A wind turbine blade fault diagnosis method and related device

Abstract

The present application belongs to the field of wind turbine fault diagnosis, and discloses a wind turbine blade fault diagnosis method and related device. The method obtains each segment band node of the blade vibration signal through wavelet packet decomposition, calculates the energy value proportion and derives the energy entropy, and then combines the vector regression model and Granger causality measure analysis to calculate the coupling causality measure value between signals. Finally, the energy entropy and the coupling causality measure value are comprehensively used for fault diagnosis. The method solves the problem that the existing diagnosis method cannot effectively identify early defects of the blade, effectively improves the detection sensitivity and accuracy of early structural damage of the blade, can early warn hidden dangers such as cracks and delamination, reduces the operation and maintenance blind area and the failure risk after maintenance, and guarantees the safe and stable operation of the wind turbine.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine fault diagnosis technology, specifically relating to the field of wind turbine blade fault diagnosis, and particularly to a wind turbine blade fault diagnosis method and related device. Background Technology

[0002] In the global transition of energy structure towards clean and renewable energy, wind energy, as a key component, is playing an increasingly important role. Wind turbine blades, as the core component for capturing wind energy, play a decisive role in the operation of wind turbines. However, the wind power industry currently faces many challenges, with blade failure being particularly prominent.

[0003] Wind turbine blades operate in a complex and harsh environment, enduring complex aerodynamic loads, mechanical stresses, and environmental erosion, making them highly susceptible to structural damage such as cracks, corrosion, and delamination. With the increasing size and flexibility of wind turbine blades, inadequate design certification has led to frequent serious accidents such as blade breakage and tower sweeping. Furthermore, existing online monitoring systems for blades are immature, and the operation and maintenance quality assurance system is incomplete, resulting in batch defects, untimely monitoring and early warning of operational damage, and repeated failures after repairs, seriously threatening the safe and stable operation of wind turbines. Currently, there are various methods for fault detection and diagnosis of wind turbine blades, mainly including vibration signal analysis, strain signal analysis, acoustic emission technology, and machine vision technology. However, strain signal analysis methods are greatly affected by the blade surface environment and have limited monitoring range; acoustic emission technology methods are costly and severely affected by environmental interference; and machine vision technology methods are limited by weather and lighting conditions and cannot detect internal structural damage. Overall, vibration signal-based blade fault diagnosis is currently the most widely used technology. It mainly relies on vibration signals to analyze changes in the blade's natural frequency to diagnose blade faults. However, existing experimental studies have shown that early blade defects do not significantly affect their natural frequency. Changes in the natural frequency can only identify mid-to-late stage blade faults, not early blade defects.

[0004] It is evident that existing blade diagnostic methods based on vibration signals cannot effectively identify early-stage blade defects, thus failing to provide effective guidance for blade operation and maintenance. Summary of the Invention

[0005] This invention provides a method and related device for diagnosing wind turbine blade faults. This method can effectively identify early and mid-to-late stage faults in the blades, thereby providing effective guidance for the operation and maintenance of the blades.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for diagnosing wind turbine blade faults includes:

[0008] Based on the predetermined optimal decomposition level, wavelet packet decomposition is performed on the acquired blade vibration signal to obtain the frequency band nodes corresponding to the blade vibration signal.

[0009] The energy value percentage of each frequency band node is calculated based on the frequency band node corresponding to the blade vibration signal; the energy entropy of the blade vibration signal is calculated based on the energy value percentage of each frequency band node.

[0010] Based on a pre-built vector regression model, the coupled Granger causality measure analysis method is used to calculate the coupled causality measure value between vibration signals of each blade.

[0011] Based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade, fault diagnosis is performed on the blades of the wind turbine.

[0012] Further, before performing wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition layer number to obtain the frequency band nodes corresponding to the blade vibration signal, the process includes:

[0013] Based on the noise signal energy and signal-to-noise ratio (SNR), the optimal number of decomposition layers is selected as those with low noise signal energy and high SNR. The specific formula for calculating the noise signal energy is as follows:

[0014]

[0015] In the formula, This represents the energy difference between the reconstructed vibration signal and the original vibration signal; Represents the original vibration signal; This represents the reconstructed vibration signal; N represents the amount of data in the original vibration signal.

