the closest to the first The estimated number of signal components for each real signal component, therefore, ; The center frequency of the signal component estimated by the algorithm.

[0131] The analysis is based on the relative root mean square error, and the analysis results of each signal component are evaluated separately to ensure that the performance analysis of the algorithm is quantitative, scientific, and accurate.

[0132] The results of the four methods for estimating the center frequencies of the seven signal components are as follows: Figures 7-13 As shown.

[0133] Specifically, Figure 7 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 7(a) An adaptive frequency band partitioning method based on the maximum inter-class variance criterion and empirical criteria for scale-space representation is used for signal components. Center frequency estimation results.

[0134] Figure 7 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results, including filter parameters This is the cutoff frequency of the low-pass filter.

[0135] Figure 7 (c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. The center frequency estimation results, where the parameters are... Let be the order of the Fourier basis.

[0136] Figure 7 (d) The frequency band division method based on empirical spectrum trends proposed in the embodiments of this application is used for signal components. Center frequency estimation results.

[0137] Depend on Figure 7 It can be seen that due to signal components The frequency band in question is subject to strong background noise interference. The adaptive frequency band partitioning methods based on scale space representation, low-pass filtering to extract spectral trends, and Fourier model fitting to extract spectral trends all result in large errors in the estimated center frequencies of the signal components. These errors do not improve with increasing signal-to-noise ratio (SNR), indicating that the performance of existing methods is limited by the non-uniformity of the background noise energy distribution, making them difficult to work effectively in colored noise environments. In contrast, the frequency band partitioning method based on empirical spectral trends in this application exhibits excellent noise robustness and low sensitivity to changes in algorithm parameters, i.e., parameter robustness. In particular, when the SNR is higher than -18 dB, this method can achieve highly accurate center frequency estimation under various parameter configurations.

[0138] Figure 8 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 8 (a) shows the adaptive frequency band division method based on scale space representation for signal components. Center frequency estimation results.

[0139] Figure 8 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results;

[0140] Figure 8(c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. Center frequency estimation results;

[0141] Figure 8 (d) is the frequency band division method based on empirical spectrum trends in this application for signal components. Center frequency estimation results.

[0142] Depend on Figure 8 It can be seen that, although signal components The background noise interference in the current frequency band is relatively weak. However, due to the wide energy distribution and insignificant peak of the broadband signal, existing methods directly use the sub-band center frequency or the peak frequency of the spectrum as the estimated center frequency of the signal component. Therefore, these methods have inherent limitations when processing broadband signals, resulting in a large deviation in the estimated center frequency. This deviation does not improve even with an increased signal-to-noise ratio, indicating that the performance of existing methods is limited by the energy distribution characteristics of the signal component itself and is difficult to apply to broadband signal scenarios. In contrast, the frequency band division method based on empirical spectral trends in this application accurately determines the center frequency of the signal component by effectively extracting the spectral energy distribution characteristics of the input signal. It is applicable to various signal types, including narrowband and broadband signals, and exhibits excellent broadband signal applicability, noise robustness, and parameter robustness. It can achieve highly accurate center frequency estimation under various parameter configurations.

[0143] Figure 9 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 9 (a) shows the adaptive frequency band division method based on scale space representation for signal components. Center frequency estimation results;

[0144] Figure 9 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results;

[0145] Figure 9 (c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. Center frequency estimation results;

[0146] Figure 9 (d) is the frequency band division method based on empirical spectrum trends in this application for signal components. Center frequency estimation results.

[0147] Figure 10 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 10(a) shows the adaptive frequency band division method based on scale space representation for signal components. Center frequency estimation results;

[0148] Figure 10 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results;

[0149] Figure 10 (c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. Center frequency estimation results;

[0150] Figure 10 (d) is the frequency band division method based on empirical spectrum trends in this application for signal components. Center frequency estimation results.

[0151] Figure 11 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 11 (a) shows the adaptive frequency band division method based on scale space representation for signal components. Center frequency estimation results;

[0152] Figure 11 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results;

[0153] Figure 11 (c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. Center frequency estimation results;

[0154] Figure 11 (d) is the frequency band division method based on empirical spectrum trends in this application for signal components. Center frequency estimation results.

[0155] Figure 12 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 12 (a) shows the adaptive frequency band division method based on scale space representation for signal components. Center frequency estimation results;

[0156] Figure 12 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results;

[0157] Figure 12(c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. Center frequency estimation results;

[0158] Figure 12 (d) is the frequency band division method based on empirical spectrum trends in this application for signal components. Center frequency estimation results.

