Wind turbine blade fault sound-vibration bimodal collaborative diagnosis method and system

Abstract

The application discloses a wind turbine blade fault sound-vibration dual-mode collaborative diagnosis method and system, comprising: obtaining original sound signals and vibration signals and working condition parameters of a wind turbine; performing collaborative noise reduction on the original sound signals and vibration signals; performing feature extraction and feature fusion on the sound signals and vibration signals subjected to the collaborative noise reduction to obtain a joint feature vector; and inputting the joint feature vector and the working condition parameters into a deep learning model to obtain a health state evaluation result and a confidence degree of the wind turbine blade, so that the sound and vibration dual modes can give full play to respective advantages, and the intelligent diagnosis can be comprehensively and collaboratively performed from data preprocessing, feature extraction to final decision making.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine generator condition monitoring and fault diagnosis technology, and relates to a method and system for collaborative diagnosis of wind turbine blade faults using both acoustic and vibration modes. Background Technology

[0002] Wind energy, as a crucial component of clean and renewable energy, has seen its installed capacity continuously grow globally. Wind turbines, as core equipment for energy conversion, typically operate in harsh and complex natural environments, enduring alternating loads, extreme weather, and other adverse factors. Among these components, the wind turbine blades are critical for capturing wind energy and are also among the most prone to failure. Operating at high speeds for extended periods, blades are highly susceptible to cracking, icing, lightning strike damage, structural cracking, and loose bolts. If these faults are not detected and diagnosed promptly, they can lead to anything from turbine shutdown and power loss to catastrophic accidents such as blade breakage and tower collapse, resulting in significant economic losses and safety risks. Therefore, condition monitoring and early fault diagnosis of wind turbine blades, along with predictive maintenance, are of paramount importance for ensuring the safe, stable, and efficient operation of wind farms.

[0003] Currently, fault diagnosis technologies for wind turbine blades can be mainly categorized as follows: Vibration analysis-based diagnostic methods collect vibration signals during structural operation by installing vibration acceleration sensors at critical locations such as blade roots, bearings, or nacelles, and analyze changes in amplitude, frequency, and modal characteristics to diagnose faults. Vibration analysis is one of the most traditional and commonly used techniques in rotating machinery fault diagnosis. However, this method has significant limitations when applied to blade diagnosis: Limited sensing range: Vibration sensors are typically contact-mounted, with numerous points and a narrow surface area, making it difficult to fully cover an entire blade that can stretch for tens of meters. For early localized damage occurring at the blade tip or leading edge, the vibration signal attenuates significantly after long-distance transmission, resulting in an extremely low signal-to-noise ratio and insufficient diagnostic sensitivity.

[0004] Insensitive to specific fault types: For faults such as icing, which involve a slow increase in mass, the early vibration characteristics change very subtly and are difficult to detect effectively. Furthermore, the inherent vibrations of the blade during operation often mask these early, subtle fault characteristics.

[0005] Strong environmental interference: The operating conditions of wind turbines are complex and variable. Fluctuations in wind speed and rotational speed can cause strong background vibration noise, which overlaps with the fault characteristic frequency band, making feature extraction and fault separation extremely difficult.

[0006] Acoustic / acoustic emission (AE)-based diagnostic methods acquire stress wave signals released by the material during damage processes (such as crack propagation) using acoustic wave sensors or acoustic emission sensors installed on or inside the blade surface. Acoustic emission technology is extremely sensitive to microscopic changes in materials.

[0007] Advantages: Sensitive to early damage, enabling non-directional global monitoring.

[0008] shortcoming: Susceptible to environmental noise pollution: The environmental noise at the site of wind turbines is enormous, including wind noise, rain noise, mechanical friction noise, generator electromagnetic noise, etc. These noises severely overlap with the fault acoustic emission signals in the frequency domain, and traditional filtering methods are very likely to cause the loss of useful signals.

[0009] Signal propagation attenuation and distortion: When sound waves propagate in composite blades, they will attenuate and be distorted, which will introduce errors into the location and identification of fault sources.

[0010] Cost and reliability: High-performance acoustic emission sensors are expensive, and system deployment and data analysis are complex.

