A wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics

Abstract

The present invention belongs to the technical field of wind turbine fault diagnosis and discloses a wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics, comprising the following steps: obtaining the time-frequency ridge of the wind turbine gearbox vibration signal and the impeller instantaneous rotational frequency curve based on a local peak search algorithm; performing angular domain resampling on the extracted impeller instantaneous rotational frequency curve to obtain the impeller instantaneous rotational frequency in the angular domain; applying the least squares principle to perform polynomial fitting on the angular domain instantaneous rotational frequency curve to remove the trend term and obtain the fluctuation term of the impeller instantaneous rotational frequency; performing Fourier transform on the angular domain impeller instantaneous rotational frequency fluctuation term to obtain spectrum analysis, extracting and analyzing the frequency components in the signal, and realizing wind turbine impeller mass imbalance fault diagnosis. The present invention adopts the above-mentioned wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics to achieve accurate diagnosis of wind turbine impeller imbalance faults and improve the operational reliability and safety of wind turbines.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine fault diagnosis, and in particular to a wind turbine impeller imbalance diagnosis method based on rotation frequency fluctuation characteristics. Background Art

[0002] As a vital component of renewable energy, wind turbines' operational stability and reliability directly impact the safety and efficiency of power supply. The impeller, one of a wind turbine's core components, is susceptible to various faults due to long-term exposure to complex natural environments. Impeller imbalance is a common and serious type of failure. Impeller imbalance not only causes vibration problems but can also damage other components, impacting the overall performance and lifespan of the turbine.

[0003] Traditional impeller imbalance fault diagnosis methods primarily rely on analyzing vibration signals. However, due to the complex and changing operating environment of wind turbines, vibration signals are easily affected by noise and other factors, which can affect the accuracy and reliability of fault diagnosis. Furthermore, the rotor frequency of wind turbines fluctuates rapidly, and the fluctuation characteristics of the instantaneous frequency are crucial for fault diagnosis. However, most current methods fail to fully utilize this characteristic for fault diagnosis.

[0004] The instantaneous rotational frequency fluctuation characteristic can reflect the actual state of a wind turbine impeller during operation and can be used to effectively identify and diagnose impeller imbalance faults. However, extracting and analyzing the instantaneous rotational frequency fluctuation characteristic to improve the accuracy and real-time performance of fault diagnosis remains a pressing technical challenge. Summary of the Invention

[0005] The purpose of the present invention is to provide a wind turbine impeller imbalance diagnosis method based on the rotational frequency fluctuation characteristics, so as to achieve accurate diagnosis of wind turbine impeller imbalance faults and improve the operational reliability and safety of the wind turbine.

[0006] To achieve the above object, the present invention provides a wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics, comprising the following steps:

[0007] S1. Obtain the time-frequency ridge line of the wind turbine gearbox vibration signal and the instantaneous frequency curve of the impeller based on the local peak search algorithm;

[0008] S2. resampling the extracted impeller instantaneous rotation frequency curve in the angular domain to obtain the impeller instantaneous rotation frequency in the angular domain;

[0009] S3. Using the least squares method principle, perform polynomial fitting on the instantaneous rotation frequency curve in the angular domain to remove the trend term and obtain the fluctuation term of the impeller instantaneous rotation frequency;

[0010] S4. Perform Fourier transform on the instantaneous frequency fluctuation of the impeller in the angular domain to obtain spectrum analysis, extract and analyze the frequency components in the signal, and realize the fault diagnosis of mass imbalance of the wind turbine impeller.

