A wind turbine gearbox output end bearing comprehensive diagnosis method based on multi-model cooperation

Abstract

The application relates to the technical field of wind turbine fault diagnosis, and discloses a wind turbine gearbox output end bearing comprehensive diagnosis method based on multi-model cooperation. The method predicts bearing temperature trends through a dynamic power-shaft temperature model, carries out temperature early warning by using a power-temperature difference model, and cooperates with multidimensional data such as mechanical vibration characteristics (time domain / frequency domain impact energy reflecting bearing mechanical damage), bearing temperature trends and temperature difference analysis (eliminating environmental and load interference to position real fault temperature rise) to carry out confidence calculation. The method can distinguish mechanical faults such as roller peeling and inner ring cracking from operation faults such as heat dissipation failure, breaks through the traditional "single-point alarm" mode, constructs a "root cause analysis-turbine power reduction / triggered shutdown" whole-process closed loop, and significantly improves the safety, reliability and operation efficiency of the turbine operation.

Description

Technical Field

[0001] This invention relates to wind turbine generator fault diagnosis technology, specifically a comprehensive diagnostic method for the output bearing of a wind turbine gearbox based on multi-model collaboration. Background Technology

[0002] The domestic wind power industry is developing rapidly. Many wind turbines, installed in open, high-altitude environments, are subjected to harsh conditions such as wind and sand erosion and extreme weather, leading to frequent gearbox output bearing failures. Existing research mostly employs traditional bearing overheating early warning and diagnostic methods or single-parameter monitoring modes. These methods cannot accurately distinguish the root cause of overheating (such as mechanical or operational failures) and are difficult to tailor maintenance strategies. For example, overheating caused by mechanical wear requires immediate shutdown and replacement of spare parts, while overheating caused by operational failures (such as abnormal lubrication or heat dissipation failure) should be addressed by limiting power output to investigate the cause.

[0003] The core shortcomings of existing technologies are:

[0004] 1. Lack of multi-source data correlation analysis: Traditional diagnostics rely on a single temperature threshold or vibration spectrum, failing to link temperature with operating parameters such as power and speed, resulting in the inability to achieve dynamic adaptive adjustment of the temperature threshold. For example, a fixed temperature threshold cannot follow power changes—when the power of a 2MW unit suddenly drops to 1MW, the bearing temperature prediction value based on the operating condition model should decrease accordingly. However, without dynamic correction based on power changes, the fixed threshold may still misjudge normal cooling as abnormal.

[0005] 2. Insufficient fault type discrimination capability: Vibration analysis focuses only on a single frequency band (such as BPFO / BPF I) without considering temperature trends. In the early stages of bearing failure, the vibration energy in the resonant frequency band increases slightly but the temperature does not change significantly, making it easy for a single vibration model to miss the fault. Conversely, if abnormal temperature is not correlated with power (e.g., vibration energy should be lower under low power), the fault level or type may be misjudged, leading to inappropriate operation and maintenance strategies.

[0006] Therefore, for the scenario of overheating of the output bearing of the gearbox, there is an urgent need to develop a comprehensive diagnostic method that integrates multi-dimensional data and accurately distinguishes between mechanical faults and operational faults causing overheating, in order to solve the core problems of "fuzzy identification of overheating causes and inappropriate operation and maintenance decisions" in the existing technology. Summary of the Invention

[0007] The purpose of this invention is to provide a comprehensive diagnostic method for the output bearing of a wind turbine gearbox based on multi-model collaboration. By using an early warning model to perform temperature warnings based on the acquired temperature data, and by combining the temperature warning model with the mechanical fault diagnosis model, the problem of inaccurate fault cause identification under a single model is solved.