[0011] Diagnostic methods based on machine vision (such as drone inspections) involve using drones equipped with high-definition cameras or thermal imagers to conduct regular inspections of blade surfaces and using image recognition technology to detect surface defects such as cracks and lightning strike damage.

[0012] shortcoming: Non-real-time: Usually involves periodic inspections (such as once every six months), making continuous online monitoring impossible and making it difficult to detect sudden faults or developing damage in a timely manner.

[0013] Operations are hampered by weather conditions: rain, snow, fog, dim lighting, and other adverse weather conditions prevent operation, affecting the testing cycle and results.

[0014] Unable to detect internal damage: It can only identify visible surface defects and is powerless to detect structural damage inside the blade, such as skin delamination and web cracking.

[0015] The shortcomings of existing multimodal fusion technologies: In recent years, researchers have attempted to combine information from multiple sensors for diagnostics, such as using vibration and acoustic sensors simultaneously. However, most existing "fusion" methods often remain at a rudimentary stage, merely piecing together data at the data level or simply voting at the decision-making level, failing to achieve deep synergy and complementarity. The core problem lies in: Lack of effective collaborative noise reduction mechanism: Failure to utilize the differences between acoustic and vibration signals in noise source and propagation path to design a collaborative signal processing algorithm that can specifically suppress environmental interference while enhancing real fault characteristics.

[0016] Superficial feature fusion: It simply concatenates acoustic and vibration feature vectors together, ignoring the heterogeneity of the two modal features in terms of physical meaning, scale, and dimension, as well as the deep spatiotemporal correlations and causal relationships between them. This "crude" fusion may not only lead to information redundancy, but may even introduce coupling noise, reducing the performance and generalization ability of the diagnostic model.

[0017] Poor model adaptability: Most diagnostic models are based on fixed thresholds or shallow machine learning algorithms, which make it difficult to adapt to the complex and ever-changing working conditions of wind turbines (variable speed, variable load), resulting in a significant reduction in the robustness and accuracy of the models in practical applications.

[0018] In summary, existing technologies either suffer from blind spots or insufficient sensitivity due to their single sensing dimension, or, although they employ multiple sensors, they fail to achieve efficient, in-depth, and collaborative information fusion. Consequently, the reliability, accuracy, and real-time performance of diagnosing early minor faults in wind turbine blades cannot meet the high requirements of predictive maintenance in industrial settings.

[0019] Therefore, there is an urgent need for a new type of intelligent diagnostic system and method that can fully leverage the advantages of both acoustic and vibration modes and can coordinate data preprocessing, feature extraction and final decision-making in a comprehensive manner, so as to achieve more accurate, earlier and more reliable online monitoring and fault diagnosis of wind turbine blade status. Summary of the Invention

[0020] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for collaborative diagnosis of acoustic and vibration modes of wind turbine blade faults. This method and system can give full play to the advantages of both acoustic and vibration modes and can perform comprehensive collaborative intelligent diagnosis from data preprocessing, feature extraction to final decision-making.

[0021] To achieve the above objectives, this invention discloses a dual-mode collaborative diagnosis method for acoustic and vibration modes of wind turbine blade faults, comprising: Acquire the original acoustic and vibration signals and operating parameters of the wind turbine; Collaborative noise reduction is performed on the original acoustic and vibration signals; Feature extraction and feature fusion are performed on the acoustic and vibration signals after collaborative noise reduction to obtain a joint feature vector; The joint feature vector and operating parameters are input into the deep learning model to obtain the health status assessment results and confidence level of the wind turbine blades.

[0022] Furthermore, acoustic and vibration sensor arrays deployed at the root, leading edge, and trailing edge of the wind turbine blades are used to synchronously acquire the original acoustic and vibration signals of the wind turbine using a data acquisition card.

[0023] Furthermore, the operating parameters of the wind turbine are obtained through the operating condition sensing unit.

[0024] Furthermore, time-domain features, frequency-domain features, and time-frequency-domain features are extracted from the collaboratively denoised acoustic and vibration signals. Then, feature fusion is performed based on a feature fusion network based on channel and spatiotemporal attention to obtain a joint feature vector.

[0025] Furthermore, the deep learning model is a hybrid model of convolutional neural network (CNN) and long short-term memory network (LSTM) or a Transformer model.

[0026] Furthermore, the health status assessment results include normal, cracked, icing, and lightning damage.

