A Wavelet Multi-Scale Defect Detection Method for Wind Turbine Blades Based on Rotation Speed ​​Synchronization Time Difference Correction

Abstract

This invention relates to the field of wind turbine blade defect detection, and discloses a wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction. The method includes: installing a microphone at the root of the wind turbine blade; installing a speed sensor on the wind turbine shaft; collecting and digitizing the acoustic signal of the wind turbine blade rotation; using the speed sensor to eliminate the time difference of the acoustic signal between different blades to obtain a correction signal; extracting features from each correction signal to obtain a feature signal; obtaining the correlation coefficient between the feature signals; detecting wind turbine blade defects based on the correlation coefficient and triggering an alarm. This invention overcomes the limitations of traditional monitoring methods and provides a high-precision, real-time, and economical solution, applicable to various complex environments and different types of wind turbines, and has broad application prospects.

Description

Technical Field

[0001] This invention relates to the field of wind turbine blade defect detection, and in particular to a wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction. Background Technology

[0002] In recent years, with the continuous growth of global demand for renewable energy, wind power has developed rapidly as a clean and sustainable energy form. However, wind turbines are usually installed in remote locations with harsh environmental conditions, such as mountains, wilderness, beaches, and islands. These environments not only increase the difficulty of equipment maintenance but also place higher demands on the reliability and durability of wind turbines.

[0003] In wind turbine generators, blades are one of the most critical components, their performance directly impacting power generation efficiency and system stability. Blades are typically designed and manufactured using composite materials. While these materials are lightweight and high-strength, they are susceptible to various environmental factors during long-term operation, such as wind, sand, salt spray, ultraviolet radiation, and lightning strikes, leading to surface damage and internal structural fatigue. These problems not only reduce blade efficiency but can also pose serious safety risks, such as blade breakage. Statistics show that blade failures account for up to 34% of total wind turbine generator downtime, severely impacting power generation efficiency and significantly increasing maintenance costs. Therefore, real-time and accurate health monitoring of wind turbine blades is crucial.

[0004] Traditional blade monitoring methods mainly include technologies such as acoustic emission, vibration analysis, fiber Bragg gratings, and infrared thermal imaging. While these methods each have their advantages, they also have limitations, such as complex data acquisition, difficult analysis, and high costs. Especially in practical applications, the diversity of wind farm location, turbine type, fault type, and background environment makes traditional rule- and feature-based monitoring methods difficult to meet the requirements of high precision and real-time performance. Furthermore, although deep learning-based methods have good recognition capabilities, their application in practical engineering still faces challenges due to limitations in data acquisition scale and insufficient model generalization ability. Summary of the Invention

[0005] The purpose of this invention is to propose a wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction, thereby solving the technical problems of high difficulty in existing wind turbine blade defect monitoring and analysis, and difficulty in meeting high precision and real-time requirements.

[0006] Specifically, this invention provides a wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction, comprising the following steps:

[0007] S1. Install a microphone at the root of the fan blades; install a speed sensor on the fan shaft;

[0008] S2. Collect the acoustic signal of the fan blade rotation and digitize it;

[0009] S3. Use a speed sensor to eliminate the time difference of acoustic signals from different blades to obtain a correction signal;

[0010] S4. Perform feature extraction on each correction signal to obtain the feature signal;

[0011] S5. Obtain the correlation coefficient between the feature signals;

[0012] S6. Detect wind turbine blade defects based on the correlation coefficient and trigger an alarm.

[0013] A storage device that stores instructions and data for implementing a wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction.

[0014] A wavelet multi-scale wind turbine blade defect detection device based on speed synchronization time difference correction includes: a processor and a storage device; the processor loads and executes instructions and data in the storage device to implement a wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction.

[0015] The beneficial effects provided by this invention are: it can significantly improve the health monitoring level of wind turbine blades, reduce operation and maintenance costs, and ensure the safe and stable operation of wind turbines. This technology not only overcomes the limitations of traditional monitoring methods but also provides a high-precision, real-time, and economical solution, applicable to various complex environments and different types of wind turbines, and has broad application prospects. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction of the present invention;

[0017] Figure 2 This is a schematic diagram of the hardware device used in this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0019] Before formally describing the present invention, a general description of the solution of the present invention will be given first to facilitate understanding.

