A method and device for extracting steady-state signals of wind turbine gearbox signals

By using discrete time-frequency curves and time-frequency analysis, the gearbox signal of the wind turbine is isolated and spliced, solving the problem of inaccurate steady-state signal extraction in the existing technology and improving diagnostic accuracy.

CN117763333BActive Publication Date: 2024-11-05CHINA RESOURCES POWER TECH RES INST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311701622.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-11
Publication Date
2024-11-05
Estimated Expiration
2043-12-11

AI Technical Summary

Technical Problem

Existing FFT algorithms and order analysis methods cannot accurately extract steady-state signals in wind turbine gearbox fault diagnosis, resulting in insufficient signal resolution and signal-to-noise ratio, which affects diagnostic accuracy.

Method used

By acquiring the discrete time-frequency curve of the wind turbine gearbox signal, the intervals of frequency fluctuation within a preset threshold are determined, and these intervals are separated and spliced. Combined with time-frequency analysis and Gabor transform, the steady-state signal is extracted.

Benefits of technology

It successfully extracts the stable speed portion from non-stationary signals, reduces information filtering, improves the accuracy of wind turbine gearbox and main bearing diagnosis, and is suitable for long waveform signal processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117763333B_ABST
    Figure CN117763333B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and device for extracting a steady-state signal of a wind turbine gearbox signal. The extraction method comprises: obtaining a discrete time-frequency curve according to the wind turbine gearbox signal; determining a plurality of intervals according to the discrete time-frequency curve, wherein the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold in each of the plurality of intervals; and isolating and splicing the wind turbine gearbox signals corresponding to the plurality of intervals to obtain a steady-state signal of the wind turbine gearbox signal. The present invention performs a time-frequency analysis on the wind turbine gearbox signal, and then finds the part of the wind turbine gearbox signal waveform with a stable speed through the time-frequency analysis, thereby solving the technical problem of inaccurate extraction of the steady-state signal of the wind turbine gearbox signal in the prior art.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine gearbox signal steady-state signal extraction, and more specifically, to a wind turbine gearbox signal steady-state signal extraction method and device. Background Art

[0002] The transmission chain (the main shaft, gearbox and generator bearing of the wind turbine) is a key component of the wind turbine. Its health status directly affects the operating reliability and safety of the unit. At present, the fault diagnosis is generally carried out by measuring and analyzing the vibration signal of the transmission chain. Affected by the operating characteristics of wind turbines with variable wind speed and working conditions, the vibration signals of wind turbines are mostly non-stationary signals with frequency modulation.

[0003] Currently, the mainstream methods for wind turbine gearbox fault diagnosis and vibration analysis are FFT (Fast Fourier Transform) analysis and order analysis methods.

[0004] The FFT algorithm has limitations in analyzing non-stationary signals. It cannot describe the change of frequency over time, that is, there will be frequency modulation and energy leakage in the spectrum. Since the fault diagnosis of gearboxes has high requirements for signal resolution and signal-to-noise ratio, it is difficult to obtain sufficient fault information through FFT algorithm analysis.

[0005] The order analysis method requires a speed waveform with sufficient accuracy as a resampling method. The CMS system currently used in wind power does not have a speed waveform with corresponding accuracy. The extracted speed waveform lacks accuracy, resulting in the order analysis accuracy involved in the current wind power vibration analysis being generally insufficient. Summary of the invention

[0006] One of the purposes of the present invention is to provide a method for extracting steady-state signals of a wind turbine gearbox signal, so as to solve the technical problem that the existing FFT algorithm is inaccurate in extracting steady-state signals of a wind turbine gearbox signal; another purpose of the present invention is to provide a device for extracting steady-state signals of a wind turbine gearbox signal; another purpose of the present invention is to provide a method for self-diagnosis of faults of a wind turbine gearbox

[0007] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0008] The present invention provides a method for extracting a steady-state signal of a wind turbine gearbox signal, comprising the following steps:

[0009] Obtain discrete time-frequency curves based on wind turbine gearbox signals;

[0010] Determine a plurality of intervals according to the discrete time-frequency curve, in each of the plurality of intervals, the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold;

[0011] The wind turbine gearbox signals corresponding to the plurality of intervals are isolated and spliced ​​to obtain steady-state signals of the wind turbine gearbox signals.