[0027] Furthermore, an adaptive joint noise reduction filter is used to perform synergistic noise reduction on the original acoustic and vibration signals.

[0028] This invention discloses a dual-mode collaborative diagnostic system for acoustic and vibration modes of wind turbine blade faults, comprising: The data acquisition layer is used to acquire the raw acoustic and vibration signals and operating parameters of the wind turbine. A data preprocessing and collaborative noise reduction layer is used to collaboratively reduce noise from the original acoustic and vibration signals. The feature extraction and fusion layer is used to extract and fuse features from the acoustic and vibration signals after collaborative noise reduction to obtain a joint feature vector. The intelligent diagnosis and decision-making layer is used to input the joint feature vector and operating parameters into the deep learning model to obtain the health status assessment results and confidence level of the wind turbine blades.

[0029] This invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the wind turbine blade fault acoustic-vibration dual-mode collaborative diagnosis method.

[0030] The present invention discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the wind turbine blade fault acoustic-vibration dual-mode collaborative diagnosis method.

[0031] The present invention has the following beneficial effects: The wind turbine blade fault acoustic-vibration dual-mode collaborative diagnosis method and system of the present invention, in specific operation, performs collaborative noise reduction on the original acoustic and vibration signals, extracts and fuses features of the collaboratively noise-reduced acoustic and vibration signals to obtain a joint feature vector, and inputs the joint feature vector and operating parameters into a deep learning model to obtain the health status assessment result and confidence level of the wind turbine blade. It fully leverages the advantages of both acoustic and vibration modes and can perform comprehensive collaborative intelligent diagnosis from data preprocessing, feature extraction to final decision-making, making it highly practical. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a system architecture diagram of the method of the present invention; Figure 2 This is a flowchart of the method of the present invention; Figure 3 Attention fusion structure diagram. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0036] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0037] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0038] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0039] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0041] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0042] Example 1 The wind turbine blade fault acoustic-vibration dual-mode collaborative diagnosis method of the present invention includes the following steps: Acquire the original acoustic and vibration signals and operating parameters of the wind turbine; Collaborative noise reduction is performed on the original acoustic and vibration signals; Feature extraction and feature fusion are performed on the acoustic and vibration signals after collaborative noise reduction to obtain a joint feature vector; The joint feature vector and operating parameters are input into the deep learning model to obtain the health status assessment results and confidence level of the wind turbine blades.

[0043] Example 2 refer to Figure 1 , Figure 2 and Figure 3 The wind turbine blade fault acoustic-vibration dual-mode collaborative diagnosis system of the present invention includes a data acquisition layer, a data preprocessing and collaborative noise reduction layer, a feature extraction and fusion layer, and an intelligent diagnosis and decision-making layer. Data acquisition layer: This includes acoustic and vibration sensor arrays deployed at the root, leading edge, and trailing edge of the wind turbine blades, as well as a condition sensing unit for collecting SCADA data (such as wind speed, rotational speed, power, and pitch angle). All sensor data are collected synchronously through a data acquisition card to ensure the spatiotemporal synchronization of acoustic and vibration signals.

[0044] Data preprocessing and collaborative noise reduction layer: used for acoustic-vibration signal collaborative noise reduction. Specifically, it receives the original acoustic and vibration signals, and utilizes the characteristic that the vibration signal is sensitive to structural noise in a specific frequency band while the acoustic signal is sensitive to air noise to construct an adaptive joint noise reduction filter. The adaptive joint noise reduction filter uses the signal of another mode as a reference input, and effectively separates and suppresses common environmental noise through improved adaptive filtering algorithms (such as LMS / NLMS) or blind source separation algorithms (such as independent component analysis ICA), while retaining the specific and common characteristics related to the fault.

[0045] Feature extraction and fusion layer: includes a dual-modal feature extraction module and a spatiotemporal feature fusion module.

[0046] The dual-modal feature extraction module extracts time-domain features (such as root mean square value, kurtosis, and peak factor), frequency-domain features (such as the spectrum and spectral centroid obtained by Fast Fourier Transform (FFT), and time-frequency-domain features (such as energy entropy and envelope spectrum features obtained by Wavelet Packet Transform (WPT) or Empirical Mode Decomposition (EMD)) from the denoised acoustic and vibration signals, respectively.

