Wind turbine blade fault detection method based on aeroacoustics and deep learning

Abstract

The invention discloses a wind turbine blade fault detection method based on aeroacoustics and deep learning, and relates to the technical field of wind turbine blade damage detection, and the detection method comprises the following steps: obtaining aeroacoustics data of a to-be-detected wind turbine blade operating in a healthy state and a damaged state; carrying out zero mean value and filtering processing on the aeroacoustic data based on an aeroacoustic theory; based on the aeroacoustic data subjected to zero mean and filtering processing, calculating power spectral density and extracting power spectral density under a characteristic frequency index, and establishing a BP neural network detection model based on a power spectral density characteristic; inputting to-be-detected data subjected to feature extraction into the BP neural network detection model based on the power spectral density feature for calculation, and establishing an evaluation system based on an output result to judge the state of the blade. According to the method, overfitting of the detection model is avoided, and the blade damage identification accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind turbine blade damage detection, and more particularly to a wind turbine blade fault detection method based on aeroacoustics and deep learning. BACKGROUND

[0002] As a key component of wind turbine operation, the blade often operates in complex natural environments such as strong wind, lightning, icing, and rain erosion, and is prone to damage. Due to the lack of effective real-time blade damage detection means, the traditional manual inspection method cannot timely detect blade damage, and such damage, if not timely detected and repaired, will lead to reduced power generation efficiency, shutdown, and even accidents such as damage of the wind turbine, greatly increasing the operation and maintenance cost of the wind turbine. Therefore, wind turbine blade damage detection is of great significance to ensure safe and efficient operation of the wind turbine and reduce the power generation cost of the wind turbine during its life cycle.

[0003] At present, the existing wind turbine damaged blade diagnosis methods mainly include: (1) Strain monitoring: when using a resistance strain gauge to monitor wind turbine blade damage, the wiring is complex, the installation points are numerous, and it is difficult to monitor the damage in the internal structure of the wind turbine blade. When using an optical fiber strain gauge, the optical fiber sensor needs to be pre-embedded in the wind turbine blade, which increases the difficulty of manufacturing the wind turbine blade, and in addition, the special structure of the wind turbine blade also affects the detection accuracy of the optical fiber sensor; (2) Vibration monitoring: the size of the wind turbine blade is large, and a large number of sensors are needed to realize global detection, which will increase the monitoring cost, and vibration monitoring is not sensitive to early minor defects of the wind turbine blade; (3) Machine vision detection: the damage information of the wind turbine blade is easily disturbed by background information, and due to the lack of effective means to extract the blade damage features, the detection model input too many feature parameters, which leads to overfitting of the detection model, reduces the detection accuracy, and increases the detection time.

[0004] Therefore, it is necessary to develop a real-time and efficient wind turbine damaged blade diagnosis method to monitor the blade state in real time and reduce the operation and maintenance cost. SUMMARY

[0005] To solve the above problems, the present application provides a wind turbine blade fault detection method based on aeroacoustics and deep learning.

[0006] To achieve the above purpose, the technical scheme adopted by the present application is: A wind turbine blade fault detection method based on aeroacoustics and deep learning, comprising the following steps: Step 1: Obtain the aerodynamic acoustic data of the wind turbine blade under healthy state and damaged state; the aerodynamic acoustic data contains real-time sound pressure data X(j) of the wind turbine from start to rated operating condition; Step 2: Based on the aerodynamic acoustic theory, the aerodynamic acoustic data is zero-mean and filtered; Step 3: Based on the aerodynamic acoustic data preprocessed in step 2, the power spectral density is calculated and the power spectral density under the characteristic frequency index is extracted, and the BP neural network detection model based on the power spectral density characteristics is established; Step 4: The data to be tested is input into the BP neural network detection model based on the power spectral density characteristics after feature extraction, and the output result is used to establish an evaluation system to judge the state of the blade.