[0012] In the above-mentioned technical means, a time-frequency analysis is performed on the wind turbine gearbox signal, and then the part with stable speed in the wind turbine gearbox signal waveform is found through the time-frequency analysis, and then the spectrum analysis of the signal is performed. The signal obtained in this way is a standard stable signal, which solves the technical problem that the existing FFT algorithm extracts inaccurate steady-state signals of wind turbine gearbox signals.

[0013] Furthermore, the step of isolating and splicing the wind turbine gearbox signals corresponding to the plurality of intervals to obtain a steady-state signal of the wind turbine gearbox signal comprises the following steps:

[0014] Using a rectangular window function, the wind turbine gearbox signals corresponding to the plurality of intervals are isolated;

[0015] The wind turbine gearbox signals after multiple sections are spliced ​​to obtain the steady-state signal of the wind turbine gearbox signal.

[0016] Furthermore, the step of obtaining a discrete time-frequency curve according to the wind turbine gearbox signal comprises the following steps:

[0017] Obtain wind turbine gearbox signal;

[0018] Performing bandpass filtering on the wind turbine gearbox signal within a preset filtering interval;

[0019] Perform time-frequency analysis on the wind turbine gearbox signal after bandpass filtering to obtain the time-frequency analysis results;

[0020] The time-frequency analysis result is subjected to time-frequency peak search and smoothing processing to obtain a discrete time-frequency curve.

[0021] Furthermore, performing time-frequency analysis on the wind turbine gearbox signal after bandpass filtering to obtain the time-frequency analysis result includes the following steps:

[0022] Perform Gabor transform on the wind turbine gearbox signal after bandpass filtering;

[0023] The coefficients of the wind turbine gearbox signal after bandpass filtering are Gabor transformed as the time-frequency analysis results of the wind turbine gearbox signal after bandpass filtering.

[0024] Furthermore, the preset filtering interval is determined according to a frequency fluctuation range of the wind turbine gearbox signal.

[0025] Furthermore, bandpass filtering is performed on the wind turbine gearbox signal within a preset filtering interval, comprising the following steps:

[0026] The wind turbine gearbox signal is band-pass filtered within a preset filtering interval using the FFT filtering method.

[0027] Further, performing time-frequency peak search and smoothing processing on the time-frequency analysis result to obtain a discrete time-frequency curve includes the following steps:

[0028] Determine the maximum frequency value in each time node in the time-frequency analysis result, and obtain a time-frequency curve according to the maximum frequency value in each time node;

[0029] The time-frequency curve is smoothed to obtain the discrete time-frequency curve.

[0030] Further, determining a plurality of intervals according to the discrete time-frequency curve comprises the following steps:

[0031] Scanning the discrete time-frequency curve by step scanning;

[0032] A number of intervals in the discrete time-frequency curve are determined according to the result of the step scan.

[0033] Furthermore, the step-scan method is used to determine a plurality of intervals in the discrete time-frequency curve, comprising the following steps:

[0034] Set the width to The window is in Hz;

[0035] Setting a step width, wherein the step width is greater than or equal to the frequency resolution of the time-frequency analysis result;

[0036] Performing frequency domain scanning on the discrete time-frequency curve according to the window and step width;

[0037] The number of points contained in each window during the scanning process is calculated, a window containing a number of continuous points exceeding a preset number threshold is selected, and the plurality of intervals are determined according to a frequency range corresponding to the window.

[0038] Further, if there is no window including a number of consecutive points exceeding a preset number threshold, an order analysis is performed on the wind turbine gearbox signal to obtain a steady-state signal of the wind turbine gearbox signal.

[0039] The present invention also provides a device for extracting steady-state signals of a wind turbine gearbox signal, the device applies the method for extracting steady-state signals of a wind turbine gearbox signal, and the device comprises:

[0040] A discrete curve module, which is used to obtain a discrete time-frequency curve according to a wind turbine gearbox signal;

[0041] A flat interval module, the flat interval module is used to determine a number of intervals according to the discrete time-frequency curve, in each of the several intervals, the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold;

[0042] A signal output module is used to isolate and splice the wind turbine gearbox signals corresponding to the several intervals to obtain a steady-state signal of the wind turbine gearbox signal.