[0008] This invention provides the following technical solution:

[0009] This invention provides a comprehensive diagnostic method for the output bearing of a wind turbine gearbox based on multi-model collaboration, comprising:

[0010] Step S1: Obtain the unit power of the wind turbine, as well as the temperature and vibration signals of the gearbox bearings of the wind turbine;

[0011] Step S2: Generate the temperature residual ΔT and steady-state temperature T using the pre-built temperature early warning model. pred and dynamic temperature difference R ΔTP Steady-state temperature T pred and dynamic temperature difference R ΔTP Used to determine if a temperature rise has occurred;

[0012] Step S3: If a temperature rise occurs, based on the pre-built mechanical fault diagnosis model, the vibration signal is filtered, and the reconstructed signal after filtering is bandpass filtered and Hilbert transformed to extract the envelope spectrum signal. The spectral kurtosis value K is calculated at the theoretical characteristic frequency of the envelope spectrum signal, and the envelope spectrum amplitude ratio R is determined.

[0013] Step S4: Combine the mechanical fault diagnosis model with the temperature early warning model, dynamically adjust the envelope spectrum amplitude ratio R of the envelope spectrum signal according to the temperature residual ΔT; and define the cross-correlation coefficient ρ between the spectral kurtosis value K and the temperature residual ΔT.

[0014] Step S5: Based on the pre-built over-temperature diagnostic model, determine the over-temperature fault confidence level according to the normalized values ​​of the spectral kurtosis value K, envelope spectral amplitude ratio R, temperature residual ΔT, and cross-correlation coefficient ρ; determine the cause of the temperature rise based on the over-temperature fault confidence level.

[0015] Optionally, the temperature residual ΔT and the steady-state temperature T pred and dynamic temperature difference R ΔTP The calculation formulas include:

[0016] ΔT=|T real -T pred |

[0017] T pred =k·P+b·T env +c

[0018]

[0019] Where P is the real-time active power (kW); T env The cabin ambient temperature is given by (°C); k, b, c are coefficients fitted using historical steady-state data, and ΔT is the temperature of the cabin. f-rThis is the temperature difference between the front and rear bearings of the gearbox output shaft.

[0020] Optionally, the steady-state temperature T pred and dynamic temperature difference R ΔTP Methods used to determine whether a temperature rise has occurred include:

[0021] Calculate R in historical data ΔTP The mean μ and standard deviation σ;

[0022] When R ΔTP When the temperature rises beyond μ+2σ and continues for a preset time, a level 1 warning is triggered.

[0023] When R ΔTP When the temperature exceeds μ+3σ and continues for a preset time, a temperature rise is detected and a level two warning is triggered.

[0024] When T real <0.95T pred And R ΔTP The warning will be lifted when the value <μ+1.5σ remains for the preset time.

[0025] Optionally, filtering the vibration signal includes:

[0026] The original vibration signal contains the gear meshing frequency f. mesh A mixed vibration signal with bearing fault signals;

[0027] The LMS (Least Mean Square) algorithm is used to minimize the meshing frequency component in the output signal by iteratively updating the filter weights.

[0028] Optionally, the step of dynamically adjusting the envelope spectrum amplitude ratio R of the envelope spectrum signal according to the temperature residual ΔT includes:

[0029] If ΔT≤5℃, the envelope spectrum amplitude ratio R>14.0;

[0030] If 5℃ < ΔT ≤ 10℃, the envelope spectrum amplitude ratio R > 12.0;

[0031] If ΔT > 10℃, the envelope spectrum amplitude ratio R > 10.0.

[0032] Optionally, the cross-correlation coefficient ρ between the spectral kurtosis value K and the temperature residual ΔT is calculated using the following formula:

[0033]

[0034] Where t represents time and μ represents the mean.

[0035] Optionally, the formula for calculating the over-temperature fault confidence level is:

[0036] Confidence=ω1f K +ω2f R +ω3f ΔT +ω4f ρ (ω1+ω2+ω3+ω4=1)

[0037] Where ω1, ω2, ω3, and ω4 are weighting coefficients, and f K f is the normalized value of the spectral kurtosis; R f is the normalized value of the envelope spectrum amplitude ratio; ΔT f is the normalized value of the temperature residual; ρ This is the normalized value of the cross-correlation coefficient.