[0007] Further, in step 1, the sampling time of the aerodynamic acoustic data t Need to meet: ; In the formula, F s is the sampling frequency of the acoustic sensor, is the frequency resolution requirement.

[0008] Further, the step 2 specifically includes the following steps: Step 2-1: Zero-mean processing: ; In the formula, represents the i-th sound pressure value in any group of sound pressure data of the wind turbine to be tested, represents the i-th sound pressure value after mean processing; N is the number of current processing data; Step 2-2: According to the aerodynamic noise frequency band [a, b] radiated by the wind turbine blade, the band-pass filter is performed, ; ; ; In the formula, represents the filtered data spectrum; represents the transfer function of the band-pass filter; X(f) represents the spectrum of the time-domain sound pressure signal to be processed; represents the filtered time-domain data; represents the inverse Fourier transform.

[0009] Further, the step 3 specifically includes the following steps: Step 3-1: The aerodynamic acoustic data after zero-mean and filtering is divided into test set and verification set; Step 3-2: Divide the sound pressure data X(j) in the test set into K segments, each with a length of L data points, and calculate the Fourier transform result of the segment data at the frequency index K n : K j j W(j) K f n f n Step 3-3: Calculate the power spectral density difference between the healthy blade and the damaged blade: Step 3-4: Take the average of the power spectral density difference as the threshold value for peak value retrieval, and retain the peak data greater than the average value to obtain the power spectral density characteristic peak value of each working condition and the frequency index ; Step 3-5: Count the number of frequency indexes appearing in each working condition and the average of the number of times, and retain the frequency indexes greater than the average of the number of times and n 1 corresponding power spectral density characteristic values ; Step 3-6: Normalize the n 1 power spectral density characteristic values to n ​​​​​​​​​​​​​​​​​​​​​​​​1 The PSD feature is used as input, the output is set to 2, the activation function is selected as the sigmoid function, the training method is selected as the adaptive momentum gradient descent method, the hidden layer neurons are selected as the average of the sum of the input and output neurons, and the target error is set to ϵ. Steps 3-7: Use the neural network output [1,1] to represent healthy blades and the neural network output [0,1] to represent faulty blades. Establish and train the BP neural network detection model, adjust the number of iterations, and stop the model iteration when the model error reaches the set error ϵ. Derive the training model function. Steps 3-8: Validate the BP neural network detection model using data from the test set. If the output of the detection model is within the allowable error range, the model is proven to be effective, and the detection function is derived. Otherwise, return to adjust the training model parameters.

[0010] Furthermore, step 4 specifically includes the following steps: Step 4-1: Calculate the power spectral density of the time-domain data y(i) after preprocessing and filtering in Step 2 according to Step 3-2, and use the power spectral density index obtained in Step 3-5. Extract the power spectral density feature value, normalize it, and import it into the detection function obtained in step 3-8 to obtain a two-digit output result [U,V]. Step 4-2: By setting the threshold α, evaluate the output result [U,V] and judge the state of the blade.

[0011] Furthermore, in the output result [U,V], the second value close to 1 indicates that the data conforms to the aerodynamic noise characteristics, and the closer the first value is to 0, the higher the probability of blade failure. The first output is taken as 1-U, and two digits are retained before it is output as a percentage, which is the probability of failure. Finally, a reasonable threshold is set to determine whether to issue a warning message.

[0012] The above solution addresses the critical issue of accurate detection of wind turbine blade faults. Based on wind turbine aeroacoustics and deep learning theory, it uses filtering technology based on aerodynamic noise to reduce data dimensionality and accelerate model computation time. By using power spectral density calculation and peak extraction techniques to obtain feature values ​​of damaged and healthy blades, it trains a BP neural network model, avoiding overfitting of the detection model and improving the accuracy of identification. This provides a new technical approach for real-time monitoring of wind turbine blades. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating the overall process of the wind turbine blade fault detection method based on aeroacoustics and deep learning in this embodiment of the invention. Figure 2The flow chart of the wind turbine blade aerodynamic sound power spectral density feature extraction in the embodiment of the present application is shown in the figure. Figure 3 The flow chart of the BP neural network detection model construction based on the power spectral density feature in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0014] The present application will be described in detail below with specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any form. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made. These all belong to the protection scope of the present application.