[0011] Preferably, in step S1, the specific process of calculating the instantaneous rotational frequency of the impeller is as follows:

[0012] S11. Obtain the vibration signal of the wind turbine transmission system based on the vibration acceleration sensor. Obtain the frequency distribution of the signal in different time periods through the sliding window. Obtain the time-frequency spectrum distribution X(τ,f) of the vibration signal based on the STFT algorithm, as shown below:

[0013]

[0014] Where x(t) is the initial signal, h(t) is the window function, X(τ,f) is the frequency component of the signal x(t) at time τ with frequency f, * is the conjugate function, h(t-τ) is the sliding window, and the energy density spectrum of the signal is:

[0015] SP(τ,f)=|X(τ,f)| 2 (2)

[0016] S12. Randomly obtain n initial time points in the analysis period, index the frequency with the largest spectrum energy at the initial time point, and filter the frequency with the largest time-frequency spectrum energy within the analysis frequency band, as shown below:

[0017] f(t,f)=argmaxX(t,f) (3)

[0018] S13. Compare the frequency with the maximum time-frequency spectrum energy within the analysis frequency band at the current moment with the frequency at the previous moment. The judgment conditions are as follows:

[0019] (f(t,f)>αf(t-1,f))||(f(t,f)<βf(t-1,f)) (4)

[0020] Among them, α and β are obtained by expert experience; if formula (4) does not hold, then the frequency value corresponding to the maximum spectrum energy at that moment is 0, and formulas (3) and (4) are repeated until formula (4) is satisfied, thereby obtaining the frequency value at that moment;

[0021] S14, based on the local peak search algorithm, searching forward and backward from the current moment for the frequency value corresponding to the next moment until the entire analysis period is searched, and obtaining the time-frequency ridge line of the analysis period;

[0022] S15, repeat steps S11 to S14 to extract the time-frequency ridge line with higher spectral energy

[0023] S16. Confirm and the instantaneous frequency of the impeller f r The proportional coefficient k between f , calculate the instantaneous frequency of the impeller as follows:

[0024]

[0025] Preferably, in step S2, the extracted impeller instantaneous rotational frequency curve is subjected to angular domain resampling to obtain the impeller instantaneous rotational frequency in the angular domain. The specific process is as follows:

[0026] S21, impeller rotation angular displacement θ(t), is as follows:

[0027] θ(t)=at 2 +bt+c (6)

[0028] Among them, a, b, c are the coefficients to be calculated; t is the sampling time;

[0029] S22. Establish the following equations based on the relationship between the time points t0, t1, and t2 of three consecutive key phase signals and the equal angle increment Δφ:

[0030]

[0031] S23. Substitute formula (7) into formula (6) to solve and obtain a, b, c, as shown below:

[0032]

[0033] S24. Based on the obtained quadratic term coefficient, the relationship between angular displacement and time is obtained, thereby calculating the time corresponding to the equal-angle interval resampling, as shown below:

[0034]

[0035] S25. To avoid repeated sampling, θ satisfies:

[0036]

[0037] Where Δθ is the angular increment of the axis resampling, and N is the number of sampling points.

[0038] Preferably, in step S3, the least squares method is used to perform polynomial fitting on the instantaneous rotation frequency curve in the angular domain to remove the trend term, and obtain the fluctuation term of the instantaneous rotation frequency of the impeller. The specific process is as follows:

[0039] S31. The original data of the trend item to be removed is (x n ,y n ),n=1,2,K,N,x n is the independent variable, y nis the dependent variable, i.e. the output term; construct a polynomial fitting function Y(x) to represent the trend term, as shown below:

[0040]

[0041] Among them, a k is the polynomial coefficient; k is the polynomial order, which is a positive integer and its value is determined by the trend of the original signal to be analyzed; when k = 1, the linear trend term is solved;

[0042] S32, based on the least squares method, by minimizing the sum of squared errors E of all data points, find the optimal {a k}, so that the gap between the fitted curve Y(x) and the actual value y(x) is minimized, as shown below:

[0043]

[0044] S33, determine the trend term needs to solve {a k}, so that E is minimized, and the extreme value conditions are found according to the multivariate function, as shown below:

[0045]

[0046] Where j = 0, 1, 2, K, m, and the matrix expression is:

[0047]

[0048] When m=1, Y(x)=a0+a1x, and the values ​​of a0 and a1 are as follows:

[0049]

[0050] S34. Subtract the trend term represented by Y(x) from the original signal, that is, remove the trend in the original signal y(x) to be analyzed, as shown below:

[0051]

[0052] Preferably, in step S4, wind turbine impeller mass imbalance fault diagnosis is implemented, including the following steps:

[0053] S41. When the impeller has mass imbalance, the wind turbine impeller rotation frequency f r , as shown below:

[0054]

[0055] Among them, T w is the torque of the wind on the impeller, T e is the electromagnetic torque of the generator, n Pis the number of generator pole pairs, J is the moment of inertia, f wind is the impeller rotation frequency caused by random wind speed, f rg The rotation frequency fluctuation is caused by the unbalanced mass of the impeller;

[0056] S42, in f r In, f rg cos(θ) is just a fluctuation term, f wind There are both trend and fluctuation terms in the equation. After detrending by the least squares method, the frequency f r Only the fluctuation term f remains r-wave , as shown below:

[0057] f r-wave =f R-wave -f rg cos(θ) (18)

[0058] Among them, f R-wave is the fluctuation component of the rotation frequency caused by wind speed;

[0059] S43, f r-wave After spectrum analysis, the main frequency characteristics in the spectrum are mass imbalance characteristics, and the characteristic frequencies are:

[0060] 1 / 2πrad -1 (19).

[0061] Therefore, the present invention adopts the above-mentioned wind turbine impeller imbalance diagnosis method based on the rotational frequency fluctuation characteristics, and traces the cause of the wind turbine speed frequency fluctuation by analyzing the instantaneous rotational frequency fluctuation curve of the wind turbine. When the extracted rotational frequency fluctuation law is related to the mass imbalance characteristics of the wind turbine impeller, it can be diagnosed that there is a mass imbalance fault in the wind turbine impeller, so that the wind turbine blades can be repaired in time, and the accurate diagnosis of the wind turbine impeller imbalance fault can be achieved, thereby improving the operating reliability and safety of the wind turbine, which is of great significance to the safe and stable operation of the wind turbine.

[0062] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is an overall flow chart of a wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics of the present invention;

[0064] Figure 2 The present invention converts the frequency conversion from time domain expression to angular domain expression; wherein, (a) is the time domain expression of the frequency conversion; (b) is the angular domain expression of the frequency conversion;

[0065] Figure 3is the detrended result of the instantaneous rotation frequency of the impeller of the present invention; wherein, (a) is the trend term; (b) is the fluctuation term after detrending;

[0066] Figure 4 This is the frequency spectrum of the rotation frequency fluctuation item and the fault feature identification of the present invention. DETAILED DESCRIPTION

[0067] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0068] like Figure 1 As shown, the present invention provides a wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics, comprising the following steps:

[0069] S1. Obtain the time-frequency ridge line of the wind turbine gearbox vibration signal and the instantaneous frequency curve of the impeller based on the local peak search algorithm;

[0070] S2. resampling the extracted impeller instantaneous rotation frequency curve in the angular domain to obtain the impeller instantaneous rotation frequency in the angular domain;

[0071] S3. Using the least squares method principle, perform polynomial fitting on the instantaneous rotation frequency curve in the angular domain to remove the trend term and obtain the fluctuation term of the impeller instantaneous rotation frequency;

[0072] S4. Perform Fourier transform on the instantaneous frequency fluctuation of the impeller in the angular domain to obtain spectrum analysis, extract and analyze the frequency components in the signal, and realize the fault diagnosis of mass imbalance of the wind turbine impeller.

[0073] Example

[0074] S1. Based on the local peak search algorithm, the time-frequency ridge of the wind turbine gearbox vibration signal is obtained and the instantaneous rotation frequency of the impeller is calculated.

