A Signal Time-Frequency Analysis Method Based on Improved Multiple Synchronous Compression Transform
By improving the multiple synchronous compression transform method and combining short-time fractional Fourier transform and multiple synchronous compression transform, the problems of insufficient time-frequency resolution and noise sensitivity in the time-frequency analysis of non-stationary signals are solved, and higher accuracy time-frequency analysis results are achieved.
Patent Information
- Application Number
- CN202310848357.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-07-12
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-07-12
AI Technical Summary
Existing time-frequency analysis methods suffer from insufficient time-frequency resolution and sensitivity to noise when processing non-stationary signals, making it difficult to achieve high-precision time-frequency analysis.
An improved multi-synchronous compression transform method is adopted, which combines short-time fractional Fourier transform and multi-synchronous compression transform. The optimal transform order is selected by applying Gaussian window function and minimum norm method, and multi-synchronous compression energy rearrangement is performed until the iteration termination condition of Ruili entropy difference is less than the threshold is met.
It achieves higher time-frequency resolution and noise robustness, provides more accurate joint distribution information of signal time-domain and frequency-domain characteristics, and improves energy concentration.
Smart Images

Figure CN117009726B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of non-stationary signal analysis and processing technology, and in particular to a signal time-frequency analysis method based on an improved multiple synchronous compression transform. Background Technology
[0002] Non-stationary signals, such as bearing vibration signals, are common in signal processing. Time-frequency analysis (TFA), a common method for processing non-stationary signals, is an effective tool for characterizing the time-frequency characteristics of ubiquitous non-stationary signals. This type of method transforms a one-dimensional time-domain signal waveform into a two-dimensional time-frequency (TF) plane, thereby providing joint distribution information in the time and frequency domains and reflecting the local variation characteristics of non-stationary signals. Traditional TFA methods include linear TFA and quadratic TFA methods. The former provides limited time-frequency resolution, while the latter suffers from significant cross-terms when analyzing multi-component signals, severely impacting the readability of the results. To improve the performance of traditional TFA methods in processing non-stationary signals, post-processing techniques such as redistribution (RM) and multiple synchronous compression transform (MSST) have been proposed in recent years. However, these methods still suffer from insufficient time-frequency expression accuracy and sensitivity to noise.
[0003] In summary, existing time-frequency analysis methods still have shortcomings in processing non-stationary signals and need further improvement to achieve high-precision time-frequency analysis. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a signal time-frequency analysis method based on improved multiple synchronous compression transform, aiming to improve the accuracy and energy concentration of non-stationary signal processing.
[0005] The technical solution adopted in this invention is as follows: A signal time-frequency analysis method based on improved multiple synchronous compression transform includes: Step 1: Input signal The downsampling signal is obtained by performing downsampling processing. , t Represents time; Step 2: Downsample the signal Windowing is performed using a Gaussian window function to obtain a windowed segmented signal. , k The total number of windowed segmented signals; Step 3: Minimum search using second-order search Norm method determines each windowed segmented signal The optimal order of the fractional Fourier transform ; Step 4: Calculate each windowed segment signal Optimal Transformation Order The fractional Fourier transform of the signal yields the fractional spectrum of the transformed segmented signal. , f Represents frequency; Step 5: Calculate the fractional spectrum of each segmented signal after fractional Fourier transform. By combining the results, the short-time fractional Fourier transform time-frequency distribution is obtained. ; Step 6: Perform a short-time fractional Fourier transform on the time-frequency distribution. Multiple synchronous energy rearrangement is performed until the termination condition is met, and the time-frequency diagram of the output signal is obtained.
[0006] The further technical solution is as follows: In step two, the downsampled signal is... Windowing is performed using a Gaussian window function to obtain a windowed segmented signal. ,include: For downsampled signals Segmentation to obtain k The segmented signal, of which the first segment... i (1≤ i ≤ k ) segmented signals downsampled signal middle signal point to Data from signal points, L The length of the Gaussian window function and the sliding step size. H = L / a, where a is a set constant; Will and Gaussian window function Multiplying yields a windowed segmented signal. ,Right now .
[0007] In step two, the Gaussian window function used is:
[0008] in, It is the standard deviation of the Gaussian window function.