[0016] The specific formula for calculating the signal-to-noise ratio is as follows:

[0017]

[0018] In the formula, This indicates the signal-to-noise ratio.

[0019] Further, the step of performing wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition level to obtain the frequency band nodes corresponding to the blade vibration signal includes:

[0020] Based on the predetermined optimal decomposition level, the acquired blade vibration signal is subjected to wavelet packet decomposition, as shown in the following formula:

[0021]

[0022] In the formula, Indicates the first l+1 Layer, 2 k Wavelet packet coefficients of each node; Indicates the first l Layer, First k Wavelet packet coefficients of each node; Indicates the first l+1 Layer, 2 k+1 Wavelet packet coefficients of each node; and These are the filters corresponding to the R-degree function and the wavelet function, respectively; m and n represent the location indices of the vibration monitoring data; i is a positive integer.

[0023] Further, the step of calculating the energy value proportion corresponding to each frequency band node based on the frequency band node corresponding to the blade vibration signal, and calculating the energy entropy of the blade vibration signal based on the energy value proportion corresponding to each frequency band node, includes:

[0024] The specific formula for calculating the energy corresponding to each frequency band node is as follows:

[0025]

[0026] Based on the energy corresponding to each frequency band node, the proportion of energy value corresponding to each frequency band node is calculated, and the specific formula is as follows:

[0027]

[0028] The energy entropy of the blade vibration signal is calculated based on the proportion of energy values ​​corresponding to nodes in each frequency band. The specific formula is as follows:

[0029]

[0030] In the formula, Indicates the first l Layer, First k The energy of each node; Indicates the first l Layer, First k The energy percentage of each node; Entropy represents energy, which is used to measure the degree of disorder in energy distribution.

[0031] Furthermore, the construction process of the pre-built vector regression model is as follows:

[0032] The lag order of the vector regression model is obtained by calculating the information criterion values ​​under different lag orders; the parameters of the vector regression model are estimated by using the least squares method or the maximum likelihood method.

[0033] The vector regression model is constructed based on the lag order and parameters, and its specific expression is as follows:

[0034]

[0035] In the formula, Indicates the first i The vibration monitoring data of the first t One value; , , These are three different coefficients for the vector regression model; Indicates the first i The vibration monitoring data of the first tk One value; Indicates the first j The vibration monitoring data of the first tk One value; This is the error term.

[0036] Furthermore, the calculation of the coupling causality measure value between the vibration signals of each blade based on the pre-constructed vector regression model and the Granger causality measure analysis method includes:

[0037] Based on a pre-built vector regression model, the Granger causality measure analysis method is used to calculate the coupling causality measure value between the vibration signals of each blade. The specific formula is as follows:

[0038]

[0039] In the formula, This represents the coupling causality measure value; Represents the original vibration signal; This represents another original vibration signal; Indicates consideration and Historical information The mean squared prediction error;

[0040] Indicates only considering Historical information The mean squared prediction error; Indicates the first i The vibration monitoring data of the first t Values.

[0041] Furthermore, the fault diagnosis of the wind turbine blades under test based on the energy entropy of the blade vibration signals and the coupling causality measure between the vibration signals of each blade includes:

[0042] Based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade, fault diagnosis is performed on the blades of the wind turbine under test, and the judgment conditions are as follows:

[0043] Condition 1: Does the energy entropy of the blade vibration signal reach the preset energy entropy threshold?

[0044] Condition 2: If the coupling causality measure between the vibration signals of the blade under test and other blades reaches the preset measure threshold, it is determined that the blade under test of the wind turbine has a fault; wherein, the coupling causality measure between the vibration signals of the blade under test and other blades comes from the coupling causality measure between the vibration signals of each blade.

[0045] If condition one and / or condition two are met, then the tested blade of the wind turbine is determined to be faulty.

[0046] A wind turbine blade fault diagnosis system, comprising:

[0047] The decomposition module is used to perform wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition layer number, so as to obtain the frequency band nodes corresponding to the blade vibration signal.