[0047] Spatiotemporal Feature Fusion Module: An attention mechanism is introduced to construct a feature fusion network based on channel and spatiotemporal attention. This network does not simply concatenate features, but automatically learns and assigns different weights to features of different modalities, time points, and frequency bands. It focuses on the features most relevant to the fault under the current operating conditions, achieving deep fusion and adaptive weighting of heterogeneous features to generate a more discriminative joint feature vector.

[0048] The intelligent diagnosis and decision-making layer includes an adaptive intelligent diagnosis module. This module receives the fused joint feature vector and operating condition parameters input from the operating condition sensing unit. It is built based on a deep learning model (such as a hybrid model of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM), or a Transformer model). The operating condition parameters serve as conditional inputs to the model, enabling it to learn normal and fault modes under different operating conditions, thus achieving adaptive fault identification and classification. Finally, this module outputs the blade's health status assessment results and confidence levels, for example, normal, cracked, icing, and lightning damage.

[0049] This invention has the following characteristics: Synergistic effect and improved signal-to-noise ratio: This invention utilizes the innovative acoustic-vibration synergistic noise reduction technology, taking advantage of the differences between the two-mode signals in terms of noise source and propagation path, and using them as mutual references to achieve targeted noise suppression, significantly improving the signal-to-noise ratio of fault characteristics, and laying a solid foundation for subsequent accurate diagnosis.

[0050] Deep fusion enhances discriminative power: This invention employs a feature fusion strategy based on an attention mechanism, which can automatically focus on the sensitive features most relevant to the fault, overcoming the drawbacks of simple feature splicing. The generated fused features are more discriminative, greatly improving the ability to distinguish between different faults.

[0051] Highly adaptive and robust: This invention uses operating parameters as input conditions for the intelligent diagnostic model, enabling the model to sense changes in operating status. This allows it to maintain high diagnostic accuracy under different wind speeds and loads, solving the problem of performance degradation caused by changes in operating conditions in traditional methods.

[0052] Early diagnosis and high reliability: This invention combines the characteristics of acoustic emission being sensitive to microscopic damage and vibration being sensitive to macroscopic anomalies, achieving complementary advantages. It can detect early potential faults in blades, such as microcracks and early icing, earlier, providing a longer early warning window for predictive maintenance.

[0053] With strong systematicity and high practical value, this invention provides a complete and systematic solution from hardware acquisition, signal processing, feature fusion to intelligent decision-making. It has clear logic, distinct levels, and strong engineering practicality and promotional value.

[0054] Example 3 The specific implementation of this invention mainly includes the following hardware and software modules: Data acquisition layer hardware deployment: Acoustic sensor array: In this embodiment, a resonant acoustic emission (AE) sensor with a frequency response range of 20kHz-400kHz is selected. One AE sensor is installed at the leading and trailing edges of each blade root, 2 meters from the center of the hub, to capture high-frequency stress wave signals generated by damage such as crack propagation. The sensors are vacuum-coupled and fixed and sealed with stainless steel clamps and protective covers to withstand the harsh environment outside the nacelle.

[0055] Vibration sensor array: In this embodiment, an ICP-type triaxial accelerometer vibration sensor is selected, with a range of ±50g and a frequency range of 0.5Hz to 10kHz. One vibration sensor is installed in the nacelle, on the main shaft bearing housing and on the base at the gearbox input end, to monitor low-frequency structural vibrations caused by blade imbalance, cracks, and other faults. All sensors are securely mounted using magnetic mounts.

[0056] Data acquisition card: Utilizes an NI cDAQ-9188 Ethernet chassis, equipped with an NI 9234 module (for acquiring vibration sensor signals) and an NI 9232 module (for acquiring acoustic emission sensor signals). The acquisition card is set to synchronous sampling, with sampling rates configured according to signal characteristics: 1 MHz for acoustic emission signals and 25.6 kHz for vibration signals, to ensure sufficient high-frequency fault characteristics can be captured.

[0057] Operating condition sensing unit: Reads and timestamps wind speed, generator speed, output power and pitch angle data in real time from the main control system SCADA of the wind turbine through the OPC UA or Modbus protocol interface of the wind turbine.