[0043] The present invention also provides a fault self-diagnosis method for a wind turbine gearbox, comprising the following steps:

[0044] The steady-state signal of the wind turbine gearbox signal is obtained by using the steady-state signal extraction method of the wind turbine gearbox signal;

[0045] According to the steady-state signal of the wind turbine gearbox signal, the fault area of ​​the wind turbine gearbox is determined by comparing the gearbox frequency table of the wind turbine gearbox.

[0046] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0047] The present invention performs video analysis on the wind turbine gearbox signal, analyzes the frequency change of the wind turbine gearbox signal waveform, and then extracts the part with stable speed from the waveform, extracting it as a waveform signal with stable speed; at the same time, if there are multiple waveforms with stable speed and consistent frequency, these waveforms are truncated and spliced ​​together. The present invention directly extracts the steady-state component from the original signal, reducing the filtering of the signal information itself. The present invention performs time-frequency analysis on the wind turbine gearbox signal, extracts the part with weak frequency change, and successfully solves the technical problem that when processing long waveform signals, the frequency fluctuation is more obvious due to the influence of speed fluctuation, resulting in spectrum energy leakage, and the part with weak frequency change cannot be extracted. The present invention is suitable for long waveform processing with a collection time of more than 10 seconds, and is suitable for the diagnosis of wind turbine gearboxes and main shaft bearings, and has strong practical applicability. The present invention can locate the accurate frequency information of the signal and improve the accuracy of the diagnosis of the main bearing and gearbox in wind power. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0049] Figure 1 A schematic flow chart of a method for extracting steady-state signals of a wind turbine gearbox signal provided by an embodiment of the present invention;

[0050] Figure 2 A schematic diagram of a specific process of a method for extracting steady-state signals of a wind turbine gearbox signal provided by an embodiment of the present invention;

[0051] Figure 3 Three simulation signal schematic diagrams provided for embodiments of the present invention;

[0052] Figure 4 for Figure 3 Schematic diagram of the waveform spectrum of the simulated time-varying frequency modulation signal LFM composed of three signals;

[0053] Figure 5 For Figure 4 A schematic diagram showing the relationship between the frequency of the simulated time-varying frequency modulation signal LFM and time;

[0054] Figure 6 For Figure 5 Find the schematic diagram of the part of the curve where the frequency changes slowly;

[0055] Figure 7 For Figure 6 Schematic diagram of spectrum analysis for the part of the middle curve where the frequency changes gently;

[0056] Figure 8 For two Figure 4 Schematic diagram of waveform spectrum after splicing the simulated time-varying frequency modulation signal LFM;

[0057] Fig. 9 For Figure 8 Schematic diagram of the relationship between the frequency of the waveform and time;

[0058] Fig.10 For Fig. 9 Find the schematic diagram of the part of the curve where the frequency changes slowly;

[0059] Fig.11 For Fig.10 Schematic diagram of spectrum analysis for the part of the middle curve where the frequency changes gently;

[0060] Fig.12 A schematic diagram of a wind turbine gearbox signal steady-state signal extraction device provided by an embodiment of the present invention;

[0061] Fig.13 A schematic flow chart of a fault self-diagnosis method for a wind turbine gearbox provided by an embodiment of the present invention;

[0062] Fig.14This is a schematic diagram of the installation of wind turbine gearbox signal measurement points;

[0063] Fig.15 It is a schematic diagram of the original data waveform spectrum;

[0064] Fig.16 A schematic diagram of spectrum analysis of raw data;

[0065] Fig.17 This is a schematic diagram of the results of performing FFT filtering on the 600-680Hz frequency band in the original data and using Gabor filtering for time-frequency analysis;

[0066] Fig.18 For Fig.15 Find the schematic diagram of the part of the curve where the frequency changes slowly;

[0067] Fig.19 For Fig.16 Schematic diagram of spectrum analysis for the part of the middle curve where the frequency changes gently;

[0068] Fig. 20 For Fig.17 A schematic diagram showing the comparison of the signal data including the analysis;

[0069] Fig.21 This is a schematic diagram showing that the intermediate stage driven gear of the gearbox has slight tooth breakage. DETAILED DESCRIPTION

[0070] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0071] In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product;

[0072] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0073] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0074] Example 1

[0075] The embodiment of the present invention provides a method for extracting a steady-state signal of a wind turbine gearbox signal. Figure 1 As shown, the following steps are included:

[0076] Obtain discrete time-frequency curves based on wind turbine gearbox signals;

[0077] Determine a plurality of intervals according to the discrete time-frequency curve, in each of the plurality of intervals, the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold;

[0078] The wind turbine gearbox signals corresponding to the plurality of intervals are isolated and spliced ​​to obtain steady-state signals of the wind turbine gearbox signals.