[0159] Depend on Figure 9 , 10 As can be seen from 11 and 12, due to the signal components , , , In continuous frequency bands, the energy distribution is similar to that of background noise, making it difficult for existing methods to effectively detect signal components. This leads to inaccurate sub-band division: either insufficient division results in the merging of multiple adjacent components into the same sub-band, or excessive division generates multiple invalid sub-bands without effective signals. Consequently, the accuracy of signal component center frequency estimation is insufficient, and this does not improve with increased signal-to-noise ratio. This indicates that the performance of existing methods is limited by the distinguishability of signal and noise frequency domain energy distributions, making them difficult to apply effectively to scenarios with complex background noise. In contrast, the frequency band division method based on empirical spectrum trends in this application exhibits excellent environmental adaptability, noise robustness, and parameter robustness. It can achieve highly accurate center frequency estimation under various parameter configurations.

[0160] Figure 13 Four methods for analyzing signal components A schematic diagram of the center frequency estimation results. Figure 13 (a) shows the adaptive frequency band division method based on scale space representation for signal components. Center frequency estimation results;

[0161] Figure 13 (b) shows the adaptive frequency band division method based on low-pass filtering extraction trend for signal components. Center frequency estimation results;

[0162] Figure 13 (c) shows the adaptive frequency band division method for signal components based on Fourier model fitting to extract spectral trends. Center frequency estimation results;

[0163] Figure 13 (d) is the frequency band division method based on empirical spectrum trends in this application for signal components. Center frequency estimation results.

[0164] Depend on Figure 13 It can be seen that although signal components The background noise interference in the frequency band is relatively weak, but its frequency domain energy distribution is complex and exhibits a multi-peak state. Existing methods are prone to generating multiple invalid sub-bands due to over-division, resulting in a large error in the estimated center frequency of the signal components. Moreover, the improvement is limited when the signal-to-noise ratio is increased. This indicates that the performance of existing methods is limited by the complexity of the energy distribution of the signal components themselves, making them difficult to apply to scenarios with non-stationary characteristics such as amplitude-modulated signals. In contrast, the frequency band division method based on empirical spectrum trends in this application is applicable to various signal types, including narrowband signals, wideband signals, and amplitude-modulated signals. It exhibits excellent adaptability to non-stationary signals, noise robustness, and parameter robustness, and can achieve highly accurate center frequency estimation under various parameter configurations.

[0165] In summary, existing frequency band division methods are sensitive to the energy distribution characteristics of signals and background noise in the frequency domain. When dealing with scenarios where multiple signal components, non-stationary signals, and complex background noise coexist, it is difficult to accurately detect the effective signal components and determine their center frequencies, resulting in inaccurate frequency band division. This manifests as insufficient division, such as merging multiple signal components, or excessive division, such as generating sub-bands that do not contain effective signals.

[0166] Inaccurate segmentation further degrades the center frequency estimation performance, specifically manifested as: the estimated center frequencies of some signal components deviating significantly from the true values, or the estimation results fluctuating greatly and lacking stability under different signal-to-noise ratios. It is noteworthy that even with improved signal-to-noise ratios, the improvement in these problems is limited, indicating that the performance bottleneck of existing methods does not stem from the noise intensity itself, but rather from their limited ability to effectively detect multi-component signals in complex noise backgrounds and their adaptability to non-stationary signals.

[0167] In contrast, the method provided in this application demonstrates superior performance in adaptive frequency band division and signal component center frequency estimation, and has broad signal class adaptability, strong noise robustness and strong parameter robustness. It can achieve high-precision and high-stability center frequency estimation under various noise environments and parameter configurations.

[0168] The above results fully demonstrate that the adaptive frequency band division method provided in this application can effectively improve signal analysis efficiency and signal component center frequency estimation accuracy when processing multi-component non-stationary signals under complex noise backgrounds, and is significantly better than other existing methods, showing its effectiveness and advantages in practical applications.

[0169] The empirical spectrum trend extracted in this application accurately reflects the energy distribution of the input signal, effectively overcoming the background noise interference caused by the non-uniform distribution of power spectral density, thereby achieving an ideal adaptive frequency band division result.

[0170] In this way, the frequency band division method based on empirical spectral trends proposed in this application fully utilizes the characteristic of spectral trends intrinsically reflecting the distribution of spectral energy to determine the center frequency of signal components, rather than directly using the center frequency of sub-bands or the frequency of spectral peak points to determine the center frequency of signal components.

[0171] The above is an introduction to the signal component center frequency method provided in the embodiments of this application. It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. In addition, in some possible implementations, each step in the above embodiments may be selectively executed according to the actual situation, and may be partially or fully executed, without limitation here. Furthermore, all or part of any feature of any of the above embodiments may be freely and arbitrarily combined without contradiction; the combined technical solution is also within the scope of this application.