[0058] Data processing and diagnostic software platform: The software platform is deployed on an industrial control computer or edge computing gateway and is implemented through signal processing and algorithm programming. The specific modules are as follows: Sound-vibration coordinated noise reduction module: Specific algorithm: The Normalized Least Mean Square (NLMS) adaptive filtering algorithm is adopted.

[0059] Implementation process: The vibration signal is used as the reference input, and the acoustic emission signal is used as the original input. The adaptive filter continuously adjusts its weight coefficients to make the filter output (i.e., the estimated vibration noise component) as close as possible to the associated mechanical noise component in the acoustic emission signal. This estimated noise is subtracted from the original acoustic emission signal to obtain the denoised acoustic emission signal. Similarly, the acoustic emission signal can be used as a reference to suppress the aerodynamic noise component in the vibration signal.

[0060] Feature extraction module: Temporal characteristics: Calculate the root mean square (RMS), kurtosis, waveform factor, and peak factor for each frame of data (frame length is 8192 points).

[0061] Frequency domain features: Perform Fast Fourier Transform (FFT) on the signal to calculate the 1 / 3 octave band spectrum and extract features such as the spectral centroid and mean square frequency.

[0062] Time-frequency domain features: The signal is decomposed into 16 frequency band components by wavelet packet transform (WPT) to the 4th level. The energy proportion and wavelet packet energy entropy of each frequency band component are calculated. Finally, a feature vector containing 12 features is generated for each mode.

[0063] Feature fusion module: Network Structure: A lightweight attention fusion network is constructed. This network receives acoustic feature vectors. and vibration eigenvectors ,like Figure 3 When making a claim.

[0064] The implementation process is as follows: The feature fusion module adopts an adaptive weighting strategy based on channel attention, the mathematical expression of which is as follows: Let the acoustic feature vector extracted after noise reduction be... The vibration characteristic vector is ,in and These are the dimensions of acoustic and vibration characteristics, respectively, in this embodiment... = =12.

[0065] Feature concatenation: First, concatenate the two feature vectors into a joint vector:

[0066] Attention weight generation: The concatenated vector is passed through a fully connected layer (FC) and a softmax activation function to calculate the attention weights for each feature dimension.

[0067]

[0068] in, , These are the parameters for the first layer fully connected layer; , These are the parameters for the second fully connected layer; For activation functions; Let be the attention weight vector, satisfying , satisfying ; Feature weighted fusion: The attention weights are multiplied element-wise with the concatenated feature vectors (Hadamard product) to obtain the weighted fused feature vector.

[0069] in, This indicates element-wise multiplication, a process that allows the model to automatically focus on the most critical cross-modal features, achieving more discriminative feature fusion.

[0070] Intelligent diagnostic module: Model selection: A hybrid model combining one-dimensional convolutional neural network (1D-CNN) and long short-term memory network (LSTM) is adopted.

[0071] Model structure: Input layer: Receives the fused feature vectors And operating data (wind speed, rotational speed).

[0072] The 1D-CNN component contains two convolutional layers (with kernel sizes of 3 and 5, and filter numbers of 64 and 128), used to further extract local deep patterns of features.

[0073] LSTM part: Contains an LSTM layer with 50 units, used to learn the dependencies of features over time series.

[0074] Output layer: A Softmax classifier is used to output the probability distribution of fault types, including: normal, leading edge crack, trailing edge crack, blade icing, and lightning damage.

[0075] Model training: The model is trained using historical data (including acoustic and vibration data and operating condition data with known states). The optimizer is Adam, and the loss function is classification cross-entropy.

[0076] Diagnostic results output: The diagnostic results are displayed in real time through a human-machine interface (HMI) and a diagnostic report can be automatically generated. When the diagnostic confidence exceeds 90% or the health status indicators continue to deteriorate, the system sends a four-level alarm signal (early warning, alarm, danger, emergency) to the wind farm SCADA system via the OPC UA protocol and notifies maintenance personnel.

