Search

CN-121980236-A - Frequency band dividing method and device based on empirical spectrum trend and electronic equipment

CN121980236ACN 121980236 ACN121980236 ACN 121980236ACN-121980236-A

Abstract

The application relates to the technical field of signal processing, and provides a frequency band dividing 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 steps of obtaining an observed signal spectrum trend according to an amplitude spectrum of an input signal, wherein the observed 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, an acoustic wave signal or an artificial signal, an empirical spectrum trend is obtained by calculating the difference between the observed signal spectrum trend and the background noise spectrum trend, the empirical spectrum trend indicates the overall fluctuation of the spectral energy of the target signal, a frequency band is divided by traversing a first critical point of the empirical spectrum trend, the center frequency of each signal component is determined by traversing a second critical point of the empirical spectrum trend based on the frequency band division result, and therefore the self-adaptive frequency band division and the signal component center frequency estimation of the multi-component nonstationary signal are realized.

Inventors

  • HAO CHENGPENG
  • XUE MENG
  • YAN CHENG

Assignees

  • 中国科学院声学研究所

Dates

Publication Date
20260505
Application Date
20251226

Claims (10)

  1. 1. A frequency band division method based on empirical spectrum trend, applied to multi-component non-stationary signal processing in a complex noise environment, characterized in that the method comprises: extracting an observed signal spectrum trend based on an amplitude spectrum of an input signal, wherein the observed signal spectrum trend comprises a combination of a target signal spectrum trend and a background noise spectrum trend which change along with frequency, and the target signal comprises a mechanical vibration signal, a biomedical signal, an electromagnetic wave signal, an acoustic wave signal or an artificial signal; estimating a background noise mean value according to the amplitude spectrum of the input signal; extracting a background noise spectrum trend based on the background noise mean value; calculating the difference between the observed signal spectral trend and the background noise spectral trend to obtain an empirical spectral trend, wherein the empirical spectral trend indicates the overall fluctuation of the spectral energy of the target signal; Traversing the empirical spectrum trend First critical point determination The first critical point comprises boundary frequency of a frequency spectrum, local minimum value points of empirical spectrum trend or gradient abrupt change points; A natural number greater than or equal to 1; traversing the empirical spectrum trend based on the frequency band division result Second critical point determination The second critical point comprises a spectrum peak point, an empirical spectrum trend peak point or a gradient abrupt change point, and the self-adaptive frequency band division is realized.
  2. 2. The method of claim 1, wherein calculating the difference between the observed signal spectral trend and the background noise spectral trend to obtain an empirical spectral trend comprises: By using Gaussian smooth extraction of the observed signal spectrum trend; By using Gaussian smooth extraction of the background noise spectrum trend; And solving to obtain an empirical spectrum trend based on the difference between the observed signal spectrum trend and the background noise spectrum trend.
  3. 3. The method according to claim 1 or 2, further comprising estimating a background noise mean of a single-side amplitude spectrum of the input signal: performing central symmetry image continuation and discrete Fourier transform according to a real-value digital signal of an input signal, and reserving non-negative frequency components to obtain a single-side frequency spectrum; Image continuation is carried out according to two sides of an amplitude spectrum corresponding to a single-side frequency spectrum of an input signal to obtain a first frequency sequence, wherein the length of the image continuation at each side is ; By length of Sliding window of (2) Calculating a local sample mean of the first frequency sequence; window-to-window In the inner part The data are arranged in ascending order to obtain a second frequency sequence; Removing sliding windows And calculating the average value of the background noise according to the data with the internal value larger than the first threshold value.
  4. 4. A method according to claim 1 or 2, wherein said traversing said empirical spectrum trend First critical point determination A sub-band comprising: determining empirical spectrum trends The local minimum value points indicate frequency points of the end of one energy peak value and the beginning of the other energy peak value in the amplitude spectrum of the observation signal; According to the described First critical point determination A sub-band.
  5. 5. A method according to claim 1 or 2, wherein said traversing said empirical spectrum trend First critical point determination A sub-band comprising: determining a frequency corresponding to each first critical point of the empirical spectrum , ; According to the first First critical point and second critical point The first critical point corresponding frequency is determined Sub-frequency bands The method comprises the following steps: Wherein the first The first critical point corresponds to the frequency And said first The first critical point corresponds to the frequency Is a sub-band Is not shown in the figure).
  6. 6. The method of claim 1, wherein the boundary frequencies comprise a zero frequency and a nyquist frequency.
  7. 7. A method according to claim 1 or 2, wherein said traversing said empirical spectrum trend Second critical point determination A center frequency, comprising: Determining that the empirical spectrum trend is at the first Sub-bands of individual frequency bands Peak point in corresponds to frequency ; According to the peak point corresponding frequency Determining the first Sub-bands of individual frequency bands Is set at the center frequency of (a).
  8. 8. A band division apparatus based on empirical spectrum trends, applied to multi-component non-stationary signal processing in a complex noise environment, comprising: The amplitude spectrum analysis module is used for extracting an observation signal spectrum trend based on the amplitude spectrum of the input signal, wherein the observation signal spectrum trend comprises a combination of a target signal spectrum trend and a background noise spectrum trend which change along with frequency, and the target signal comprises a mechanical vibration signal, a biomedical signal, an electromagnetic wave signal, an acoustic wave signal or an artificial signal; the system comprises an input signal, an empirical spectrum extraction module, an empirical spectrum analysis module, a background noise spectrum trend extraction module and a target signal spectrum energy analysis module, wherein the input signal is used for receiving an input signal, the background noise spectrum is used for receiving an observed signal spectrum, the empirical spectrum extraction module is used for estimating a background noise average value according to an amplitude spectrum of the input signal, extracting a background noise spectrum trend based on the background noise average value, calculating a difference between the observed signal spectrum trend and the background noise spectrum trend to obtain an empirical spectrum trend, and the empirical spectrum trend indicates overall fluctuation of the target signal spectrum energy; A frequency band dividing module for traversing the empirical spectrum trend First critical point determination The first critical point comprises boundary frequency of a frequency spectrum, local minimum value points of empirical spectrum trend or gradient abrupt change points; A natural number greater than or equal to 1; a boundary center determining module for traversing the empirical spectrum trend Second critical point determination The second critical point comprises a spectrum peak point, an empirical spectrum trend peak point or a gradient abrupt change point, and the self-adaptive frequency band division is realized.
  9. 9. A computing device comprising at least one memory for storing a program, at least one processor for executing the program stored in the memory, the processor for performing the method of any of claims 1-8 when the program stored in the memory is executed.
  10. 10. A storage medium having stored therein instructions which, when run on a terminal, cause a first terminal to perform the method of any of claims 1-8.