[0172] Next, based on the above, the frequency band division device based on empirical spectrum trends provided in the embodiments of this application will be introduced. For relevant descriptions of concepts, formulas, etc., involved in the following content, please refer to the above text.

[0173] Figure 14 This application provides a schematic diagram of a frequency band division device based on empirical spectrum trends. Figure 14 As shown, the frequency band division device 140 based on empirical spectrum trends includes an amplitude spectrum analysis module 141, an empirical spectrum extraction module 142, a frequency band division module 143, and a boundary center determination module 144.

[0174] This device can be widely applied to multi-component non-stationary signal processing scenarios in complex noise environments. The amplitude spectrum analysis module 141 is used to obtain the observed signal spectrum trend based on the amplitude spectrum of the input signal. The observed signal spectrum trend includes a combination of the target signal spectrum trend varying with frequency and the background noise spectrum trend; the target signal includes mechanical vibration signals, biomedical signals, electromagnetic wave signals, sound wave signals, or artificial signals; the empirical spectrum extraction module 142 calculates the difference between the observed signal spectrum trend and the background noise spectrum trend to obtain the empirical spectrum trend; the empirical spectrum trend indicates the overall fluctuation of the target signal's spectral energy; the frequency band division module 143 traverses the empirical spectrum trend... Determining the critical point Sub-bands; critical points include local minima or gradient abruptness points; For natural numbers greater than or equal to 1; boundary center determination module 144 determines... The boundary frequency and center frequency of each sub-band are determined to achieve adaptive frequency band division.

[0175] As an example of a software functional unit, the amplitude spectrum analysis module 141 may include code running on a computing instance. The computing instance may include at least one of a physical host, such as a computing device, a virtual machine, or a container. Further, the aforementioned computing instance may be one or more. For example, the amplitude spectrum analysis module 141 may include code running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers used to run the code may be distributed in the same region or in different regions. Further, the multiple hosts / virtual machines / containers used to run the code may be distributed in the same availability zone (AZ) or in different AZs, each AZ including one data center or multiple geographically proximate data centers. Typically, a region may include multiple AZs.

[0176] Similarly, multiple hosts / virtual machines / containers used to run this code can be distributed within the same Virtual Private Cloud (VPC) or across multiple VPCs. Typically, a VPC is set up within a region. Communication between two VPCs within the same region, as well as between VPCs in different regions, requires a communication gateway to be set up within each VPC to enable interconnection between VPCs.

[0177] This application also provides a computing device 150. For example... Figure 15 As shown, the computing device 150 includes a bus 152, a processor 154, a memory 156, and a communication interface 158. The processor 154, the memory 156, and the communication interface 158 communicate with each other via the bus 152. The computing device 150 can be a computing device or a terminal device. It should be understood that this application does not limit the number of processors and memories in the computing device 150.

[0178] Bus 152 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 15 A single line may be used to represent a bus, but this does not mean that there is only one bus or one type of bus. Bus 154 may include a path for transmitting information between various components of computing device 150, such as memory 156, processor 154, and communication interface 158.

[0179] Processor 154 may include any one or more processors such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0180] Memory 156 may include volatile memory, such as random access memory (RAM). Processor 154 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0181] The memory 156 stores executable program code, and the processor 154 executes the executable program code to implement the aforementioned functions. Figure 13 The amplitude spectrum analysis module 131 shown functions to implement all or part of the steps of the method in the above embodiments. That is, the memory 156 stores instructions for executing all or part of the steps in the method of the above embodiments.

[0182] The communication interface 158 uses transceiver modules, such as, but not limited to, network interface cards and transceivers, to enable communication between the computing device 150 and other devices or communication networks.

[0183] This application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to perform the method as described in any of the first aspects.

[0184] It is understood that the processor in the embodiments of this application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor may be a microprocessor or any conventional processor.

[0185] The method steps in the embodiments of this application can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0186] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means, such as coaxial cable, optical fiber, digital subscriber line (DSL), or wireless means, such as infrared, wireless, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid-state disk (SSD), etc.

[0187] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application.