[0048] The first calculation module is used to calculate the energy value ratio of each frequency band node based on the frequency band node corresponding to the blade vibration signal; and to calculate the energy entropy of the blade vibration signal based on the energy value ratio of each frequency band node.

[0049] The second calculation module is used to calculate the coupled causal measure value between the vibration signals of each blade based on the pre-built vector regression model and the Granger causal measure analysis method.

[0050] The fault diagnosis module is used to diagnose faults in the blades of the wind turbine based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade.

[0051] A wind turbine blade fault diagnosis device, comprising:

[0052] Memory, used to store computer programs;

[0053] A processor is used to implement the steps of the above-described wind turbine blade fault diagnosis method when executing the computer program.

[0054] A computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement the steps of the above-described wind turbine blade fault diagnosis method.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] This invention provides a method for diagnosing wind turbine blade faults. This method obtains the frequency band nodes of the blade vibration signal through wavelet packet decomposition, calculates the energy value proportion and derives the energy entropy, and then combines vector regression model and Granger causality measure analysis to calculate the coupling causality measure value between signals. Finally, it integrates the energy entropy and the coupling causality measure value for fault diagnosis. Wavelet packet decomposition finely segments high-frequency signals, capturing local frequency changes caused by early micro-defects, thus overcoming the insensitivity to natural frequencies in traditional vibration analysis. Energy entropy quantifies the complexity changes in signal energy distribution, reflecting nonlinear disturbances caused by early damage. The coupling causality measure reveals the dynamic causal relationship between different signals, enhancing the correlation identification of multi-source vibration data. This method effectively improves the detection sensitivity and accuracy of early structural damage to blades, providing early warning of potential problems such as cracks and delamination, reducing blind spots in operation and maintenance and the risk of post-repair failure, and ensuring the safe and stable operation of wind turbines. Attached Figure Description

[0057] Figure 1 A framework diagram of a wind turbine blade fault diagnosis method provided in an embodiment of the present invention;

[0058] Figure 2 A flowchart of a wind turbine blade fault diagnosis method provided in an embodiment of the present invention;

[0059] Figure 3 This is a schematic diagram of a wind turbine blade fault diagnosis system provided in an embodiment of the present invention. Detailed Implementation

[0060] To further understand the content of this invention, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0061] As described in the background section, machine vision-based methods are limited by weather conditions (such as rain, snow, and fog) and lighting conditions. They perform poorly at night or in inclement weather, and cannot detect internal structural damage (such as delamination or degumming). High-resolution images require significant computational resources and have poor real-time performance, thus they are primarily used for offline detection in practical engineering applications. Overall, vibration signal-based blade fault diagnosis is currently the most widely used technology. It primarily diagnoses blade faults by analyzing changes in the blade's natural frequency based on vibration signals. However, existing experimental studies have shown that early-stage blade defects generally do not affect their natural frequency. Therefore, changes in natural frequency can only identify mid-to-late-stage blade faults, not early-stage defects.

[0062] To address the aforementioned issues, this embodiment provides a method for diagnosing wind turbine blade faults. This method can effectively identify early and mid-to-late stage faults in the blades, providing effective guidance for blade operation and maintenance.

[0063] For example, such as Figure 2 As shown, this embodiment provides a method for diagnosing wind turbine blade faults, including:

[0064] Based on the predetermined optimal decomposition level, wavelet packet decomposition is performed on the acquired blade vibration signal to obtain the frequency band nodes corresponding to the blade vibration signal.

[0065] The energy value percentage of each frequency band node is calculated based on the frequency band node corresponding to the blade vibration signal; the energy entropy of the blade vibration signal is calculated based on the energy value percentage of each frequency band node.

[0066] Based on a pre-built vector regression model, the coupled Granger causality measure analysis method is used to calculate the coupled causality measure value between vibration signals of each blade.

[0067] Based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade, fault diagnosis is performed on the blades of the wind turbine.

[0068] The calibration method provided in this embodiment will be further explained below with reference to the accompanying drawings:

[0069] This embodiment provides a method for diagnosing wind turbine blade faults, the specific steps of which are as follows:

[0070] S1. Collect and process wind turbine blade vibration data: Collect wind turbine blade vibration monitoring data (blade vibration signal), and preprocess it to remove invalid and interference data.