[0075] S11. For the signal x(t), the window function h(t) is used to obtain the vibration signal of the wind turbine transmission system based on the vibration acceleration sensor. The frequency distribution of the signal in different time periods is obtained through the sliding window. The time-frequency spectrum distribution X(τ,f) of the vibration signal is obtained based on the STFT algorithm, as shown below:

[0076]

[0077] Where X(τ,f) is the frequency component of the signal x(t) at time τ with frequency f; * is the conjugate function; and h(t-τ) is the sliding window. The energy density spectrum of the signal at this time is:

[0078] SP(τ,f)=|X(τ,f)| 2 (2)

[0079] S12. Randomly obtain n initial time points in the analysis period, index the frequency with the largest spectrum energy at the initial time point, and filter the frequency with the largest time-frequency spectrum energy within the analysis frequency band, as shown below:

[0080] f(t,f)=argmaxX(t,f) (3)

[0081] S13. Compare the frequency with the maximum time-frequency spectrum energy within the analysis frequency band at the current moment with the frequency at the previous moment. The judgment conditions are as follows:

[0082] (f(t,f)>αf(t-1,f))||(f(t,f)<βf(t-1,f)) (4)

[0083] Among them, α and β are obtained by expert experience. If formula (4) does not hold, then the frequency value corresponding to the maximum spectrum energy at that moment is 0. Formulas (3) and (4) are repeated until formula (4) is satisfied, thereby obtaining the frequency value at that moment.

[0084] S14. Based on the local peak search algorithm, search forward and backward from this moment for the frequency value corresponding to the next moment until the entire analysis period is searched, and obtain the time-frequency ridge line of the analysis period.

[0085] S15, repeat steps S11 to S14 to extract the time-frequency ridge line with higher spectral energy

[0086] S16. Confirm and the instantaneous frequency of the impeller f r The proportional coefficient k between f , calculate the instantaneous frequency of the impeller as follows:

[0087]

[0088] Among them, the instantaneous frequency extraction results are as follows: Figure 2 shown.

[0089] S2. Resample the extracted impeller instantaneous rotation frequency curve in the angular domain to obtain the impeller instantaneous rotation frequency in the angular domain.

[0090] S21. The impeller rotational angular displacement θ(t) is a function of time. After Taylor expansion, the relationship between angle and time can be approximated by a quadratic equation as shown below:

[0091] θ(t)=at 2 +bt+c (6)

[0092] Where a, b, and c are the coefficients to be determined; and t is the sampling time.

[0093] S22. Establish the following equations based on the relationship between the time points t0, t1, and t2 of three consecutive key phase signals and the equal angle increment Δφ:

[0094]

[0095] S23. Substitute formula (7) into formula (6) to solve and obtain a, b, c, as shown below:

[0096]

[0097] S24. Based on the obtained quadratic term coefficient, the relationship between angular displacement and time is obtained, thereby calculating the time corresponding to the equal-angle interval resampling, as shown below:

[0098]

[0099] S25. To avoid repeated sampling, θ satisfies:

[0100]

[0101] Where Δθ is the angular increment of the axis resampling, and N is the number of sampling points.

[0102] S3. Use the least squares method to perform polynomial fitting on the instantaneous frequency curve in the angular domain to remove the trend term and obtain the fluctuation term of the impeller instantaneous frequency, such as Figure 3 shown.

[0103] S31. Use the least square method to eliminate the trend term in the frequency conversion and enhance the quality imbalance fault characteristics. The original data of the trend term to be removed is (x n ,y n ),n=1,2,K,N,x n is the independent variable, y n is the dependent variable, i.e. the output term. Construct a polynomial fitting function Y(x) to represent the trend term, as shown below:

[0104]

[0105] Among them, a k are polynomial coefficients; k is the polynomial order, which is a positive integer. Its value is determined by the trend of the original signal to be analyzed. When k = 1, the linear trend term is solved; the value of k shall not exceed m (m≤N).