[0015] In this embodiment, a 450W wind turbine with SD2030 airfoil is taken as a typical analysis object, and its sound pressure data under the conditions of blade damage and health are collected to establish a blade fault diagnosis model, thereby specifically describing a wind turbine blade fault detection method based on aerodynamic acoustic signal features. As shown in the figure, it includes the following steps: Figure 1 Step 1: Artificially set the common damage forms of the wind turbine blade, arrange acoustic sensors according to the far-field test standard, and collect aerodynamic sound signals of the same type of wind turbine under different working conditions of healthy blades and damaged blades for multiple times. The collected data contains real-time sound pressure data x(j) of the wind turbine from start to rated working condition. Step 2: The frequency band of the turbulent boundary layer trailing edge aerodynamic noise of the 450W small wind turbine with SD2030 airfoil is about 1000-2000Hz, and the collected acoustic signals are processed by zero mean value and 1000-2000Hz band-pass filter with cutoff frequency; Specifically: Step 2-1: Zero mean value processing: ; In the formula, represents the i-th sound pressure value in the sound pressure data of any group of wind turbines to be tested, represents the i-th sound pressure value after mean value processing.

[0016] Step 2-2: 1000-2000Hz band-pass filter processing with cutoff frequency; ; ; ; In the formula, represents the frequency spectrum of the filtered data; represents the transfer function of the band-pass filter; X(f) represents the frequency spectrum of the time-domain sound pressure signal to be processed.​ represents filtered time domain data; represents inverse Fourier transform.

