Polymorphic signal processing method based on neighborhood maximum extraction transformation

Abstract

The invention discloses a polymorphic signal processing method based on neighborhood maximum extraction transformation, and the method comprises the following steps: S1, setting a proper sampling frequency, and collecting a vibration signal of a planetary gear box through an acceleration sensor; s2, processing the planetary gear box vibration signal by using short-time Fourier transform, and obtaining a corresponding time-frequency spectrum result; s3, defining a multi-neighborhood maximum detection algorithm based on the local energy characteristics of the time-frequency spectrum obtained in S2; s4, defining a neighborhood maximum extraction operator in combination with the multi-neighborhood maximum detection algorithm in the S3 and the time-frequency trajectory features; s5, adaptively extracting a time-frequency coefficient on a short-time Fourier transform time-frequency plane by using the neighborhood maximum extraction operator constructed in the step S4 to obtain neighborhood maximum extraction transform; and S6, judging the working condition state of the planetary gearbox according to the time-frequency spectrum result of the maximum extraction transformation of the neighborhood. According to the method, the energy aggregation of the time-frequency spectrum of the polymorphic signal is improved, the noise robustness is relatively high, and a powerful guarantee is provided for judging the working condition state of a machine containing the polymorphic characteristic vibration signal.

Description

Technical Field

[0001] This invention belongs to the field of signal processing, specifically relating to a multimorphic signal processing method based on neighborhood maximum extraction transform. Background Technology

[0002] Rotating machinery plays a vital role in aviation, transportation, agriculture, and other fields. Because these machines operate under heavy loads and fatigue conditions for extended periods, their core components and critical structures inevitably suffer varying degrees of damage. Condition monitoring of rotating machinery primarily relies on sensors to collect vibration signals and then analyzing these signals to obtain relevant fault information. With the increasing complexity of the equipment's operating environment, monitoring signals often exhibit multi-modal characteristics, combining harmonic and transient signals. Effectively identifying and extracting key information from these multi-modal signals remains a significant technical challenge.

[0003] Time-frequency analysis methods are widely used to process multi-mode signals because they can simultaneously characterize time-domain information and frequency components. Traditional time-frequency analysis methods, however, cannot provide highly readable time-frequency spectra due to the Heisenberg uncertainty principle and cross-term interference. Therefore, researchers have proposed post-processing techniques to obtain highly concentrated energy time-frequency spectra by rearranging time-frequency coefficients on the time-frequency plane to the instantaneous frequency or group delay trajectory. Redistribution methods simultaneously redistribute time-frequency coefficients along both frequency and time directions, improving the energy concentration of the time-frequency spectrum. However, because these methods lose crucial phase information, they cannot reconstruct the signal back to the time domain, and their time-frequency resolution still has room for improvement. Time-frequency multiple compression transform and time-frequency extraction transform, based on frequency modulation partitioning criteria, perform mode partitioning at each time-frequency position, and then use unidirectional time-frequency analysis methods for processing and fusion, improving the resolution of the time-frequency spectrum of multi-mode signals. Both of these algorithms require mode partitioning and unidirectional processing at each time-frequency point, reducing overall computational efficiency, and energy concentration still needs further improvement.

[0004] In summary, although existing time-frequency analysis methods have been applied to multi-mode signal processing in the condition monitoring of rotating machinery, their energy concentration and computational efficiency still need further improvement. Summary of the Invention

[0005] To address the shortcomings of the existing technologies, this invention provides a multimorphic signal processing method based on neighborhood maximum extraction transform, which can significantly improve the energy concentration of the time spectrum and the computational efficiency of the algorithm, and achieve accurate representation of multimorphic signals.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a multimorphic signal processing method based on neighborhood maximum extraction transform, comprising the following steps:

[0007] Step S1: Set an appropriate sampling frequency and use an accelerometer to collect vibration signals from the planetary gearbox;

[0008] Step S2: Process the planetary gearbox vibration signal using short-time Fourier transform and obtain the corresponding time-frequency spectrum results;