[0009] In step three, a second-order search for the minimum is performed. Norm method determines each windowed segmented signal The optimal order of the fractional Fourier transform ,include: The first order performs a coarse search, and then the order is transformed. The search interval is obtained at equal intervals of 0.1 steps. For windowed segmented signals Calculate each in sequence The fractional Fourier transform of the order of the signal yields the fractional spectrum of the transformed signal. (1≤ n ≤21), select norm Minimum value corresponds to order ; The second order performs a fine search, and the transformation order is then... The search interval is obtained at equal intervals of 0.001 steps. For windowed segmented signals Calculate each in sequence The fractional Fourier transform of the order of the signal yields the fractional spectrum of the transformed signal. (1≤ n ≤199), select norm Minimum value corresponds to order As the optimal transformation order .
[0010] In step five, the fractional spectrum of the segmented signal after fractional Fourier transform is... The combination method is as follows: Fractional spectrum of each segmented signal As 3D column vectors, forming 3D matrix , L is the length of the Gaussian window function.
[0011] In step six, the time-frequency distribution is based on the short-time fractional Fourier transform. The formula for iterative synchronous compression energy rearrangement is as follows:
[0012] in, yes After iteration N The time-frequency representation obtained by the next synchronous compressed energy rearrangement. , They are After iteration N The time and frequency variables of the time-frequency representation obtained by the subsynchronous compression energy rearrangement; It is the Dirac function. for The frequency variable represented by time-frequency, yes After iteration N Instantaneous frequency estimate of the signal obtained by subsynchronous compressed energy rearrangement:
[0013] in, The signal passes through The instantaneous frequency estimate of the transformation is calculated using the following formula: .
[0014] In step six, the iteration termination condition is: Ruili entropy difference between two adjacent iterations Less than the threshold ,in It is the first i The entropy difference in Ruili corresponding to the next iteration , This represents the final number of iterations. , They are pass Second-rate, The Ruili entropy of the time-frequency distribution obtained by performing synchronous compression energy rearrangement operation in -1 iterations. It is a constant.
[0015] The beneficial effects of this invention are as follows: This invention achieves a more accurate time-frequency representation of non-stationary signals by combining the Short-Time Fractional Fourier Transform (STFrFT) and Multiple Synchronous Compression Transform (MSST). In the implementation of the STFrFT, a minimum... By combining norms with a second-order search method to select the optimal transformation order, better transformation results are achieved.
[0016] Compared to traditional methods, this invention provides higher time-frequency resolution, better noise robustness, and more accurate joint distribution information of the signal's time and frequency domain characteristics. It also exhibits high accuracy and energy concentration.
[0017] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. Attached Figure Description
[0018] Figure 1 This is a flowchart of the method according to an embodiment of the present invention.
[0019] Figure 2 This is a time-frequency analysis result diagram of the signal multiple synchronous compression transformation in Verification Example 1 of this invention.
[0020] Figure 3 This is a time-frequency analysis result diagram of the improved multiple synchronous compression transformation of the signal in Verification Example 1 of this invention.
[0021] Figure 4 This is a time-frequency analysis result diagram of the signal multiple synchronous compression transformation in Verification Example 2 of this invention.
[0022] Figure 5 This is a time-frequency analysis result diagram of the improved multiple synchronous compression transformation of the signal in Verification Example 2 of this invention. Detailed Implementation
[0023] The specific embodiments of the present invention are described below with reference to the accompanying drawings.
[0024] See Figure 1 A signal time-frequency analysis method based on improved multiple synchronous compression transform includes: Step 1: Input signal The downsampling signal is obtained by performing downsampling processing. , t Represents time.
[0025] Step 2: Downsample the signal Windowing is performed using a Gaussian window function to obtain a windowed segmented signal. , k The total number of windowed segmented signals, specifically including: For downsampled signals Segmentation to obtain k The segmented signal, of which the first segment... i (1≤ i ≤ k ) segmented signals downsampled signal middle signal point to Data from signal points, L The length of the Gaussian window function and the sliding step size. H = L / a, where a is a set constant; Will and Gaussian window function Multiplying yields a windowed segmented signal. ,Right now .
[0026] Gaussian window function used for:
[0027] in, It is the standard deviation of the Gaussian window function.