[0017] Step 3: see Figure 3 , based on acoustic data of blade health and damage, calculate power spectral density and extract power spectral density under feature frequency index, establish BP neural network detection model based on power spectral density characteristics, that is, blade detection model; Specifically: Step 3-1: divide the collected sound pressure data x(j) into training set and test set two parts, wherein the test set contains a group of acoustic data of healthy blades and fault blades under different working conditions, and the remaining data is composed of training set; Step 3-2: divide the sound pressure data X(j) in the test set into K segments, each segment has L data points, calculate the Fourier transform of each segment at frequency index n And power spectral density (PSD); Wherein: ; In the formula, represents the j K th sound pressure data in the K j th segment, here j is the time domain sample index, W(j) is the window function, represents the base function of converting time domain sound pressure signal to frequency domain signal; Then the power spectral density (PSD) of the K K th segment signal at frequency f n is: ; In the formula, U is the average energy of the window function, which is used for compensation in power spectral density; It is calculated by the following formula: ; The estimation of power spectral density at frequency f n is: ; Step 3-3: on the basis of obtaining the power spectral density result, calculate the power spectral density difference between healthy blades and damaged blades in each working condition training set: ; In the formula, represents the power spectral density difference between healthy blades and damaged blades, represents the power spectral density value of healthy blades, represents the power spectral density value of damaged blades; ​Step 3-4: Calculate the mean value of the peak value of the power spectrum density difference, take the mean value of the power spectrum density difference as the threshold for peak value retrieval, retain the peak value data greater than the average value, and obtain the power spectrum density characteristic peak value of each working condition With the frequency index ; Step 3-5: Count the number of times each power spectrum density difference peak value corresponding frequency index appears and the mean value of the number of times, and retain the frequency index greater than the mean value of the number of times , according to the retained frequency index Obtain the peak value of the main power spectrum density difference, and obtain n 1 corresponding power spectrum density characteristic value ; Step 3-6: Normalize n 1 power spectrum density characteristic values, take n 1 PSD characteristic values as input neurons, set the number of output neurons to 2, select traingdx as the training function, select the adaptive momentum gradient descent method as the training method, select the average of the sum of input and output neurons as the hidden layer neuron, and set the target error to 0.001. Calculate the number of hidden layer neurons according to the number of input and output neurons, and obtain the BP neural network training model. Step 3-7: Take the neural network output result [1,1] to represent a healthy blade, and the neural network output result [0,1] to represent a faulty blade. Normalize the peak value data set of the main power spectrum density difference and input it into the BP neural network for iterative calculation. Determine the number of iterations according to the fitting degree of the training error and the target error until the error of the training model is equal to the set target error. Obtain the detection model for judging the health state of the blade. Step 3-8: Verify the detection model using data in the test set. When the output result of the detection model is within the error tolerance range, it proves that the model is effective, and the detection function is derived. Otherwise, return to adjust the training model parameters. Specifically: extract the power spectrum density characteristic values under the frequency index of the test set data and normalize the processing. Input the power spectrum density characteristic values of each verification data into the blade health state detection model, and the output result is within the set error tolerance range. It is proved that the detection model is effective, and the detection function is derived. Otherwise, return to adjust the training model parameters. Step 4: Input the data to be measured after feature extraction into the blade detection model based on BP neural network for calculation, and establish an evaluation system based on the output result to judge the state of the blade. Specifically: Step 4-1: the time domain data y(i) filtered in step 2 is calculated according to step 3-2 to obtain the power spectrum density index The power spectrum density characteristic value is extracted, normalized and introduced into the detection function obtained in step 3-8 to obtain a two-digit output result [U, V]; Step 4-2: the output result [U, V] is evaluated according to the following table by setting a threshold value a, and the state of the blade is judged. In the two-digit output of the detection model, the second digit value close to 1 indicates that the data conforms to the aerodynamic noise characteristics, and the first digit value close to 0 indicates that the probability of blade failure is higher. The first output is taken as 1-U, and the two digits are reserved to output in percentage form, which is the probability of failure. Finally, a reasonable threshold is set to determine whether to send a warning message.

[0018] The wind turbine blade fault detection method based on aerodynamic acoustic signal characteristics provided by the application filters the acoustic data of the wind turbine to be measured based on aerodynamic acoustic theory, compares and analyzes the aerodynamic sound power spectrum density characteristics of the damaged and healthy wind turbine to establish a blade state characteristic extraction method, inputs the blade state characteristics into the BP neural network for training by means of the deep learning theory, obtains a detection model, verifies the detection model using a verification data set, inputs the acoustic characteristics of the wind turbine to be measured into the detection model, and finally judges the health state of the wind turbine blade by combining the evaluation system with the output result of the model. The method can be applied to fault identification of various types of wind turbine blades, has fast detection timeliness and high accuracy, so as to early detect the damage of the wind turbine blade, reduce the safety accidents of the wind turbine, and reduce the operation and maintenance cost of the wind farm.

[0019] The specific embodiments of the application are described above. It should be understood that the application is not limited to the above specific embodiments, and those skilled in the art can make various modifications or changes within the scope of the claims, which does not affect the essential content of the application.

Claims

1. A wind turbine blade fault detection method based on aeroacoustics and deep learning, characterized in that, It comprises the following steps: Step 1: obtaining aerodynamic acoustic data of the wind turbine blade to be tested under healthy state and damaged state; Step 2: zero-mean and filtering processing of the aerodynamic acoustic data based on aerodynamic acoustic theory; Step 3: calculating power spectral density and extracting power spectral density under the index of characteristic frequency based on the aerodynamic acoustic data preprocessed in step 2, and establishing a BP neural network detection model based on power spectral density characteristics; Step 4: inputting the data to be tested after feature extraction into the BP neural network detection model based on power spectral density characteristics for calculation, and establishing an evaluation system based on the output results to judge the state of the blade.