[0106] S32. The purpose of using the least squares method is to find the best k}, so that the gap between the fitted curve Y(x) and the actual value y(x) is minimized. The least squares method achieves this goal by minimizing the sum of squared errors E of all data points, as shown below:

[0107]

[0108] S33, determine the trend term needs to solve {a k}, so that E is minimized, and the extreme value conditions are found according to the multivariate function, as shown below:

[0109]

[0110] Where j = 0, 1, 2, K, m, and the matrix expression is:

[0111]

[0112] The coefficient matrix of the above equations is a positive definite matrix and will have a unique solution. When m = 1, Y(x) = a0 + a1x. The values ​​of a0 and a1 are as follows:

[0113]

[0114] S34. Finally, the trend term represented by Y(x) is subtracted from the original signal to remove the trend in the original signal y(x) to be analyzed, as shown below:

[0115]

[0116] S4. Perform Fourier transform on the instantaneous frequency fluctuation of the impeller in the angular domain to obtain spectrum analysis, extract and analyze the frequency components in the signal, and realize the fault diagnosis of mass imbalance of the wind turbine impeller.

[0117] S41. When the impeller has mass imbalance, the wind turbine impeller rotation frequency f r , as shown below:

[0118]

[0119] Among them, T w is the torque of the wind on the impeller, T e is the electromagnetic torque of the generator, n P is the number of generator pole pairs, J is the moment of inertia, f wind is the impeller rotation frequency caused by random wind speed, f rg The rotation frequency fluctuation is caused by the unbalanced impeller mass.

[0120] S42, in f r In, f rg cos(θ) is just a fluctuation term, and f wind There are both trend terms and fluctuation terms in the equation. After detrending by the least squares method, the frequency conversion f r Only the fluctuation term f remains r-wave , as shown below:

[0121] f r-wave =f R-wave -f rg cos(θ) (18)

[0122] Among them, f R-wave is the fluctuation component of the rotation frequency caused by wind speed.

[0123] S43, such as Figure 4 As shown, due to f R-wave It has the characteristics of small amplitude and non-periodicity, so f r-wave After spectrum analysis, the main frequency characteristics in the spectrum are mass imbalance characteristics, and the characteristic frequencies are:

[0124] 1 / 2πrad -1 (19).