[0038] Optionally, when Confidence > 0.7, the cause of the temperature rise is confirmed to be a failure of the high-speed shaft bearing;

[0039] When 0.4 < Confidence ≤ 0.7, the cause of the temperature rise is confirmed to be a suspected fault in the high-speed shaft bearing, and the high-speed shaft bearing is continuously tested.

[0040] When Confidence ≤ 0.4, it is confirmed that the temperature rise is not caused by mechanical failure, and the unit is checked for abnormal heat dissipation or operational failure.

[0041] Compared to existing technologies, this invention has the following significant advantages: Addressing the difficulty of distinguishing between over-temperature during turbine operation and over-temperature due to mechanical faults in existing over-temperature diagnostic methods, this invention constructs a comprehensive diagnostic system for wind turbine gearbox bearings based on multi-model collaboration. By introducing an over-temperature fault confidence algorithm based on vibration signal characteristics and temperature signal data, it quantifies and analyzes the bearing status from a global perspective, accurately identifying over-temperature caused by bearing mechanical faults. This effectively eliminates the influence of external interference factors such as ambient temperature fluctuations and load changes, improving accuracy and reliability, and providing efficient protection for the safe and stable operation of the turbine. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the gearbox condition monitoring method provided in Example 1;

[0043] Figure 2 The shaft temperature curve predicted by the power-shaft temperature model in Example 1;

[0044] Figure 3 The power-temperature ratio curve predicted in Example 1;

[0045] Figure 4 The image shows the outer ring fault envelope spectrum of the gearbox output bearing before filtering, as provided in Example 1.

[0046] Figure 5This is the no-fault envelope spectrum of the gearbox output bearing before filtering, as provided in Example 1.

[0047] Figure 6 The bandpass filtered outer ring fault envelope spectrum of the gearbox output bearing provided in Example 1;

[0048] Figure 7 The fault-free envelope spectrum of the gearbox output bearing after bandpass filtering provided in Example 1;

[0049] Figure 8 The adaptive resonant frequency band filtering of the gearbox output bearing provided in Example 1 is the outer ring fault envelope spectrum of the bearing.

[0050] Figure 9 The fault-free envelope spectrum of the gearbox output bearing after adaptive resonant band filtering provided in Example 1;

[0051] Figure 10 This is a line graph showing the fault-free peak factor, envelope spectrum amplitude ratio, and spectral kurtosis value for the overheating root cause diagnosis in Example 1.

[0052] Figure 11 This is a line graph showing the peak factor of the outer ring fault, the amplitude ratio of the envelope spectrum, and the kurtosis value in the overheating root cause diagnosis of Example 1.

[0053] Figure 12 This is a schematic diagram illustrating the synergistic effect of the power-shaft temperature model and the power-temperature difference ratio model in Example 1;

[0054] Figure 13 This is a schematic diagram of the collaborative diagnosis process in Example 1. Detailed Implementation

[0055] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0056] Example 1

[0057] Combination Figure 1 This invention provides a comprehensive diagnostic method for the output bearing of a wind turbine gearbox based on multi-model collaboration to address this issue. The method includes:

[0058] Step S1: Collect the temperature of the front / rear bearings of the gearbox output shaft using a temperature sensor, and obtain the ambient temperature using an external sensor of the unit.

[0059] The sampling frequency of the vibration sensor in step S1 is 25600Hz, the unit of the rotation speed data is r / min, and the temperature can be directly displayed on the PLC.

[0060] Multi-source data is synchronized with the vibration data clock via a PLC, and real-time power is calculated. Outliers (such as sudden temperature rises >10℃ / min) are removed based on the 3σ principle. Short-term missing data (<5s) are supplemented using linear interpolation, while long-term missing data (≥30s) are removed, triggering sensor alarms.