[0079] In this embodiment, a time-frequency analysis is performed on the wind turbine gearbox signal, and then the part of the wind turbine gearbox signal waveform with a stable speed is found through the time-frequency analysis, and then the spectrum analysis of the signal is performed. The signal obtained in this way is a standard stable signal, which solves the technical problem that the existing FFT algorithm extracts the steady-state signal of the wind turbine gearbox signal inaccurately.

[0080] Example 2

[0081] This embodiment further discloses the following contents based on the embodiment 1:

[0082] The step of isolating and splicing the wind turbine gearbox signals corresponding to the plurality of intervals to obtain a steady-state signal of the wind turbine gearbox signal comprises the following steps:

[0083] Using a rectangular window function, the wind turbine gearbox signals corresponding to the plurality of intervals are isolated;

[0084] The wind turbine gearbox signals after multiple sections are spliced ​​to obtain the steady-state signal of the wind turbine gearbox signal.

[0085] In a further embodiment, the step of obtaining a discrete time-frequency curve according to a wind turbine gearbox signal comprises the following steps:

[0086] Obtain wind turbine gearbox signal;

[0087] Performing bandpass filtering on the wind turbine gearbox signal within a preset filtering interval;

[0088] Perform time-frequency analysis on the wind turbine gearbox signal after bandpass filtering to obtain the time-frequency analysis results;

[0089] The time-frequency analysis result is subjected to time-frequency peak search and smoothing processing to obtain a discrete time-frequency curve.

[0090] The specific process is as follows Figure 2 As shown:

[0091] After obtaining the wind turbine gearbox signal, the frequency to be analyzed needs to be selected first, which is generally the prominent frequency component in the original signal. This embodiment analyzes the wind turbine gearbox signal, selects the wind turbine gearbox high-speed stage meshing frequency as the extraction frequency, and performs a narrowband filter to extract this frequency.

[0092] The step of performing time-frequency analysis on the wind turbine gearbox signal after bandpass filtering to obtain a time-frequency analysis result comprises the following steps:

[0093] Perform Gabor transform on the wind turbine gearbox signal after bandpass filtering;

[0094] The coefficient c(m,n) after Gabor transformation of the wind turbine gearbox signal after bandpass filtering is used as the time-frequency analysis result of the wind turbine gearbox signal after bandpass filtering.

[0095] Gabor transform exists as a special case of STFT transform, and the discrete Gabor expansion is defined as:

[0096]

[0097] Where f(l) is the time domain signal, a is the time interval, N is the number of time segments, L = dN; m is the frequency domain sample point, n is the time domain sample point, γ(l-an) is the dual function of the window function. c(m,n) is the Gabor expansion coefficient, the expression is as follows:

[0098]

[0099] Where g(l-an) is the window function, the data length is L, and its dual function γ(l-an) satisfies the following relationship:

[0100]

[0101] Where F is the frequency domain interval and δ is the pulse function. Gabor transform is an STFT transform that uses Gauss function as the window function. The expression of Gauss function and its dual function is:

[0102]

[0103]

[0104] Where T is the sampling time interval.

[0105] In this embodiment, the main frequency curve is obtained by GABOR transformation, and the time location of the stationary signal in the signal is obtained by extracting the interval with slow frequency change from the main frequency curve.

[0106] In a further embodiment, the preset filter interval is determined according to the frequency fluctuation range of the wind turbine gearbox signal. In this embodiment, the fluctuation range of the frequency to be analyzed is observed from the frequency domain of the wind turbine gearbox signal to set the filter interval [fa, fb].

[0107] In a further embodiment, bandpass filtering is performed on the wind turbine gearbox signal within a preset filtering interval, comprising the following steps:

[0108] The wind turbine gearbox signal is band-pass filtered within a preset filtering interval using the FFT filtering method.

[0109] In this embodiment, FFT (Fast Fourier Transform) filtering has the advantages of thorough filtering effect and small frequency truncation residual compared to traditional FIR and IIR filtering, but the amount of calculation is relatively large, and it is suitable for analyzing sampled data.