[0009] Step S3: Define a multi-neighborhood maximum detection algorithm based on the local energy features of the time spectrum obtained in S2;

[0010] Step S4: Combine the multi-neighbor maximum detection algorithm and time-frequency trajectory features in S3 to define the neighborhood maximum extraction operator;

[0011] Step S5: Adaptively extract the time-frequency coefficients on the short-time Fourier transform time-frequency plane using the neighborhood maximum extraction operator constructed in S4 to obtain the neighborhood maximum extraction transform;

[0012] Step S6: Determine the operating condition of the planetary gearbox based on the time-frequency spectrum result of the neighborhood maximum extraction transform.

[0013] As a further improvement of the present invention, in step S2, the time-spectrum result of the short-time Fourier transform is as follows:

[0014]

[0015] in, This indicates the collected vibration signal of the planetary gearbox. Indicates the center of time. Indicates the frequency center. It represents the imaginary unit. Let Gaussian window function be defined in the time domain, and its expression is:

[0016] As a further improvement of the present invention, in step S3, a multi-neighborhood maximum detection algorithm is defined based on the local energy features of the time spectrum, and its expression is:

[0017]

[0018] in, and These represent discrete indices for the frequency center and time center, respectively. yes The modulus.

[0019] As a further improvement of the present invention, in step S4, a neighborhood maximum extraction operator is defined by combining the multi-neighborhood maximum detection algorithm and time-frequency trajectory features:

[0020]

[0021] in, This represents the summation function.

[0022] As a further improvement of the present invention, in step S5, the time-frequency coefficients on the short-time Fourier transform time-frequency plane are adaptively extracted based on the neighborhood maximum extraction operator to obtain the neighborhood maximum extraction transform, the expression of which is:

[0023]

[0024] in, It is the maximum extraction operator in the neighborhood.

[0025] Compared with existing technologies, the multimorphic signal processing method based on neighborhood maximum extraction transform proposed in this invention has the following advantages: It uses short-time Fourier transform to process the planetary gearbox vibration signal and obtains the corresponding time-frequency spectrum results; it defines a multi-neighborhood maximum detection algorithm based on the local energy characteristics of the time-frequency spectrum obtained in S2; it defines a neighborhood maximum extraction operator by combining the multi-neighborhood maximum detection algorithm in S3 and the time-frequency trajectory characteristics; it adaptively extracts the time-frequency coefficients on the short-time Fourier transform time-frequency plane using the neighborhood maximum extraction operator constructed in S4, obtaining the neighborhood maximum extraction transform; and it determines the operating condition of the planetary gearbox based on the time-frequency spectrum results of the neighborhood maximum extraction transform. This invention provides a novel multimorphic signal processing method that can achieve accurate characterization of multimorphic signals, with good energy concentration and high computational efficiency. Attached Figure Description

[0026] Figure 1 This is a flowchart of an embodiment of the present invention.

[0027] Figure 2 This is a diagram of the experimental setup in an embodiment of the present invention.

[0028] Figure 3 This is a time-domain waveform diagram of the vibration signal of the planetary gearbox in an embodiment of the present invention.

[0029] Figure 4 This is the time-frequency spectrum result obtained by performing time-frequency analysis on the vibration signal of the planetary gearbox according to the present invention.

[0030] Figure 5 The time-frequency spectrum results are obtained by performing time-frequency analysis on the vibration signal of the planetary gearbox using the redistribution method.

[0031] Figure 6 The time-frequency spectrum results are obtained by performing time-frequency multiple compression transform on the vibration signal of the planetary gearbox.

[0032] Figure 7 The time-frequency spectrum results are obtained by performing time-frequency extraction and transformation on the vibration signal of the planetary gearbox.

[0033] Figure 8This is a slice diagram of the transient components of the time spectrum obtained using four methods in an embodiment of the present invention.