[0071] S2. Wavelet packet decomposition of the blade vibration signal: For the blade vibration signal collected in S1, wavelet packet function is used to decompose it; wherein, the optimal number of decomposition layers is determined according to the noise signal energy and signal-to-noise ratio to obtain the nodes of each frequency band of the blade vibration signal.

[0072] S3. Extraction of blade vibration energy entropy: Based on the blade vibration nodes obtained from S2 decomposition, calculate the energy value and energy ratio of each frequency band node, and calculate the energy entropy of the blade vibration signal according to the energy value ratio of each frequency band node.

[0073] S4. Coupling Measure Analysis between Blade Vibration Signals. For each blade vibration signal collected in step S1, the coupled Granger causality measure analysis method is used to calculate the coupling causality measure between each vibration signal, and a heatmap of the coupling causality measure between blade vibration signals is constructed.

[0074] S5. Blade Fault Diagnosis: Based on the energy entropy of the blade vibration signal obtained from S3 and S4 and the coupling causality measure value between each vibration signal, fault diagnosis is performed on the blade under test.

[0075] The more specific steps are as follows:

[0076] like Figure 1 As shown in the figure, this embodiment provides a method for diagnosing wind turbine blade faults, including the following steps:

[0077] S1. Wind turbine blade vibration monitoring data collection and preprocessing. In this embodiment, vibration monitoring data of the flapping and swaying directions of the three wind turbine blades are collected, totaling blade vibration signals in six directions. The collected blade vibration signals are preprocessed to remove invalid and interfering data.

[0078] S2. Extraction of Energy Entropy from Wind Turbine Blade Vibration. Based on the blade vibration monitoring data, wavelet packet decomposition is used to obtain the energy values ​​and energy proportions of different frequency band nodes, and the energy entropy is calculated. The specific steps are as follows:

[0079] S21. Perform wavelet packet decomposition on the acquired blade vibration monitoring data and read the vibration data at each measuring point. Perform wavelet packet decomposition:

[0080] (1)

[0081] In the formula: Indicates the first l+1 Layer, 2 k Wavelet packet coefficients of each node; Indicates the first l Layer, First k Wavelet packet coefficients of each node; Indicates the first l+1 Layer, 2 k+1 Wavelet packet coefficients of each node; and These are the filters corresponding to the R-degree function (low-pass) and the wavelet function (high-pass), respectively; m and n represent the location index of the vibration monitoring data; i is a positive integer.

[0082] S22. Determine the number of wavelet packet decomposition layers. To determine the optimal number of wavelet packet decomposition layers, it is necessary to consider the signal-to-noise ratio. Signal sampling frequency The target frequency band range is jointly determined. This applies to the reconstructed signal after wavelet packet decomposition at different levels. Signal-to-noise ratio The calculation is as follows:

[0083] (2)

[0084] In the formula: This is the original vibration signal. If the vibration signal is reconstructed... precise, This indicates the noise that has been removed.

[0085] For different signals and signal-to-noise ratios, wavelet packet decomposition will have an appropriate number of decomposition layers to achieve the best or near-best denoising effect while also considering the amount of information. If the number of wavelet packet layers is appropriate, the reconstructed signal can filter out noise and retain the denoised signal; otherwise, it will affect the signal integrity. To quantify the denoising effect under different decomposition layers, this invention quantifies the denoising effect by calculating the energy difference between the reconstructed signal and the original signal under different decomposition layers, i.e., the noise energy, as shown in the following formula:

[0086] (3)

[0087] In the formula: This represents the energy difference between the reconstructed vibration signal and the original vibration signal; N This indicates the amount of data in the signal.

[0088] When selecting the number of decomposition layers, the signal-to-noise ratio and noise energy should be considered, and a decomposition layer with low noise energy and high signal-to-noise ratio should be selected.

[0089] S23. Extract the energy proportion and energy entropy of each frequency band node after wavelet packet decomposition. Calculate the energy of each frequency band node after wavelet packet decomposition, as shown in the following formula:

[0090] (4)

[0091] In the formula: Indicates the first l Layer, First k The energy of each node.