[0061] Step S2: Employ a dual-model collaborative framework that balances steady-state prediction and dynamic response. The specific architecture is as follows: Figure 12 As shown:

[0062] Define a dynamic power-shaft temperature regression model: T pred =k·P+b·T env +c, and use the sliding window least squares method to solve for the correlation coefficient. Select the most recent 72 hours of data and substitute them into the corresponding objective function: Simultaneously arrange k, b, and c into matrices and solve: in:

[0063] Meanwhile, to ensure more accurate model parameters, k, b, and c are recalculated every 72 hours to adapt to seasonality and unit aging.

[0064] Define the power-temperature difference model: (∈=1kW to prevent division by zero), and simultaneously calculate R in historical data. ΔTP The mean μ and standard deviation σ. When R ΔTP A level 1 warning is triggered when R > μ+2σ and continues for a preset time. ΔTP >μ+3σ indicates a level two warning. When T real <0.95T pred And R ΔTP The warning will be lifted when the value <μ+1.5σ remains for the preset time.

[0065] like Figure 2 and Figure 3 The figures show the shaft temperature curve and power-temperature difference ratio curve predicted by the power-shaft temperature model.

[0066] Step S3: Construct a mechanical fault diagnosis model. Relevant vibration data are measured using vibration sensors; rotational speed data is obtained from the gearbox output shaft speed f provided by the main control PLC. r (Hz) or speed estimation algorithm.

[0067] When the fault characteristic frequencies of high-speed shaft bearings (such as BPFO and BPFI) are completely masked by the gear meshing frequencies and their harmonics, traditional envelope spectrum analysis may fail. In this case, preprocessing of the raw vibration data is necessary.

[0068] This patent employs an adaptive resonant frequency band to track and filter out gear meshing frequency harmonics and related noise in real time, while retaining the fault frequency band of the bearing resonant frequency. The core technology uses the LMS (Least Mean Square) algorithm to iteratively update the filter weights and minimize interference components in the output signal. The input signal contains the gear meshing frequency f. mesh The vibration signal is a mixture of the bearing fault signal and the vibration signal. The purpose of the LMS algorithm in step S3 is to update the filter coefficients and minimize the meshing frequency component in the output signal. The corresponding transfer function is: (λ is the bandwidth control parameter).

[0069] The reference signal is The goal is to suppress the target frequency. (Filter output) Where ω k (n) is the time-varying weight, and the error signal is e(n) = d(n) - y(n).

[0070] Spectral kurtosis of time-domain features Peak factor Bandpass filtering and Hilbert transform are applied to the reconstructed signal z[n]=x bp [n]+j·Η(x bp [n]), where H is the Hilbert transform, and H(x) = IFFT(-j·sign(f)·FFT(x)) is implemented in the frequency domain to obtain the envelope signal. The envelope spectrum of the reconstructed signal is obtained by calculating its envelope spectrum: The spectral kurtosis value is calculated at the theoretical characteristic frequency in the envelope spectrum to determine whether a fault has occurred.

[0071] Finally, the amplitude ratio of the fault characteristic frequency envelope spectrum was obtained (taking the bearing outer ring fault as an example):

[0072] A BPFI =S[f BPFI ].

[0073] like Figures 5-10 The image shows the envelope spectra of the bearings inside the gearbox, both before and after processing.

[0074] Step S4: The mechanical fault diagnosis model and the temperature early warning model achieve deep collaboration through data sharing, dynamic threshold adjustment, and temporal causal analysis, forming a collaborative system of "temperature early warning triggering vibration analysis → dynamic adjustment of vibration characteristic thresholds." The specific architecture is as follows... Figure 13In step S4, the collaborative diagnosis mainly refers to the Pearson correlation coefficient:

[0075] In step S4, the most important part of the multi-factor decision engine is the normalization process. The feature value is divided by its threshold to obtain the normalized value. If the normalized value is greater than 1, it indicates that the feature exceeds the limit. If the normalized value is less than 1, it indicates that the feature does not exceed the limit.