[0028] Step 3: Minimum search using second-order search Norm method determines each windowed segmented signal The optimal order of the fractional Fourier transform Specifically, it includes: The first order performs a coarse search, and then the order is transformed. The search interval is obtained at equal intervals of 0.1 steps. For windowed segmented signals Calculate each in sequence (1≤ n The fractional Fourier transform of order ≤21 yields the transformed signal. (1≤ n ≤21), select ( p =0.8) norm Minimum value corresponds to order ; The second order performs a fine search, and the transformation order is then... The search interval is obtained at equal intervals of 0.001 steps. For windowed segmented signals Calculate each in sequence (1≤ n The fractional Fourier transform of order ≤199 yields the fractional spectrum of the transformed signal. (1≤ n ≤199), select ( p =0.8) norm Minimum value corresponds to order As the optimal transformation order .
[0029] in, , , This represents the fractional spectrum of the signal after fractional Fourier transform. The first in m Each element.
[0030] At the optimal transformation order down signal It can achieve optimal energy concentration within the fractional domain, thus achieving the best energy concentration effect.
[0031] Step 4: Calculate each windowed segment signal Optimal Transformation Order The fractional Fourier transform of the signal yields the fractional spectrum of the transformed segmented signal. Specifically, the formula for calculating the fractional Fourier transform is as follows:
[0032] in It is the kernel function of the fractional Fourier transform:
[0033] in , p It is the order of the fractional Fourier transform. n It is an integer; Step 5: Calculate the fractional spectrum of each segmented signal after fractional Fourier transform. By combining the results, the short-time fractional Fourier transform time-frequency distribution is obtained. Specific combination methods: Fractional spectrum of each segmented signal As 3D column vectors, forming 3D matrix .
[0034] Step 6: Perform a short-time fractional Fourier transform on the time-frequency distribution. Multiple synchronous energy rearrangement is performed until the termination condition is met, and the time-frequency diagram of the output signal is obtained.
[0035] Specifically, the formula for iterative synchronous compression energy rearrangement is:
[0036] in, yes After iteration N The time-frequency representation obtained by the next synchronous compressed energy rearrangement. , They are After iteration N The time and frequency variables of the time-frequency representation obtained by the subsynchronous compression energy rearrangement; It is the Dirac function. for The frequency variable represented by time-frequency, yes After iteration N Instantaneous frequency estimate of the signal obtained by subsynchronous compressed energy rearrangement:
[0037] in, The signal passes through The instantaneous frequency estimate of the transformation is calculated using the following formula: .
[0038] Specifically, the iteration termination condition is: Ruili entropy difference between two adjacent iterations Less than the threshold ,in, It is the first i The entropy difference in Ruili corresponding to the next iteration , This represents the final number of iterations. , They are pass Second-rate, The Ruili entropy of the time-frequency distribution obtained by performing synchronous compression energy rearrangement operation in -1 iterations. It is a constant.
[0039] The formula for calculating Ruili entropy is as follows:
[0040] in, It is a time-frequency distribution. It is the order of Ruili entropy, which is taken as 3 in this embodiment.
[0041] This embodiment uses the Short-Time Fractional Fourier Transform (STFrFT) as the time-frequency basis of the Multiple Synchronous Compression Transform (MSST). By introducing the energy concentration capability of STFrFT, it overcomes the shortcomings of MSST in processing emphasis-frequency signals. Furthermore, the synchronous compression iteration termination criterion based on Rayleigh entropy difference makes the algorithm more flexible and applicable. Therefore, the improved multiple synchronous compression transform can better focus on the emphasis frequency of the signal, providing more accurate time-frequency energy concentration and a more reliable time-frequency description. It can be widely used for processing non-stationary signals such as bearing vibration signals.
[0042] The effectiveness of the method in this embodiment will be verified below with examples.
[0043] Verification Example 1 The input signal is a single-component emphasis simulation signal, and its analytical expression is:
[0044] The signal sampling frequency is 4096Hz and the sampling time is 1s.
[0045] Figure 2 This is a time-frequency analysis result graph after multiple synchronous compression transformations. As can be seen from the graph, the signal exhibits energy divergence at points of rapid frequency change, resulting in low time-frequency resolution.
[0046] Figure 3 The figure shows the time-frequency analysis results after the improved multiple synchronous compression transform. As can be seen from the figure, the improved method proposed in this embodiment can greatly improve the energy concentration of the signal and make the time-frequency distribution more readable.
[0047] Verification Example 2 The input signal is a multi-component frequency-modulated simulation signal, and its analytical expression is:
[0048] in , , , , , .
[0049] The signal sampling frequency is 16384Hz and the sampling time is 1s.