Claims

1. A frequency band division method based on empirical spectral trends, applied to multi-component non-stationary signal processing in complex noise environments, characterized in that, The method includes: The observed signal spectrum trend is extracted based on the amplitude spectrum of the input signal. The observed signal spectrum trend includes a combination of the target signal spectrum trend and the background noise spectrum trend that vary with frequency. The target signal includes mechanical vibration signals, biomedical signals, electromagnetic wave signals, sound wave signals, or artificial signals. Estimate the mean background noise based on the amplitude spectrum of the input signal; Extract the background noise spectral trend based on the background noise mean; An empirical spectral trend is obtained by calculating the difference between the observed signal spectral trend and the background noise spectral trend; the empirical spectral trend indicates the overall fluctuation of the target signal spectral energy. Traversing the aforementioned empirical spectrum trends The first critical point is determined. Each sub-band; the first critical point includes the boundary frequency of the spectrum, the local minimum point of the empirical spectrum trend, or the gradient abrupt change point; It is a natural number greater than or equal to 1; Based on the frequency band division results, the empirical spectrum trend is traversed. The second critical point is determined. A center frequency; the second critical point includes the peak point of the spectrum, the peak point of the empirical spectrum trend, or the gradient abrupt change point; to achieve adaptive frequency band division.

2. The method according to claim 1, characterized in that, Calculating the difference between the observed signal spectral trend and the background noise spectral trend to obtain the empirical spectral trend includes: use Gaussian smoothing was used to extract the spectral trend of the observed signal; use Gaussian smoothing was used to extract the spectral trend of the background noise; The empirical spectral trend is obtained by solving for the difference between the observed signal spectral trend and the background noise spectral trend.

3. The method according to claim 1 or 2, characterized in that, The method further includes estimating the mean background noise of the one-sided amplitude spectrum of the input signal: The input signal is subjected to a centrally symmetric mirror extension and discrete Fourier transform, and non-negative frequency components are retained to obtain a single-sided spectrum. The first frequency sequence is obtained by mirroring the amplitude spectrum on both sides of the single-sided spectrum of the input signal; the length of each mirror extension is... ; Using a length of sliding window Calculate the local sample mean of the first frequency sequence; For window within The second frequency sequence is obtained by arranging the data in ascending order. Remove sliding window For data with an internal value greater than the first threshold, calculate the mean background noise.

4. The method according to claim 1 or 2, characterized in that, The traversal of the empirical spectrum trend The first critical point is determined. Each sub-band includes: Determining the trend of empirical spectrum A first critical point, such as a local minimum point; the local minimum point indicates the frequency point at which one energy peak ends and another energy peak begins in the amplitude spectrum of the observed signal; According to the above The first critical point is determined. Sub-band.

5. The method according to claim 1 or 2, characterized in that, The traversal of the empirical spectrum trend The first critical point is determined. Each sub-band includes: Determine the frequency corresponding to each first critical point of the empirical spectral trend. , ; According to the The first critical point and the first critical point The frequency corresponding to the first critical point determines the first... Sub-band for: Among them, the first The first critical point corresponds to the frequency and the first The first critical point corresponds to the frequency Sub-band The two boundary frequencies.

6. The method according to claim 1, characterized in that, The boundary frequencies include the zero frequency and the Nyquist frequency.

7. The method according to claim 1 or 2, characterized in that, The traversal of the empirical spectrum trend The second critical point is determined. One center frequency, including: Determine the empirical spectral trend in the first Sub-bands of each frequency band The peak point within corresponds to the frequency ; According to the frequency corresponding to the peak point Determine the first Sub-bands of each frequency band The center frequency.

8. A frequency band division device based on empirical spectral trends, applied to multi-component non-stationary signal processing in complex noise environments, characterized in that, The device includes: The amplitude spectrum analysis module is used to extract the spectrum trend of the observed signal based on the amplitude spectrum of the input signal. The observed signal spectrum trend includes a combination of the target signal spectrum trend and the background noise spectrum trend that vary with frequency. The target signal includes mechanical vibration signals, biomedical signals, electromagnetic wave signals, sound wave signals, or artificial signals. An empirical spectrum extraction module is used to estimate the mean background noise based on the amplitude spectrum of the input signal; extract the background noise spectrum trend based on the mean background noise; calculate the difference between the observed signal spectrum trend and the background noise spectrum trend to obtain the empirical spectrum trend; the empirical spectrum trend indicates the overall fluctuation of the target signal spectral energy. The frequency band division module is used to traverse the empirical spectrum trend. The first critical point is determined. Each sub-band; the first critical point includes the boundary frequency of the spectrum, the local minimum point of the empirical spectrum trend, or the gradient abrupt change point; It is a natural number greater than or equal to 1; The boundary center determination module is used to traverse the empirical spectrum trend. The second critical point is determined. A center frequency; the second critical point includes the peak point of the spectrum, the peak point of the empirical spectrum trend, or the gradient abrupt change point; to achieve adaptive frequency band division.

9. A computing device, comprising: At least one memory for storing programs; At least one processor is configured to execute a program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to perform the method as described in any one of claims 1-8.

10. A storage medium storing instructions that, when executed on a terminal, cause a first terminal to perform the method as described in any one of claims 1-8.

Images