[0034] Figure 9 This is a schematic diagram of the normalized energy of the time spectrum obtained using four methods in an embodiment of the present invention. Detailed Implementation

[0035] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0036] Example 1:

[0037] See Figure 1 A multimorphic signal processing method based on neighborhood maximum extraction transform includes the following steps:

[0038] Step S1: Set an appropriate sampling frequency and use an accelerometer to collect vibration signals from the planetary gearbox;

[0039] Step S2: Process the planetary gearbox vibration signal using short-time Fourier transform and obtain the corresponding time-frequency spectrum results;

[0040] Step S3: Define a multi-neighborhood maximum detection algorithm based on the local energy features of the time spectrum obtained in S2;

[0041] Step S4: Combine the multi-neighbor maximum detection algorithm and time-frequency trajectory features in S3 to define the neighborhood maximum extraction operator;

[0042] Step S5: Adaptively extract the time-frequency coefficients on the short-time Fourier transform time-frequency plane using the neighborhood maximum extraction operator constructed in S4 to obtain the neighborhood maximum extraction transform;

[0043] Step S6: Determine the operating condition of the planetary gearbox based on the time-frequency spectrum result of the neighborhood maximum extraction transform.

[0044] As a further improvement of the present invention, in step S2, the time-spectrum result of the short-time Fourier transform is as follows:

[0045]

[0046] in, This indicates the collected vibration signal of the planetary gearbox. Indicates the center of time. Indicates the frequency center. It represents the imaginary unit. Let Gaussian window function be defined in the time domain, and its expression is:

[0047] As a further improvement of the present invention, in step S3, a multi-neighborhood maximum detection algorithm is defined based on the local energy features of the time spectrum, and its expression is:

[0048]

[0049] in, and These represent discrete indices for the frequency center and time center, respectively. yes The modulus.

[0050] As a further improvement of the present invention, in step S4, a neighborhood maximum extraction operator is defined by combining the multi-neighborhood maximum detection algorithm and time-frequency trajectory features:

[0051]

[0052] in, This represents the summation function.

[0053] As a further improvement of the present invention, in step S5, the time-frequency coefficients on the short-time Fourier transform time-frequency plane are adaptively extracted based on the neighborhood maximum extraction operator to obtain the neighborhood maximum extraction transform, the expression of which is:

[0054]

[0055] in, It is the maximum extraction operator in the neighborhood.

[0056] Example 2:

[0057] The planetary gearbox vibration signal used in this embodiment was acquired from... Figure 2 The planetary gearbox experimental platform shown is composed of a speed-increasing gearbox, a reduction gearbox, a load device, a motor, a coupling, and an accelerometer. Vibration signals are measured using an accelerometer mounted on top of the gearbox. The sampling frequency of the planetary gearbox vibration signal is 16384 Hz. A 0.9 s segment of the measurement signal is extracted for further analysis, and its time-domain waveform is shown below. Figure 3 As shown. To improve computational efficiency and reveal the low-frequency characteristics of the signal, the selected signal is downsampled, and the new sampling frequency is 2400 Hz.

[0058] The time-frequency energy distribution results obtained by the method provided by this invention are as follows: Figure 4 As shown in the figure, the time-frequency ridges in the time-frequency spectrum obtained by the neighborhood maximum extraction transform are continuous and clear, with high energy concentration, and can accurately characterize both harmonic and transient signals simultaneously. The natural frequency fluctuations remain relatively stable within a reasonable range, indicating that the natural frequencies are normal and there is no resonance or fault. The time-frequency spectrum results of the planetary gearbox vibration signal processed using the redistribution method are shown in the figure. Figure 5As shown, energy is dispersed around the time-frequency ridge, indicating poor time-frequency cohesion. The results of the time-frequency multiple compression transform and time-frequency extraction transform are as follows. Figure 6 and 7 As shown, although it can provide a time-frequency spectrum with concentrated energy for harmonic and transient components, a large number of non-redistribution points still exist, reducing the time-frequency resolution.