[0110] The analytical formula of discrete Fourier transform (DFT) is:

[0111]

[0112] In the formula, x(n) is the discrete signal, n is the sampling sequence, and N is the number of sampling points. The discrete Fourier transform is transformed into the fast Fourier transform through computer disk operation, and the operation efficiency is improved. X(k) is the spectrum obtained by the discrete Fourier transform.

[0113] Set the unnecessary frequency range to zero and only keep the spectral components within the required frequency band:

[0114]

[0115] In the formula, f a , f b is the frequency segment, fs is the sampling frequency of the signal, k×fs represents the frequency component represented by the kth point in the frequency domain. D(k) is the spectrum of the FFT filtered signal.

[0116] Then perform inverse FFT transform on D(k), and the analytical formula is:

[0117]

[0118] In a further embodiment, performing time-frequency peak search and smoothing processing on the time-frequency analysis result to obtain a discrete time-frequency curve includes the following steps:

[0119] Determine the maximum frequency value in each time node in the time-frequency analysis result, and obtain a time-frequency curve according to the maximum frequency value in each time node;

[0120] The time-frequency curve is smoothed to obtain the discrete time-frequency curve.

[0121] In a further embodiment, determining a plurality of intervals according to the discrete time-frequency curve comprises the following steps:

[0122] Scanning the discrete time-frequency curve by step scanning;

[0123] A number of intervals in the discrete time-frequency curve are determined according to the result of the step scan.

[0124] In a further embodiment, the step-scanning method is used to determine a plurality of intervals in the discrete time-frequency curve, comprising the following steps:

[0125] Set the width to The window is in Hz;

[0126] Setting a step width, wherein the step width is greater than or equal to the frequency resolution of the time-frequency analysis result;

[0127] Performing frequency domain scanning on the discrete time-frequency curve according to the window and step width;

[0128] The number of points contained in each window during the scanning process is calculated, a window containing a number of continuous points exceeding a preset number threshold is selected, and the plurality of intervals are determined according to a frequency range corresponding to the window.

[0129] In a further embodiment, if there is no window including a number of consecutive points exceeding a preset number threshold, an order analysis is performed on the wind turbine gearbox signal to obtain a steady-state signal of the wind turbine gearbox signal.

[0130] In the specific implementation process, simulation verification is given:

[0131] First, create three simulation signals, such as Figure 3 As shown, Signal1 is the time-varying linear frequency modulation signal X1, and the expression is as follows:

[0132] X1=e 2πftT

[0133] Among them, f is the initial frequency, T is the time axis series, and the expression is:

[0134]

[0135] Among them, fs is the sampling frequency, which is set to 256 Hz; N is the number of sampling points, which is set to 1280.

[0136] Signal2 is the standard cosine signal X2, and its expression is as follows:

[0137] X2=cos(2πft)

[0138] Signal3 is obtained by flipping Signal1.

[0139] Create a simulated time-varying frequency modulation signal LFM. The LFM signal is composed of the above three signals, and the waveform spectrum is shown in Figure 4 .

[0140] Depend on Figure 5It can be seen that since the frequency in LFM changes over time, there is frequency ambiguity in spectrum analysis, which can be alleviated by adding windows.

[0141] Perform Gabor transform on LFM to obtain the relationship between signal frequency and time, see Figure 6 .

[0142] The frequency change curve is obtained by peak finding and fitting methods, and the part of the curve with gentle frequency change is found by frequency window scanning. Figure 7 .

[0143] Depend on Figure 7 It can be seen that the extracted signal is mainly composed of the cosine component of Signal2 in the LFM signal, which is the steady-state component in the LFM signal. It can be seen that the method of this embodiment successfully extracts the steady-state component of the analog signal.

[0144] Since the steady-state components of the non-steady-state signal are distributed in multiple places, the LFM signal is repeated, such as Figure 8 shown.

[0145] Figure 8 The signal in is the concatenation of the two original LFM signals. The signal is transformed by Gabor, and the time-frequency diagram is as follows: Fig. 9 shown.

[0146] The frequency change curve is obtained by peak finding and fitting methods, and the part of the curve with gentle frequency change is found by frequency window scanning. Fig.10 .