[0020] Please refer to Figure 1The present invention provides a wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction, comprising the following steps:

[0021] S1. Install a microphone at the root of the fan blades; install a speed sensor on the fan shaft;

[0022] It should be noted that this invention installs a microphone near the root of each blade of the wind turbine to capture noise signals generated during blade rotation. A speed sensor is installed on the turbine shaft to monitor the turbine's rotational speed in real time.

[0023] S2. Collect the acoustic signal of the fan blade rotation and digitize it;

[0024] Specifically, the signal collected by the microphone in time The instantaneous value at that point . These are the blade numbers. These signals are sampled into discrete-time series. ,in This is the sampling point index. The sampling frequency is set to... The sampling interval is .

[0025] The digitized signal is represented as: .

[0026] S3. Use a speed sensor to eliminate the time difference of acoustic signals from different blades to obtain a correction signal;

[0027] Step S3 is as follows:

[0028] S31. Obtain the real-time speed of the fan based on the speed sensor and calculate the fan rotation cycle;

[0029] S32. Determine the blade time delay based on the fan rotation cycle;

[0030] Delay in step S32 The calculation formula is as follows:

[0031]

[0032] in, i This refers to the numbering of the wind turbine blades. t i For the first i The time points collected by the microphones on each wind turbine blade. t 1 represents the data collection time point for the first wind turbine blade, and also serves as a reference time point. r ( t 1) Real-time speed information of the fan

[0033] S33. Correct the generation signals of different leaves based on the calculated time delay to obtain the corrected signals.

[0034] The corrected signal in step S33 is as follows:

[0035]

[0036] in The corrected signal; n Sampling point index

[0037] As one embodiment, the speed information is obtained as follows: the speed sensor provides the real-time speed information of the fan. Or the number of rotations per unit time. The fan's rotational period is calculated based on the rotational speed information. , .

[0038] For the case of three blades, the theoretical time interval between each blade and the pickup is: .

[0039] Time difference calculation: Assume there are three blades passing through the microphone at the following times: ,in Use the actual time point as a reference. Calculate the difference between the actual time point and the theoretical time point to determine the time delay. :

[0040] The delay of the second blade :

[0041]

[0042] The delay of the third blade :

[0043]

[0044] Time delay correction: The signal is corrected based on the calculated time delay so that all signals can be compared at the same rotation angle. The corrected signal is represented as follows:

[0045]

[0046]

[0047] S4. Perform feature extraction on each correction signal to obtain the feature signal;

[0048] It should be noted that step S4 is as follows:

[0049] S41. Perform discrete wavelet transform on the corrected signal to obtain signals at different scales. l The following signal;

[0050] S42, For different scales l The signal is subjected to feature extraction to obtain a feature signal, which includes wavelet coefficients, energy distribution and peak position.

[0051] As one embodiment, for each corrected noise signal Perform Discrete Wavelet Transform (DWT) to obtain multi-level wavelet coefficients. This refers to the decomposition level. Using appropriate wavelet basis functions, such as the Haar wavelet or Daubechies wavelet, multi-scale analysis is performed. The discrete wavelet transform can decompose a signal into a series of approximate and detail coefficients, each corresponding to a different scale. Signal characteristics under [the following conditions].

[0052]

[0053] in , It is the largest wavelet scale.

[0054] Feature extraction:

[0055] Extract signal features at different scales, such as energy distribution and peak location.

[0056] For discrete wavelet transform, we can start from the wavelet coefficients at each stage. Features can be extracted from this. The energy distribution at each scale can then be calculated.

[0057]

[0058] It can also calculate the peak location at each scale:

[0059]

[0060] These features can define the local properties of a signal at different scales.