[0092] Calculate the energy percentage of all frequency band nodes to construct the frequency band energy feature vector of each vibration signal of the blade:

[0093] (5)

[0094] In the formula: Indicates the first l Layer, First k The energy percentage of each node.

[0095] Calculate the energy entropy of the blade vibration signal.

[0096] (6)

[0097] In the formula: Entropy represents energy, which is used to measure the degree of disorder in energy distribution.

[0098] S3. Blade Vibration Signal Coupling Measurement Analysis. Based on vibration monitoring data from three blades in six directions, the coupling Granger causality measurement relationships between each pair of blades are analyzed. The specific steps are as follows:

[0099] S31, Constructing Vector Autoregression ( VAR )Model.

[0100] In establishing VAR Before modeling, it is necessary to determine the lag order of the model. p Common methods for determining the lag order include the information criterion method, such as the Akaike information criterion and the Bayesian information criterion. By calculating the information criterion values ​​under different lag orders, the order that minimizes the information criterion value is selected as the optimal order.

[0101] After determining the model lag order, then The model is:

[0102] (7)

[0103] In the formula: Indicates the first i The vibration monitoring data of the first t One value; , , These are the model coefficients; Indicates the first i The vibration monitoring data of the first tk One value; Indicates the first j The vibration monitoring data of the first tk One value; This is the error term.

[0104] for The parameters in the model can be estimated using the least squares method or the maximum likelihood method.

[0105] S32, Granger causality test

[0106] Define the null hypothesis: For the variable Is it a variable? The Granger cause, the null hypothesis is:

[0107] Coefficients of all lagged terms , ,Right now Past values ​​for prediction The current value is not helpful.

[0108] Construct both a restricted model and a complete model; the restricted model contains only... Its own lag term:

[0109] (8)

[0110] The complete model includes Self-lag term and Lag term:

[0111] (9)

[0112] Calculate the test statistic, compare the residual sums of squares (RSS) of the restricted model and the complete model, and calculate the F-statistic:

[0113] (10)

[0114] In the formula: T represents the total sum of squares of the vibration monitoring data; p represents the number of independent variables in the model; n represents the sample size, i.e. the total number of vibration monitoring data.

[0115] The significance level is usually chosen. Based on the F-distribution table, find the degree of freedom. The critical value is determined by the F-statistic. If the calculated F-statistic is greater than the critical value, the null hypothesis is rejected, and Granger causality is considered to exist; otherwise, the null hypothesis is accepted, and Granger causality is considered not to exist.

[0116] S33, Calculation of Coupled Granger Causality Measure

[0117] Coupled Granger causality measures can be compared by... VAR The prediction error of the model is used for definition. This invention uses the Granger Causality Index to quantify the causal measure between different variables. For variables The Granger causality index is defined as:

[0118] (11)

[0119] In the formula: It is a consideration and Historical information The mean square prediction error, Is it only considering Historical information The mean square prediction error.

[0120] S4. Blade Fault Diagnosis. The blade fault diagnosis is performed by analyzing the energy percentage of each frequency band node in the blade vibration signal, energy entropy, and the coupling causality measure between various vibration signals. The specific diagnostic process is as follows:

[0121] S41. For normal blades, the vibration signal energy is more concentrated in the low-frequency range, and the energy entropy value is relatively small. For faulty blades, the high-frequency vibrations excited by the fault cause the energy distribution to gradually shift to the high-frequency part, increasing the proportion of energy at high-frequency nodes and thus increasing the energy entropy. When the blade vibration signal energy entropy reaches the preset energy entropy threshold, it is determined that the current blade is faulty.

[0122] S42. For normal blades, the coupling causality between signals is weak, and the coupling causality measure value is small. When a blade is damaged, the nonlinear vibration and phase difference caused by the damage will lead to changes in the coupling causality between signals, specifically manifested as a significant increase in its causality measure value. When the coupling causality measure value between a blade vibration signal and other blade vibration signals reaches a preset threshold, it is determined that the blade is currently faulty.

[0123] S43. For the fault judgment conditions of S41 and S42, it can be judged that the blade under test has a fault as long as any one of them is met or both of them are met at the same time.