[0076] Steps S2 and S3 have detailed the data synchronization and preprocessing procedures. This step focuses on the temperature warning model-triggered vibration analysis and temperature-vibration time-series causal analysis. First, the dynamic threshold adjustment rules are defined:

[0077]

[0078] Simultaneously define the cross-correlation coefficient: When ρ≥0.7, the temperature rise is determined to be caused by mechanical failure (abnormal vibration precedes temperature rise). When ρ<0.3, the temperature rise is determined to be caused by external heat source (abnormal temperature dominates).

[0079] Step S5: The overheating diagnostic model uses multi-source data such as temperature and vibration, combined with dynamic thresholds and a multi-factor decision engine, to pinpoint the root cause of overheating. The overheating diagnostic model is similar to the temperature warning model in Step S2, with the biggest difference being the use of a multi-factor decision engine to calculate the confidence level.

[0080] Confidence=ω1f K +ω2f R +ω3f ΔT +ω4f ρ

[0081] Where ω1+ω2+ω3+ω4=1, the weight allocation is based on the fault mechanism and historical data.

[0082] To eliminate dimensional differences, the original features need to be standardized, specifically for f. i A common method for normalizing values ​​is threshold relativization: For ω i The weights are allocated using the entropy weighting method, and the specific steps are as follows: Figures 11-12 The image shows a line graph of four indicators in the diagnosis of root causes of overheating.

[0083] Suppose we have m samples (historical data) and n features, construct a data matrix X = [x ij ] m×n Standardization process:

[0084] Treat the standardized values ​​as a probability distribution: Entropy calculation: Next, calculate the coefficient of variation, which reflects the information content of the indicator: d j =1-E j Finally, determine the weights and calculate the normalization parameters:

[0085] When Confidence > 0.7, the cause of the temperature rise is confirmed to be a failure of the high-speed shaft bearing; when 0.4 < Confidence ≤ 0.7, the cause of the temperature rise is confirmed to be a suspected failure of the high-speed shaft bearing, and monitoring will continue; when Confidence ≤ 0.4, the cause of the temperature rise is confirmed to be not a mechanical failure, but may be due to abnormal heat dissipation or operational failure.

[0086] The principle of this model is based on the idea of ​​weighted comprehensive scoring, which combines the weights and normalized values ​​of each input feature to finally output a confidence score between 0 and 1.

[0087] Step S4 mentions the cross-correlation coefficient ρ, whose core function is to resolve the "first and second" problem, determining whether the temperature anomaly is caused by a mechanical failure. Step S5 mentions the confidence score, whose core function is to coordinate multi-dimensional evidence and comprehensively determine the "fault type and severity." The cross-correlation coefficient focuses on causal verification, while the confidence score focuses on the synthesis of multi-dimensional evidence and accurate fault diagnosis. The two complement each other, jointly improving the accuracy and reliability of the diagnosis.

[0088] This embodiment is based on a multi-model collaborative comprehensive diagnosis of the output bearing of the wind turbine gearbox. It predicts temperature trends using a dynamic power-shaft temperature model, provides temperature early warning using a power-temperature difference model, and calculates confidence levels by combining multi-dimensional data such as mechanical vibration characteristics, bearing temperature trends, and temperature difference analysis. It supports mechanical fault types such as roller spalling, inner ring cracks, and cage breakage, as well as operational faults such as lubrication abnormalities and heat dissipation failures. Furthermore, it breaks through the traditional "single-point alarm" mode, constructing a closed-loop process of "root cause analysis - unit power reduction / triggered shutdown," significantly improving the safety, reliability, and maintenance efficiency of unit operation.