[0061] in: It is mainly used to represent the position index in the wavelet coefficient sequence, and is used to calculate the correlation between wavelet coefficients, energy distribution and peak position.

[0062] S5. Obtain the correlation coefficient between the feature signals;

[0063] It should be noted that step S5 uses a cross-correlation function to calculate the correlation coefficient and further standardizes the correlation coefficient.

[0064] As one example, for each scale The correlation between the wavelet coefficients of the three blade noise signals was calculated.

[0065] Correlation can be estimated using the cross-correlation function:

[0066]

[0067] in yes .

[0068] in, It is also the leaf number. This allows us to express the number of... The first leaf and the first The correlation coefficient of each leaf.

[0069] To eliminate the impact of amplitude differences, standardized correlation can be calculated:

[0070]

[0071] in and They are and The autocorrelation function.

[0072] For each scale The correlation between the energy distributions of the three blade noise signals was calculated.

[0073]

[0074] To eliminate the impact of amplitude differences, standardized correlation can be calculated:

[0075]

[0076] in It is the autocorrelation function of energy distribution.

[0077] For each scale The correlation between the peak positions of the three blade noise signals was calculated.

[0078]

[0079] To eliminate the impact of amplitude differences, standardized correlation can be calculated:

[0080]

[0081] in It is the autocorrelation function of the peak position.

[0082] S6. Detect wind turbine blade defects based on the correlation coefficient and trigger an alarm.

[0083] Step S6 specifically involves setting a threshold based on the wavelet coefficient characteristics of the normal blade signal, for each scale. Set the corresponding threshold If a certain leaf signal is at the scale l If the correlation coefficient on the blade is lower than the set threshold, it is considered that the blade may have a defect and an alarm is triggered.

[0084] As one embodiment, a threshold is set: the threshold is set based on the wavelet coefficient characteristics of the normal blade signal. For each scale Set the corresponding threshold .

[0085] Anomaly detection: If the characteristics of a certain blade signal at a certain scale exceed the set threshold, the blade is considered to have a possible defect.

[0086] For example, if the correlation of wavelet coefficients, energy distribution, or peak position of a certain blade is significantly lower than that of other blades at a certain scale, it may indicate that there is a problem with that blade.

[0087] Set alarm threshold An alarm is triggered when the correlation falls below this threshold:

[0088] For any wavelet scale , , ,

[0089] Then the first One or The wind turbine blades are defective.

[0090] Please see Figure 2 , Figure 2 This is a schematic diagram of the hardware device in an embodiment of the present invention. The hardware device specifically includes: a wavelet multi-scale wind turbine blade defect detection device 401 based on rotational speed synchronization time difference correction, a processor 402, and a storage device 403.

[0091] A wavelet multi-scale wind turbine blade defect detection device 401 based on speed synchronization time difference correction: The wavelet multi-scale wind turbine blade defect detection device 401 based on speed synchronization time difference correction implements the wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction.

[0092] Processor 402: The processor 402 loads and executes the instructions and data in the storage device 403 to implement the wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction.

[0093] Storage device 403: The storage device 403 stores instructions and data; the storage device 403 is used to implement the wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction.

[0094] The key point of this invention is:

[0095] 1. High-precision detection: By correcting signal time difference using real-time rotational speed information, the consistency of compared signals is ensured at the same rotational angle, thereby improving the accuracy and reliability of detection. This method overcomes the signal time difference problem caused by rotational speed fluctuations in traditional methods, improving the accuracy and reliability of signal comparison. Furthermore, wavelet transform is used to extract signal features at different scales, which can more comprehensively reflect the state of the blade within different frequency ranges and identify defects of different types and sizes. This method can capture subtle changes in the blade at different frequencies, improving detection sensitivity.

[0096] 2. Real-time Monitoring: Utilizing real-time speed information provided by a speed sensor, signals can be corrected in real time under dynamic conditions, enabling instant monitoring of the blade's condition. This method can promptly capture any abnormal changes in the blade during operation, ensuring the real-time nature of the detection results. Simultaneously, through multi-scale feature extraction and correlation calculation, potential blade problems can be identified promptly, and a rapid response can be initiated. This method can quickly identify and locate defects, improving the system's response speed and detection efficiency.