[0124] like Figure 3 As shown in the figure, this embodiment also provides a wind turbine blade fault diagnosis system, including: a decomposition module, used to perform wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition level to obtain the frequency band nodes corresponding to each segment of the blade vibration signal; a first calculation module, used to calculate the energy value ratio corresponding to each frequency band node based on the frequency band nodes corresponding to each segment of the blade vibration signal; and calculate the energy entropy of the blade vibration signal based on the energy value ratio corresponding to each frequency band node; a second calculation module, used to calculate the coupling causality measure value between each blade vibration signal based on a pre-constructed vector regression model coupled with the Granger causality measure analysis method; and a fault diagnosis module, used to perform fault diagnosis on the blade under test of the wind turbine based on the energy entropy of the blade vibration signal and the coupling causality measure value between each blade vibration signal.

[0125] The present invention also provides a wind turbine blade fault diagnosis device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the wind turbine blade fault diagnosis method.

[0126] The present invention also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the wind turbine blade fault diagnosis method.

[0127] When the processor executes the computer program, it implements the steps for diagnosing wind turbine blade faults, such as: performing wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition level to obtain the frequency band nodes corresponding to each segment of the blade vibration signal; calculating the energy value proportion corresponding to each frequency band node based on the frequency band nodes corresponding to each segment of the blade vibration signal; calculating the energy entropy of the blade vibration signal based on the energy value proportion corresponding to each frequency band node; calculating the coupling causality measure value between each blade vibration signal based on a pre-constructed vector regression model coupled with the Granger causality measure analysis method; and diagnosing the faults of the blades under test of the wind turbine based on the energy entropy of the blade vibration signal and the coupling causality measure value between each blade vibration signal.

[0128] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing preset functions, wherein the instruction segments describe the execution process of the computer program in the wind turbine blade fault diagnosis device. For example, the computer program can be divided into a decomposition module, a first calculation module, a second calculation module, and a fault diagnosis module; the specific functions of each module are as follows: the decomposition module is used to perform wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition level to obtain the frequency band nodes corresponding to each segment of the blade vibration signal; the first calculation module is used to calculate the energy value ratio corresponding to each frequency band node based on the frequency band nodes corresponding to each segment of the blade vibration signal; and calculate the energy entropy of the blade vibration signal based on the energy value ratio corresponding to each frequency band node; the second calculation module is used to calculate the coupling causality measure value between each blade vibration signal based on a pre-constructed vector regression model coupled with the Granger causality measure analysis method; the fault diagnosis module is used to perform fault diagnosis on the blade under test of the wind turbine based on the energy entropy of the blade vibration signal and the coupling causality measure value between each blade vibration signal.

[0129] The wind turbine blade fault diagnosis device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The wind turbine blade fault diagnosis device may include, but is not limited to, processors and memory. Those skilled in the art will understand that the above are examples of wind turbine blade fault diagnosis devices and do not constitute a limitation on wind turbine blade fault diagnosis devices. It may include more components than described above, or combine certain components, or use different components. For example, the wind turbine blade fault diagnosis device may also include input / output devices, network access devices, buses, etc.

[0130] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or any conventional processor, etc. The processor is the control center for the wind turbine blade fault diagnosis, connecting various parts of the entire wind turbine blade fault diagnosis equipment through various interfaces and lines.

[0131] The memory can be used to store the computer program and / or modules. The processor implements various functions of the wind turbine blade fault diagnosis device by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory.

[0132] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function (such as sound playback, image playback, etc.). The data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0133] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the wind turbine blade fault diagnosis method described above.

[0134] If the modules / units integrated in the wind turbine blade fault diagnosis system are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0135] Based on this understanding, the present invention can implement all or part of the processes in the above-mentioned wind turbine blade fault diagnosis method, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-mentioned wind turbine blade fault diagnosis method. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or a preset intermediate form, etc.