[0089] Those skilled in the art will understand that the modules in the apparatus of the implementation scenario can be distributed within the apparatus of the implementation scenario as described, or they can be located in one or more apparatuses different from this implementation scenario, with corresponding changes. The modules of the above-described implementation scenario can be combined into one module, or they can be further divided into multiple sub-modules.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

[0091] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A comprehensive diagnostic method for the output bearing of a wind turbine gearbox based on multi-model collaboration, characterized in that, include: Step S1: Obtain the unit power of the wind turbine, as well as the temperature and vibration signals of the gearbox bearings of the wind turbine; Step S2: Generate temperature residuals using a pre-built temperature early warning model. steady-state temperature and dynamic temperature difference The steady-state temperature and dynamic temperature difference Used to determine if a temperature rise has occurred; Step S3: If a temperature rise occurs, based on the pre-built mechanical fault diagnosis model, the vibration signal is filtered, and the reconstructed signal after filtering is bandpass filtered and Hilbert transformed to extract the envelope spectrum signal. The spectral kurtosis value K is calculated at the theoretical characteristic frequency of the envelope spectrum signal, and the envelope spectrum amplitude ratio R is determined. Step S4: Combine the mechanical fault diagnosis model with the temperature early warning model, based on the temperature residual... The envelope spectrum amplitude ratio R of the envelope spectrum signal is dynamically adjusted; and the spectral kurtosis value K and temperature residual are defined. Cross-correlation coefficient between ; Step S5: Based on the pre-built overtemperature diagnostic model, according to the spectral kurtosis value K, envelope spectrum amplitude ratio R, and temperature residual... and cross-correlation coefficient The normalized value is used to determine the over-temperature fault confidence level; the cause of the temperature rise is determined based on the over-temperature fault confidence level. The temperature residual steady-state temperature and dynamic temperature difference Calculation formula include: Where P is the real-time active power. ; cabin ambient temperature ; The coefficients are fitted using historical steady-state data. This is the temperature difference between the front and rear bearings of the gearbox output shaft. The formula for calculating the confidence level of the over-temperature fault is: in, These are the weighting coefficients. This represents the normalized value of the spectral kurtosis. This is the normalized value of the envelope spectrum amplitude ratio; This is the normalized value of the temperature residual; This is the normalized value of the cross-correlation coefficient.

2. The comprehensive diagnostic method for the output bearing of a wind turbine gearbox according to claim 1, characterized in that, The steady-state temperature and dynamic temperature difference Methods used to determine whether a temperature rise has occurred include: Calculating historical data mean and standard deviation ; when And after a preset time, it will determine if the temperature rises and trigger a level 1 warning; when And after a preset time, it will determine if the temperature rises and trigger a level 2 warning; when and The warning will be lifted after the preset time has elapsed.

3. The comprehensive diagnostic method for the output bearing of a wind turbine gearbox according to claim 1, characterized in that, The temperature residual Dynamically adjusting the envelope spectrum amplitude ratio R of the envelope spectrum signal includes: like Envelope spectrum amplitude ratio ; like Envelope spectrum amplitude ratio ; like Envelope spectrum amplitude ratio .

4. The comprehensive diagnostic method for the output bearing of a wind turbine gearbox according to claim 1, characterized in that, The spectral kurtosis value K in step S4 and the temperature residual in step S2 Cross-correlation coefficient between The calculation formula is: in, Indicates time, This represents the mean.

5. The comprehensive diagnostic method for the output bearing of a wind turbine gearbox according to claim 1, characterized in that, The causes of the temperature rise, determined based on the over-temperature fault confidence level, include: when Upon investigation, it was confirmed that the temperature rise was caused by a failure in the high-speed shaft bearing. when At that time, it was confirmed that the cause of the temperature rise was a suspected failure of the high-speed shaft bearing, and the high-speed shaft bearing was continuously tested; when If the temperature rise is not caused by mechanical failure, check whether the unit has abnormal heat dissipation or malfunction.

Images