[0097] 3. Economy and Ease of Use: Compared to high-cost technologies such as fiber Bragg gratings, the hardware required by this invention is relatively simple, easy to deploy and maintain. This method reduces the initial investment and maintenance costs of the system, improving economic efficiency. Furthermore, through standardized correlation calculations, the data analysis process is simplified, reducing the requirements for professional personnel and enhancing the system's ease of use. This method makes data analysis simpler and faster, lowering the technical threshold.

[0098] 4. Adaptability to Complex Environments: It exhibits environmental robustness, maintaining stable detection results even in complex and variable natural environments through time-of-flight correction and multi-scale analysis. This method can adapt to various harsh environmental conditions, ensuring the stability and reliability of the detection results. It also boasts diverse applicability: suitable for different types of wind turbines and various wind farm environments, demonstrating broad applicability. This method can be widely applied to various types of wind turbine generator sets, improving the technology's versatility and practicality.

[0099] In summary, the beneficial effects of this invention are: it can significantly improve the health monitoring level of wind turbine blades, reduce operation and maintenance costs, and ensure the safe and stable operation of wind turbines. This technology not only overcomes the limitations of traditional monitoring methods but also provides a high-precision, real-time, and economical solution, applicable to various complex environments and different types of wind turbines, and has broad application prospects.

[0100] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A wavelet multi-scale wind turbine blade defect detection method based on rotational speed synchronization time difference correction, characterized in that: The method includes the following steps: S1. Install a microphone at the root of the fan blades; install a speed sensor on the fan shaft; S2. Collect the acoustic signal of the fan blade rotation and digitize it; S3. Use a speed sensor to eliminate the time difference of acoustic signals from different blades to obtain a correction signal; S4. Perform feature extraction on each correction signal to obtain the feature signal; S5. Obtain the correlation coefficient between the feature signals; S6. Detect wind turbine blade defects based on correlation coefficients and trigger an alarm; Step S3 is as follows: S31. Obtain the real-time speed of the fan based on the speed sensor and calculate the fan rotation cycle; S32. Determine the blade time delay based on the fan rotation cycle; S33. Correct the generation signals of different leaves based on the calculated time delay to obtain the corrected signals; Delay in step S32 The calculation formula is as follows: in, i This refers to the numbering of the wind turbine blades. t i For the first i The time points collected by the microphones on each wind turbine blade. t 1 represents the data collection time point for the first wind turbine blade, and also serves as a reference time point. r ( t 1) Real-time speed information of the fan.

2. The wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction as described in claim 1, characterized in that: The corrected signal in step S33 is as follows: in The corrected signal; n This is the index for the sampling points.

3. The wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction as described in claim 1, characterized in that: Step S4 is as follows: S41. Perform discrete wavelet transform on the corrected signal to obtain signals at different scales. l The following signal; S42, For different scales l The signal is subjected to feature extraction to obtain a feature signal, which includes wavelet coefficients, energy distribution and peak position.

4. The wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction as described in claim 1, characterized in that: Step S5 uses a cross-correlation function to calculate the correlation coefficient and further standardizes the correlation coefficient.

5. The wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction as described in claim 3, characterized in that: Step S6 specifically involves setting a threshold based on the wavelet coefficient characteristics of the normal blade signal, for each scale. Set the corresponding threshold If the correlation coefficient of a certain blade signal on scale l is lower than the set threshold, it is considered that the blade may have a defect and an alarm is triggered.

6. A storage device, characterized in that: The storage device stores instructions and data to implement the wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction as described in any one of claims 1 to 5.

7. A wavelet multi-scale wind turbine blade defect detection device based on rotational speed synchronization time difference correction, characterized in that: include: A processor and a storage device; the processor loads and executes instructions and data in the storage device to implement the wavelet multi-scale wind turbine blade defect detection method based on speed synchronization time difference correction as described in any one of claims 1 to 5.

Images