[0136] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0137] It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0138] In summary, this invention provides a method and related apparatus for diagnosing wind turbine blade faults, which has the following advantages compared to existing diagnostic methods:

[0139] First, this invention performs wavelet packet decomposition on the blade vibration signal to extract the energy proportion and energy entropy of nodes in different frequency bands, and realizes blade fault diagnosis based on the changes in energy proportion and energy entropy. For normal blades, the vibration signal energy is more concentrated in the low-frequency range, and the energy entropy value is smaller; for faulty blades, the high-frequency vibrations excited by the fault cause the energy distribution to gradually shift to the high-frequency part, the energy proportion of high-frequency nodes increases, and the energy entropy also increases accordingly.

[0140] Secondly, this invention analyzes the coupling causality between different vibration signals and aerodynamic noise signals of the blade using the Coupled Granger Causality Measurement (FCCM) method. Fault diagnosis of the blade is achieved based on the changes in the FCCM values ​​between different signals. For normal blades, the coupling causality between signals is weak, and the FCCM values ​​are small. When a blade is damaged, the nonlinear vibration and phase difference caused by the damage lead to changes in the coupling causality between signals, specifically manifested as a significant increase in the FCCM values.

[0141] Third, the wind turbine blade fault diagnosis method based on vibration energy entropy and coupling metric analysis proposed in this invention can effectively identify early and mid-to-late stage blade faults. Among them, the blade fault diagnosis method based on wavelet packet energy entropy has a better identification effect on mid-to-late stage blade faults, while the blade fault diagnosis method based on acoustic-vibration coupling can identify early stage blade faults.

[0142] Fourth, this invention analyzes data from the standard vibration monitoring system for blades, making it highly feasible. It requires no additional monitoring equipment, resulting in low implementation costs. Through centralized deployment, it enables real-time online diagnosis of large-scale wind turbine blades.

[0143] Fifth, the present invention is reasonable and easy to implement, and provides a flexible algorithm system that can give full play to the advantages of data analysis, providing a good foundation for subsequent correction and improvement.

[0144] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for diagnosing faults in wind turbine blades, characterized in that, include: Based on the predetermined optimal decomposition level, wavelet packet decomposition is performed on the acquired blade vibration signal to obtain the frequency band nodes corresponding to the blade vibration signal. The energy value percentage of each frequency band node is calculated based on the frequency band node corresponding to the blade vibration signal; the energy entropy of the blade vibration signal is calculated based on the energy value percentage of each frequency band node. Based on a pre-constructed vector regression model, the coupled Granger causality measure analysis method is used to calculate the coupled causality measure value between vibration signals of each blade. The input data consists of the original vibration signals of each blade in multiple directions, and the coupled causality measure value is obtained based on the coupled Granger causality measure relationship analysis between each pair of original vibration signals. The coupled Granger causality measure value is defined by comparing the prediction errors of different vector regression models, and the Granger causality index is used to quantify the causality measure value between different variables. The Granger causality index is calculated based on the mean square prediction error of vibration monitoring data considering two original vibration signals and the mean square prediction error of vibration monitoring data considering only a single original vibration signal. Based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade, fault diagnosis is performed on the blades of the wind turbine.

2. The wind turbine blade fault diagnosis method according to claim 1, characterized in that, Before performing wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition level to obtain the corresponding frequency band nodes of the blade vibration signal, the process includes: Based on the noise signal energy and signal-to-noise ratio (SNR), the optimal number of decomposition layers is selected as those with low noise signal energy and high SNR. The specific formula for calculating the noise signal energy is as follows: In the formula, This represents the energy difference between the reconstructed vibration signal and the original vibration signal; Represents the original vibration signal; This represents the reconstructed vibration signal; N represents the amount of data in the original vibration signal. The specific formula for calculating the signal-to-noise ratio is as follows: In the formula, This indicates the signal-to-noise ratio.

3. The wind turbine blade fault diagnosis method according to claim 1, characterized in that, The process of performing wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition level to obtain the frequency band nodes corresponding to the blade vibration signal includes: Based on the predetermined optimal decomposition level, the acquired blade vibration signal is subjected to wavelet packet decomposition, as shown in the following formula: In the formula, Indicates the first l+1 Layer, 2 k Wavelet packet coefficients of each node; Indicates the first l Layer, First k Wavelet packet coefficients of each node; Indicates the first l+1 Layer, 2 k+1 Wavelet packet coefficients of each node; and These are the filters corresponding to the R-degree function and the wavelet function, respectively; m and n represent the location indices of the vibration monitoring data; i is a positive integer.