Description

Frequency band dividing method and device based on empirical spectrum trend and electronic equipment Technical Field The present invention relates to the field of signal processing, and in particular, to a frequency band division method, apparatus and electronic device based on empirical spectrum trend. Background The core task of signal processing is to extract and identify the effective components from complex observations and suppress or remove the mixed interference components in the signals, so as to provide reliable input for subsequent applications such as parameter estimation, target identification and decision support. Along with the coming of the whole large data age and the rapid development of intelligent information detection technology, the informatization level of the modern society is continuously improved, and the requirements on higher efficiency, accuracy, safety, self-adaptability and the like are provided for the digital signal processing technology of various instruments and equipment. In order to meet the requirements and various requirements such as real-time performance and robustness in engineering application, domestic and foreign scholars research and develop modern signal processing methods such as time sequence analysis, spectral kurtosis, envelope modulation, sparse representation, self-adaptive signal decomposition and the like. The adaptive signal decomposition method can decompose any complex multi-component signal into a series of single-component signals without considering prior transformation. At present, the conventional adaptive signal decomposition method can be roughly divided into four types of time domain decomposition method, iterative filtering decomposition method, time-frequency decomposition method and frequency domain decomposition method from the aspect of the principle of signal decomposition. The frequency domain decomposition method decomposes signals from the angle of the frequency domain, and has direct and effective processing effect on signals without time-frequency cross. However, the existing classical frequency domain decomposition method lacks the utilization of spectrum morphology information, the frequency bands are divided according to the distribution of the amplitude spectrum by methods such as empirical wavelet transformation, adaptive empirical Fourier decomposition and the like, the result is extremely easily influenced by interference information, the performance of the variation modal decomposition and the variation method thereof is closely related to the selection of model parameters, and the variation model parameters such as the decomposition modal number, the modal initial center frequency and the like are very important for the effectiveness of the variation modal decomposition method and the application of the variation modal decomposition method in various fields. Therefore, to improve the signal analysis quality and signal processing capability of modern signal processing systems, more efficient signal preprocessing methods are needed. Disclosure of Invention In view of the above, the present application provides a frequency band dividing method, apparatus and electronic device based on empirical spectrum trend, which can adaptively realize robust frequency band division and signal component center frequency estimation in a complex noise environment according to the energy distribution characteristic of an input signal in a frequency domain, and can be used for supporting the feature extraction and parameter estimation of a signal processing system in actual engineering on a multi-component non-stationary signal. In a first aspect, the application provides a frequency band division method based on an empirical spectrum trend, which is applied to multi-component non-stationary signal analysis in a complex noise environment, and comprises the steps of extracting an observed signal spectrum trend based on an amplitude spectrum of an input signal, wherein the observed signal spectrum trend comprises a combination of a target signal spectrum trend and a background noise spectrum trend which change along with frequency, the target signal comprises a mechanical vibration signal, a biomedical signal, an electromagnetic wave signal, an acoustic wave signal or an artificial signal, estimating a background noise mean value according to the amplitude spectrum of the input signal, extracting the background noise spectrum trend based on the background noise mean value, calculating a difference value between the observed signal spectrum trend and the background noise spectrum trend to obtain the empirical spectrum trend, indicating the overall fluctuation of spectral energy of the target signal, traversing the empirical spectrum trendFirst critical point determinationThe first critical point comprises a boundary frequency of a frequency spectrum, a local minimum point of an empirical spectrum trend or a gradient abrupt change point; A natural