[0050] Figure 4 This is a time-frequency analysis result graph after multiple synchronous compression transformation. As can be seen from the graph, due to the multi-component and time-varying nature of the signal, the multiple synchronous compression transformation algorithm exhibits energy divergence and cannot accurately identify the frequency components in the signal.
[0051] Figure 5 The figure shows the time-frequency analysis results after the improved multiple synchronous compression transform. As can be seen from the figure, the improved method proposed in this embodiment has superior performance in component discrimination and time-frequency concentration of multi-component time-varying frequency signals.
[0052] It will be understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A signal time-frequency analysis method based on improved multiple synchronous compression transform, characterized in that, include: Step 1: Input signal The downsampling signal is obtained by performing downsampling processing. , t Represents time; Step 2: Downsample the signal Windowing is performed using a Gaussian window function to obtain a windowed segmented signal. , k The total number of windowed segmented signals; Step 3: Minimum search using second-order search Norm method determines each windowed segmented signal The optimal order of the fractional Fourier transform ; Step 4: Calculate each windowed segment signal Optimal Transformation Order The fractional Fourier transform of the signal yields the fractional spectrum of the transformed segmented signal. , f Represents frequency; Step 5: Calculate the fractional spectrum of each segmented signal after fractional Fourier transform. By combining the results, the short-time fractional Fourier transform time-frequency distribution is obtained. ; Step 6: Perform a short-time fractional Fourier transform on the time-frequency distribution. Perform multiple synchronous compressed energy rearrangements until the termination condition is met, and output the time-frequency diagram of the signal. In step two, the downsampled signal is... Windowing is performed using a Gaussian window function to obtain a windowed segmented signal. ,include: For downsampled signals Segmentation to obtain k The segmented signal, of which the first segment... i (1≤ i ≤ k ) segmented signals downsampled signal middle signal point to Data from signal points, L The length of the Gaussian window function and the sliding step size. H = L / a, where a is a set constant; Will and Gaussian window function Multiplying yields a windowed segmented signal. ,Right now ; In step three, a second-order search for the minimum is performed. Norm method determines each windowed segmented signal The optimal order of the fractional Fourier transform ,include: The first order performs a coarse search, followed by a transformation of the order. The search interval is obtained at equal intervals of 0.1 steps. For windowed segmented signals Calculate each in turn The fractional Fourier transform of the order of the signal yields the fractional spectrum of the transformed signal. (1≤ n ≤21), select norm Minimum value corresponds to order subscript p =0.8; The second order performs a fine search, changing the order... The search interval is obtained at equal intervals of 0.001 steps. For windowed segmented signals Calculate each in turn The fractional Fourier transform of the order of the signal yields the fractional spectrum of the transformed signal. (1≤ n ≤199), select norm Minimum value corresponds to order As the optimal transformation order subscript p =0.8; In step six, the time-frequency distribution is based on the short-time fractional Fourier transform. The formula for iterative synchronous compression energy rearrangement is as follows: in, yes After iteration N The time-frequency representation obtained by the next synchronous compressed energy rearrangement. , They are After iteration N The time and frequency variables of the time-frequency representation obtained by the subsynchronous compression energy rearrangement; It is the Dirac function. for The frequency variable represented by time-frequency, yes After iteration N Instantaneous frequency estimate of the signal obtained by subsynchronous compressed energy rearrangement: in, The signal passes through The instantaneous frequency estimate of the transformation is calculated using the following formula: 。 2. The signal time-frequency analysis method based on improved multiple synchronous compression transform according to claim 1, characterized in that, In step two, the Gaussian window function used is: in, It is the standard deviation of the Gaussian window function.
3. The signal time-frequency analysis method based on improved multiple synchronous compression transform according to claim 1, characterized in that, In step five, the fractional spectrum of the segmented signal after fractional Fourier transform is... The combination method is as follows: Fractional spectrum of each segmented signal As 3D column vectors, forming 3D matrix , L is the length of the Gaussian window function.
4. The signal time-frequency analysis method based on improved multiple synchronous compression transform according to claim 1, characterized in that, In step six, the iteration termination condition is: Ruili entropy difference between two adjacent iterations Less than the threshold ,in , They are the first i sequence i- The entropy difference in Ruili corresponding to one iteration , This represents the final number of iterations. , They are pass Second-rate, The Ruili entropy of the time-frequency distribution obtained by performing synchronous compression energy rearrangement operation in -1 iterations. It is a constant.