2. The wind turbine blade fault detection method based on aeroacoustics and deep learning according to claim 1, wherein, The sampling time of the aeroacoustic data in step 1 t The following needs to be satisfied: ; wherein is the frequency resolution requirement.

3. The method of claim 1, wherein the method comprises: The step 2 specifically comprises the following steps: Step 2-1: zero-mean processing: ； In the formula, represents the i-th sound pressure value in the sound pressure data of any one group of wind turbines to be measured, represents the i-th sound pressure value after mean value processing; N is the number of current processing data; Step 2-2: band-pass filtering according to the aerodynamic noise frequency band [a, b] radiated by the wind turbine blade, ； ； ； wherein represents the filtered data spectrum; represents the transfer function of the band pass filter; X(f) represents the spectrum of the time domain sound pressure signal to be processed; represents the filtered time domain data; represents the inverse Fourier transform.

4. The method of wind turbine blade fault detection based on aeroacoustics and deep learning as claimed in claim 1, wherein, The step 3 specifically comprises the following steps: Step 3-1: dividing the aerodynamic acoustic data after zero-mean and filtering processing into test set and verification set; Step 3-2: Divide the sound pressure data X(j) in the test set into K segments, each of length L data points, and calculate the Fourier transform result of the segment data at the frequency index K ; n ; ; ; In the formula, Indicates the first K The first in the paragraph j Each sound pressure level data point, here j For time-domain sample index, W(j) For window functions, Let the basis functions be used to convert the time-domain sound pressure signal into a frequency-domain signal; then the first... K Segment signal at frequency f n Power spectral density at for: ； where U is the average energy of the window function used to compensate in the power spectral density; obtaining the power spectral density at the frequency f n n ; Step 3-3: Calculate the power spectral density difference between healthy and broken blades: ; wherein represents the difference in power spectral density between healthy and damaged blades, represents the power spectral density value of healthy blades, represents the power spectral density value of damaged blades; Step 3-4: Peak searching with the mean of the power spectral density difference as the threshold, and the power spectral density characteristic peak value of each working condition is obtained with frequency index ; Step 3-5: Count the number of times each frequency index appears and take the average of the number of times, use the average as the threshold, keep the frequency index greater than the average number of times and n 1 corresponding power spectral density characteristic value ; Step 3-6: The n 1 characteristic values of the power spectral density are normalized to n 1 characteristic values of the power spectral density are inputted, the output is set to 2, the activation function is selected as a sigmod function, the training method is selected as an adaptive momentum gradient descent method, the hidden layer neurons are selected as the average of the sum of the input and output neurons, and the target error is set to ε. Step 3-7: establishing and training the BP neural network detection model, adjusting the number of iterations, and stopping the model iteration when the error of the model reaches the set error ε, and deriving the trained model function; Step 3-8: verifying the BP neural network detection model using the data in the verification set, proving the model effective when the output results of the detection model are within the error allowable range, deriving the detection function, otherwise returning to adjust the training model parameters.

5. The method of wind turbine blade fault detection based on aeroacoustics and deep learning as claimed in claim 1, wherein, The step 4 specifically comprises the following steps: Step 4-1: The time domain data y(i) filtered in step 2 is calculated according to step 3-2 to obtain the power spectrum density index The power spectrum density characteristic value is extracted, normalized, and introduced into the detection function obtained in step 3-8 to obtain a two-digit output result [U, V]. Step 4-2, evaluating the output results [U, V] by setting threshold α, and judging the state of the blade.

6. The method of wind turbine blade fault detection based on aeroacoustics and deep learning as claimed in claim 5, wherein, In the output results [U, V], the second value close to 1 indicates that the data conforms to the aerodynamic noise characteristics, and the first value closer to 0 indicates that the probability of blade failure is higher, taking 1-U of the first output and retaining two digits to output in percentage form, which is the probability of failure, and finally setting a reasonable threshold to determine whether to issue a warning information.

Images