[0125] Therefore, the present invention adopts the above-mentioned wind turbine impeller imbalance diagnosis method based on the rotational frequency fluctuation characteristics, and traces the cause of the wind turbine speed frequency fluctuation by analyzing the instantaneous rotational frequency fluctuation curve of the wind turbine. When the extracted rotational frequency fluctuation law is related to the mass imbalance characteristics of the wind turbine impeller, it can be diagnosed that there is a mass imbalance fault in the wind turbine impeller, so that the wind turbine blades can be repaired in time, and the accurate diagnosis of the wind turbine impeller imbalance fault can be achieved, thereby improving the operating reliability and safety of the wind turbine, which is of great significance to the safe and stable operation of the wind turbine.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A wind turbine impeller imbalance diagnosis method based on rotational frequency fluctuation characteristics, characterized in that: The following steps are involved: S1. Obtain the time-frequency ridge of the wind turbine gearbox vibration signal and the instantaneous frequency curve of the impeller based on the local peak search algorithm. The specific process of calculating the instantaneous frequency of the impeller is as follows: S11. Obtain the vibration signal of the wind turbine transmission system based on the vibration acceleration sensor. Obtain the frequency distribution of the signal in different time periods through the sliding window. Obtain the time-frequency spectrum distribution X(τ,f) of the vibration signal based on the STFT algorithm, as shown below: Where x(t) is the initial signal, h(t) is the window function, X(τ,f) is the frequency component of the signal x(t) at time τ with frequency f, * is the conjugate function, h(t-τ) is the sliding window, and the energy density spectrum of the signal is: SP(τ,f)=|X(τ,f)| 2 (2); S12. Randomly obtain n initial time points in the analysis period, index the frequency with the largest spectrum energy at the initial time point, and filter the frequency with the largest time-frequency spectrum energy within the analysis frequency band, as shown below: f(t,f)=argmaxX(t,f) (3); S13. Compare the frequency with the maximum time-frequency spectrum energy within the analysis frequency band at the current moment with the frequency at the previous moment. The judgment conditions are as follows: (f(t,f)>αf(t-1,f))||(f(t,f)<βf(t-1,f)) (4); Among them, α and β are obtained by expert experience; if formula (4) does not hold, then the frequency value corresponding to the maximum spectrum energy at that moment is 0, and formulas (3) and (4) are repeated until formula (4) is satisfied, thereby obtaining the frequency value at that moment; S14, based on the local peak search algorithm, searching forward and backward from the current moment for the frequency value corresponding to the next moment until the entire analysis period is searched, and obtaining the time-frequency ridge line of the analysis period; S15, repeat steps S11 to S14 to extract the time-frequency ridge line with higher spectral energy S16. Confirm and the instantaneous frequency of the impeller f r The proportional coefficient k between f , calculate the instantaneous frequency of the impeller as follows: S2. Resample the extracted impeller instantaneous frequency curve in the angular domain to obtain the impeller instantaneous frequency in the angular domain. The specific process is as follows: S21, impeller rotation angular displacement θ(t), is as follows: θ(t)=at 2 +bt+c(6); Among them, a, b, c are the coefficients to be calculated; t is the sampling time; S22. Establish the following equations based on the relationship between the time points t0, t1, and t2 of three consecutive key phase signals and the equal angle increment Δφ: S23. Substitute formula (7) into formula (6) to solve and obtain a, b, c, as shown below: S24. Based on the obtained quadratic term coefficient, the relationship between angular displacement and time is obtained, thereby calculating the time corresponding to the equal-angle interval resampling, as shown below: S25. To avoid repeated sampling, θ satisfies: Where Δθ is the angular increment of the axis resampling, and N is the number of sampling points; S3. Use the least squares method to perform polynomial fitting on the instantaneous frequency curve in the angular domain to remove the trend term and obtain the fluctuation term of the impeller instantaneous frequency. The specific process is as follows: S31. The original data of the trend item to be removed is (x n ,y n ),n=1,2,...,N,x n is the independent variable, y n is the dependent variable, i.e. the output term; construct a polynomial fitting function Y(x) to represent the trend term, as shown below: Among them, a k is the polynomial coefficient; k is the polynomial order, which is a positive integer and its value is determined by the trend of the original signal to be analyzed; when k = 1, the linear trend term is solved; S32, based on the least squares method, by minimizing the sum of squared errors E of all data points, find the optimal {a k }, so that the gap between the fitted curve Y(x) and the actual value y(x) is minimized, as shown below: S33, determine the trend term needs to solve {a k }, so that E is minimized, and the extreme value conditions are found according to the multivariate function, as shown below: Where j = 0, 1, 2, ..., m, and is expressed as a matrix: When m=1, Y(x)=a0+a1x, and the values ​​of a0 and a1 are as follows: S34. Subtract the trend term represented by Y(x) from the original signal, that is, remove the trend in the original signal y(x) to be analyzed, as shown below: S4. Performing Fourier transform on the instantaneous frequency fluctuation of the impeller in the angular domain to obtain spectrum analysis, extracting and analyzing the frequency components in the signal, and realizing the fault diagnosis of mass imbalance of the wind turbine impeller, including the following steps: S41. When the impeller has mass imbalance, the wind turbine impeller rotation frequency f r , as shown below: Among them, T w is the torque of the wind on the impeller, T e is the electromagnetic torque of the generator, n P is the number of generator pole pairs, J is the moment of inertia, f wind is the impeller rotation frequency caused by random wind speed, f rg The rotation frequency fluctuation is caused by the unbalanced mass of the impeller; S42, in f r In, f rg cos(θ) is just a fluctuation term, f wind There are both trend terms and fluctuation terms in the equation. After detrending by the least squares method, the frequency f r Only the fluctuation term f remains r-wave , as shown below: f r-wave =f R-wave -f rg cos(θ) (18); Among them, f R-wave is the fluctuation component of the rotation frequency caused by wind speed; S43, f r-wave After spectrum analysis, the main frequency characteristics in the spectrum are mass imbalance characteristics, and the characteristic frequencies are: 1 / 2πrad -1 (19)。

Images