4. The wind turbine blade fault diagnosis method according to claim 3, characterized in that, The calculation of the energy value proportion corresponding to each frequency band node based on the frequency band node corresponding to the blade vibration signal; and the calculation of the energy entropy of the blade vibration signal based on the energy value proportion corresponding to each frequency band node, including: The specific formula for calculating the energy corresponding to each frequency band node is as follows: Based on the energy corresponding to each frequency band node, the proportion of energy value corresponding to each frequency band node is calculated, and the specific formula is as follows: The energy entropy of the blade vibration signal is calculated based on the proportion of energy values ​​corresponding to nodes in each frequency band. The specific formula is as follows: In the formula, Indicates the first l Layer, First k The energy of each node; Indicates the first l Layer, First k The energy percentage of each node; Entropy represents energy, which is used to measure the degree of disorder in energy distribution.

5. The method for diagnosing wind turbine blade faults according to claim 1, characterized in that, The construction process of the pre-built vector regression model is as follows: The lag order of the vector regression model is obtained by calculating the information criterion values ​​under different lag orders; the parameters of the vector regression model are estimated by using the least squares method or the maximum likelihood method. The vector regression model is constructed based on the lag order and parameters, and its specific expression is as follows: In the formula, Indicates the first i The vibration monitoring data of the first t One value; , , These are three different coefficients for the vector regression model; Indicates the first i The vibration monitoring data of the first tk One value; Indicates the first j The vibration monitoring data of the first t- k One value; This is the error term.

6. The wind turbine blade fault diagnosis method according to claim 1, characterized in that, The method, based on a pre-constructed vector regression model and coupled with Granger causality measure analysis, calculates the coupled causality measure values ​​between the vibration signals of each blade, including: Based on a pre-built vector regression model, the Granger causality measure analysis method is used to calculate the coupling causality measure value between the vibration signals of each blade. The specific formula is as follows: In the formula, This represents the coupled causal measure value; Represents the original vibration signal; This represents another original vibration signal; Indicates consideration and Historical information The mean square prediction error; Indicates only considering Historical information The mean square prediction error; Indicates the first i The vibration monitoring data of the first t Values.

7. The method for diagnosing wind turbine blade faults according to claim 1, characterized in that, The method for fault diagnosis of wind turbine blades based on the energy entropy of blade vibration signals and the coupling causality measure between blade vibration signals includes: Based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade, fault diagnosis is performed on the blades of the wind turbine under test, and the judgment conditions are as follows: Condition 1: Does the energy entropy of the blade vibration signal reach the preset energy entropy threshold? Condition 2: If the coupling causality measure between the vibration signals of the blade under test and other blades reaches the preset measure threshold, it is determined that the blade under test of the wind turbine has a fault; wherein, the coupling causality measure between the vibration signals of the blade under test and other blades comes from the coupling causality measure between the vibration signals of each blade. If condition one and / or condition two are met, then the tested blade of the wind turbine is determined to be faulty.

8. A wind turbine blade fault diagnosis system, characterized in that, include: The decomposition module is used to perform wavelet packet decomposition on the acquired blade vibration signal based on a predetermined optimal decomposition layer number, so as to obtain the frequency band nodes corresponding to the blade vibration signal. The first calculation module is used to calculate the energy value ratio of each frequency band node based on the frequency band node corresponding to the blade vibration signal; and to calculate the energy entropy of the blade vibration signal based on the energy value ratio of each frequency band node. The second calculation module is used to calculate the coupled causal measure value between the vibration signals of each blade based on the pre-built vector regression model and the Granger causal measure analysis method. The fault diagnosis module is used to diagnose faults in the blades of the wind turbine based on the energy entropy of the blade vibration signal and the coupling causality measure between the vibration signals of each blade.

9. A wind turbine blade fault diagnosis device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the wind turbine blade fault diagnosis method according to any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it is used to implement the steps of the wind turbine blade fault diagnosis method according to any one of claims 1-7.

Images