[0059] The transient components are sliced, and the results are as follows: Figure 8 As shown in (a), the transient component appears at 0.727 s and does not exhibit obvious periodicity. Its amplitude and frequency are not significantly high either. Therefore, it can be preliminarily determined that the planetary gearbox is in a healthy state. The transient components from the three comparison methods are sliced, and the results are shown below. Figure 8 As shown in (b)-(d), it can be observed that the occurrence time of the transient component in the comparison method is shifted, and the occurrence time cannot be accurately located.

[0060] To quantitatively evaluate the energy concentration of the proposed method, this invention introduces Rényi entropy and normalized energy as evaluation indicators. The smaller the Rényi entropy value, the better the energy concentration of the time-frequency spectrum. The Rényi entropies for the neighborhood maximum extraction transform, the redistribution method, the time-frequency multiple compression transform, and the time-frequency extraction transform are 1.545, 3.196, 1.686, and 2.422, respectively. Among them, the neighborhood maximum extraction transform has the smallest Rényi entropy value, indicating the best energy concentration. Secondly, normalized energy is used to evaluate the energy concentration of the time-frequency spectrum. The faster the normalized energy reaches 1, the higher the energy concentration of the time-frequency spectrum. The normalized energy results for different time-frequency analysis methods are shown below. Figure 9 As shown. Compared with other methods, the method proposed in this embodiment reaches the normalized energy of 1 the fastest, indicating that it has the best energy concentration.

[0061] Furthermore, to illustrate the computational efficiency of the proposed method, the computation times of different methods are presented in this invention. The computation times for the neighborhood maximum extraction transform, the redistribution method, the time-frequency multiple compression transform, and the time-frequency extraction transform are 0.34 s, 0.45 s, 0.98 s, and 0.47 s, respectively. The method proposed in this invention has the shortest computation time, indicating that its computational efficiency is superior to other comparative methods.

[0062] The embodiments described above are merely illustrative of the technical solutions of the present invention and are not intended to limit the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A multimorphic signal processing method based on neighborhood maximum extraction transform, characterized in that, Includes the following steps: Step S1: Set an appropriate sampling frequency and use an accelerometer to collect vibration signals from the planetary gearbox; Step S2: Process the planetary gearbox vibration signal using short-time Fourier transform and obtain the corresponding time-frequency spectrum results; Step S3: Define a multi-neighborhood maximum detection algorithm based on the local energy features of the time spectrum obtained in S2; Step S4: Combine the multi-neighbor maximum detection algorithm and time-frequency trajectory features in S3 to define the neighborhood maximum extraction operator; Step S5: Adaptively extract the time-frequency coefficients on the short-time Fourier transform time-frequency plane using the neighborhood maximum extraction operator constructed in S4 to obtain the neighborhood maximum extraction transform; Step S6: Determine the operating condition of the planetary gearbox based on the time-frequency spectrum result of the neighborhood maximum extraction transform.

2. The multimorphic signal processing method based on neighborhood maximum extraction transform according to claim 1, characterized in that, In step S2, the time-frequency spectrum result of the short-time Fourier transform is as follows: ； in, This indicates the collected vibration signal of the planetary gearbox; Indicates the center of time. Indicates the frequency center. Represents the imaginary unit; Let Gaussian window function be defined in the time domain, and its expression is: .

3. The multimorphic signal processing method based on neighborhood maximum extraction transform according to claim 1, characterized in that, In step S3, a multi-neighborhood maximum detection algorithm is defined based on the local energy features of the time spectrum, and its expression is: ； in, and Discrete indices representing the frequency center and time center, respectively; yes The modulus.

4. The multimorphic signal processing method based on neighborhood maximum extraction transform according to claim 1, characterized in that, In step S4, the neighborhood maximum extraction operator is defined by combining the multi-neighborhood maximum detection algorithm and time-frequency trajectory features: ； in, This represents the summation function.

5. The multimorphic signal processing method based on neighborhood maximum extraction transform according to claim 1, characterized in that, In step S5, the time-frequency coefficients on the short-time Fourier transform time-frequency plane are adaptively extracted based on the neighborhood maximum extraction operator to obtain the neighborhood maximum extraction transform, the expression of which is: ； in, It is the maximum extraction operator in the neighborhood.

Images