[0077] The workflow of this invention is as follows: The data acquisition layer continuously collects sound, vibration, and operating condition data, and transmits them to the data processing platform via a local area network. The software platform initiates a diagnostic process at a set interval (e.g., every 10 minutes): the raw data first undergoes preprocessing with sound-vibration collaborative noise reduction; subsequently, the feature extraction module calculates multi-dimensional features from the purified signal; the feature fusion module generates a fused feature vector using an attention mechanism; finally, the intelligent diagnostic module calculates the real-time health status and fault type probability of the blades based on the current operating condition data (wind speed, rotational speed) and the fused feature vector. All diagnostic results and historical data are stored in a database for model optimization and trend analysis.

[0078] The module division in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in each embodiment of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0079] Example 4 A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a dual-modal acoustic-vibration collaborative diagnosis method for wind turbine blade faults. For example, the method includes: acquiring the original acoustic and vibration signals and operating parameters of the wind turbine; performing collaborative noise reduction on the original acoustic and vibration signals; extracting and fusing features from the collaboratively noise-reduced acoustic and vibration signals to obtain a joint feature vector; and inputting the joint feature vector and operating parameters into a deep learning model to obtain a health status assessment result and confidence level of the wind turbine blades. The memory may include main memory, such as high-speed random access memory, or non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry standard architecture bus, a peripheral component interconnection standard bus, an extended industry standard architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory stores the program; specifically, the program may include program code, which includes computer operation instructions. Memory can include main memory and non-volatile memory, and provides instructions and data to the processor.

[0080] Example 5 A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of a dual-modal acoustic-vibration collaborative diagnosis method for wind turbine blade faults. For example, the method includes: acquiring the original acoustic and vibration signals and operating parameters of the wind turbine; performing collaborative noise reduction on the original acoustic and vibration signals; extracting and fusing features from the noise-reduced acoustic and vibration signals to obtain a joint feature vector; and inputting the joint feature vector and operating parameters into a deep learning model to obtain a health status assessment result and confidence level for the wind turbine blades. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0081] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0082] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0083] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0084] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0085] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and disclosure of the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0086] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

[0087] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults, characterized in that, include: Acquire the original acoustic and vibration signals and operating parameters of the wind turbine; Collaborative noise reduction is performed on the original acoustic and vibration signals; Feature extraction and feature fusion are performed on the acoustic and vibration signals after collaborative noise reduction to obtain a joint feature vector; The joint feature vector and operating parameters are input into the deep learning model to obtain the health status assessment results and confidence level of the wind turbine blades.

2. The method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults according to claim 1, characterized in that, The original acoustic and vibration signals of the wind turbine are synchronously collected by a data acquisition card using acoustic sensor arrays and vibration sensor arrays deployed at the root, leading edge, and trailing edge of the wind turbine blades.

3. The method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults according to claim 1, characterized in that, The operating parameters of the wind turbine are obtained through the operating condition sensing unit.

4. The method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults according to claim 1, characterized in that, Temporal, frequency, and time-frequency features are extracted from the acoustic and vibration signals after collaborative noise reduction. Then, feature fusion is performed based on a feature fusion network with channel and spatiotemporal attention to obtain a joint feature vector.

5. The method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults according to claim 1, characterized in that, Deep learning models are either hybrid models of Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM) or Transformer models.

6. The method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults according to claim 1, characterized in that, The health status assessment results include normal, cracked, icing, and lightning damage.

7. The method for coordinated acoustic-vibration dual-mode diagnosis of wind turbine blade faults according to claim 1, characterized in that, An adaptive joint noise reduction filter is used to perform synergistic noise reduction on the original acoustic and vibration signals.

8. A dual-mode collaborative diagnostic system for acoustic and vibration modes of wind turbine blade faults, characterized in that, include: The data acquisition layer is used to acquire the raw acoustic and vibration signals and operating parameters of the wind turbine. A data preprocessing and collaborative noise reduction layer is used to collaboratively reduce noise from the original acoustic and vibration signals. The feature extraction and fusion layer is used to extract and fuse features from the acoustic and vibration signals after collaborative noise reduction to obtain a joint feature vector. The intelligent diagnosis and decision-making layer is used to input the joint feature vector and operating parameters into the deep learning model to obtain the health status assessment results and confidence level of the wind turbine blades.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the wind turbine blade fault acoustic-vibration dual-mode collaborative diagnosis method as described in 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 implements the steps of the dual-mode acoustic-vibration collaborative diagnosis method for wind turbine blade faults as described in any one of claims 1-7.

Images