[0147] According to the boundary, the signal is truncated and spliced ​​to obtain a steady-state signal, such as Fig.11 As shown, the sampling time of the spliced ​​signal is longer, which has an advantage in signal resolution.

[0148] Example 3

[0149] This embodiment provides a device for extracting steady-state signals from a wind turbine gearbox signal. The device applies the method for extracting steady-state signals from a wind turbine gearbox signal described in Embodiment 1 and Embodiment 2. Fig.12 As shown, the device comprises:

[0150] A discrete curve module, which is used to obtain a discrete time-frequency curve according to a wind turbine gearbox signal;

[0151] A smooth interval module, the smooth interval module is used to determine a number of intervals according to the discrete time-frequency curve, in each of the several intervals, the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold; a signal output module, the signal output module is used to isolate and splice the wind turbine gearbox signals corresponding to the several intervals to obtain a steady-state signal of the wind turbine gearbox signal.

[0152] Example 4

[0153] This embodiment provides a fault self-diagnosis method for a wind turbine gearbox. Fig.13 As shown, the following steps are included:

[0154] The steady-state signal of the wind turbine gearbox signal is obtained by using the steady-state signal extraction method of the wind turbine gearbox signal described in Example 1 and Example 2;

[0155] According to the steady-state signal of the wind turbine gearbox signal, the fault area of ​​the wind turbine gearbox is determined by comparing the gearbox frequency table of the wind turbine gearbox.

[0156] Taking a wind turbine in a certain project as an example, the gearbox model is NGC FD2250GEG, and the transmission information is shown in the following table:

[0157] Table 1

[0158]

[0159] Measuring point installation Fig.14 shown.

[0160] Select the data of the gearbox axial measurement point 5, the sampling frequency is 5120Hz, and the sampling time is 52s. The waveform spectrum of the original data is as follows Fig.15 shown.

[0161] Depend on Fig.15 It can be seen that the frequency ambiguity caused by the speed fluctuation is more obvious for the wind turbine signal with a longer sampling time. When the 0-500Hz spectrum is enlarged, it can be seen that the frequency ambiguity at low frequencies is also obvious, and it is difficult to distinguish the sideband components of each frequency.

[0162] The original high-speed shaft speed of the CMS data is shown as 1744rpm. According to the transmission chain parameters, the high-speed engagement frequency is calculated to be 668.5Hz. The frequency spectrum is analyzed to locate the high-speed shaft rotation frequency. The frequency fluctuation is between 600-680Hz. Fig.16 shown.

[0163] Perform FFT filtering on the 600-680Hz frequency band and use Gabor filtering for time-frequency analysis. The results are as follows: Fig.17 shown.

[0164] The corresponding high-speed shaft meshing frequency curve is extracted through frequency fitting, and the area with two frequency changes that are moderate is located by scanning the frequency window, such as Fig.18 shown.

[0165] The frequency of the frequency change relaxation area is determined to be 672.482Hz. Through calculation, the frequencies of the gearbox are shown in the following table:

[0166] Table 2

[0167]

[0168] It can be seen from the above table that the high-speed shaft speed during this period is 1754.3Hz, which is inconsistent with the data given by the vibration data. This data is more accurate.

[0169] The slowly changing areas are cut off and spliced ​​to obtain the data under stable operation. The waveform spectrum is as follows Fig.19 shown.

[0170] The details of the data spectrum after splicing are clear, which proves that this method can handle the non-stationary data of wind turbines well. Fig. 20 .

[0171] The envelope comparison chart shows that the envelope spectrum of the original signal has a fuzzy fault frequency and a high noise floor. The frequency lines above 150Hz are difficult to distinguish, while the envelope spectrum of the signal processed by this method is clear and the noise floor is clean. By analyzing the envelope spectrum, it is found that there are 144Hz and 6Hz sideband components. By comparing the gearbox frequency table, the intermediate driven wheel gear of the gearbox is faulty.

[0172] After checking the machine and disassembling the gearbox, it was found that the intermediate driven gear of the gearbox had a slight tooth collapse phenomenon, such as Fig.21 shown.

[0173] The same or similar reference numerals correspond to the same or similar components;

[0174] The terms used in the drawings to describe positional relationships are only used for illustrative purposes and should not be construed as limiting this patent;

[0175] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for extracting steady-state signals from a wind turbine gearbox, characterized in that: The following steps are involved: Obtain discrete time-frequency curves based on wind turbine gearbox signals; Determine a plurality of intervals according to the discrete time-frequency curve, in each of the plurality of intervals, the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold; The wind turbine gearbox signals corresponding to the plurality of intervals are separated and spliced ​​to obtain steady-state signals of the wind turbine gearbox signals; The method of obtaining a discrete time-frequency curve according to the wind turbine gearbox signal comprises the following steps: Obtain wind turbine gearbox signal; Performing bandpass filtering on the wind turbine gearbox signal within a preset filtering interval; Perform time-frequency analysis on the wind turbine gearbox signal after bandpass filtering to obtain the time-frequency analysis results; Performing time-frequency peak search and smoothing processing on the time-frequency analysis result to obtain a discrete time-frequency curve; The step of performing time-frequency analysis on the wind turbine gearbox signal after bandpass filtering to obtain a time-frequency analysis result comprises the following steps: Perform Gabor transform on the wind turbine gearbox signal after bandpass filtering; The coefficients of the wind turbine gearbox signal after bandpass filtering are subjected to Gabor transformation as the time-frequency analysis results of the wind turbine gearbox signal after bandpass filtering; Performing time-frequency peak search and smoothing processing on the time-frequency analysis result to obtain a discrete time-frequency curve includes the following steps: Determine the maximum frequency value in each time node in the time-frequency analysis result, and obtain a time-frequency curve according to the maximum frequency value in each time node; The time-frequency curve is smoothed to obtain the discrete time-frequency curve.

2. The method for extracting steady-state signals from a wind turbine gearbox according to claim 1, characterized in that: The step of isolating and splicing the wind turbine gearbox signals corresponding to the plurality of intervals to obtain a steady-state signal of the wind turbine gearbox signal comprises the following steps: Using a rectangular window function, the wind turbine gearbox signals corresponding to the plurality of intervals are isolated; The wind turbine gearbox signals after multiple sections are spliced ​​to obtain the steady-state signal of the wind turbine gearbox signal.

3. The method for extracting steady-state signals from a wind turbine gearbox according to claim 1, characterized in that: Determining a plurality of intervals according to the discrete time-frequency curve comprises the following steps: Scanning the discrete time-frequency curve by step scanning; A number of intervals in the discrete time-frequency curve are determined according to the result of the step scan.

4. The method for extracting steady-state signals of wind turbine gearbox signals according to claim 3 is characterized in that: The method of determining a plurality of intervals in the discrete time-frequency curve by step scanning comprises the following steps: Set the width to The window is in Hz; Setting a step width, wherein the step width is greater than or equal to the frequency resolution of the time-frequency analysis result; Performing frequency domain scanning on the discrete time-frequency curve according to the window and step width; The number of points contained in each window during the scanning process is calculated, a window containing a number of continuous points exceeding a preset number threshold is selected, and the plurality of intervals are determined according to a frequency range corresponding to the window.

5. The method for extracting steady-state signals of wind turbine gearbox signals according to claim 4, characterized in that: If there is no window including a number of consecutive points exceeding a preset number threshold, an order analysis is performed on the wind turbine gearbox signal to obtain a steady-state signal of the wind turbine gearbox signal.

6. A wind turbine gearbox signal steady-state signal extraction device, characterized in that: The device applies the method for extracting steady-state signals of wind turbine gearbox signals according to any one of claims 1 to 5, and the device comprises: A discrete curve module, which is used to obtain a discrete time-frequency curve according to a wind turbine gearbox signal; A flat interval module, the flat interval module is used to determine a number of intervals according to the discrete time-frequency curve, in each of the several intervals, the frequency fluctuation of the discrete time-frequency curve does not exceed a preset fluctuation threshold; A signal output module is used to isolate and splice the wind turbine gearbox signals corresponding to the several intervals to obtain a steady-state signal of the wind turbine gearbox signal.

7. A fault self-diagnosis method for a wind turbine gearbox, characterized in that: The following steps are involved: Obtaining a steady-state signal of a wind turbine gearbox signal by using the method for extracting a steady-state signal of a wind turbine gearbox signal according to any one of claims 1 to 5; According to the steady-state signal of the wind turbine gearbox signal, the fault area of ​​the wind turbine gearbox is determined by comparing the gearbox frequency table of the wind turbine gearbox.