Method and related device for analyzing correlation between abnormal sound of wind turbine blade and surface defect

Abstract

The application discloses a wind turbine blade abnormal sound and surface defect correlation analysis method and related device, comprising: a signal preprocessing and feature extraction module for preprocessing the abnormal acoustic signal and surface image, and extracting acoustic features and visual defect features; a multi-modal data fusion and correlation analysis module for inputting the acoustic features and visual defect features into a pre-trained correlation model for analysis to obtain an analysis result representing the correlation between abnormal sound and defects; an intelligent diagnosis and early warning module for health state evaluation and generation of graded early warning information based on the analysis result, which can solve the problems of early warning lag, poor anti-interference ability, inability to accurately correlate acoustic features and surface defects, and low intelligentization in the prior art.

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 related device for analyzing the correlation between abnormal noise and surface defects in wind turbine blades. Background Technology

[0002] Wind energy, as a clean and renewable energy source, plays a crucial role in the global energy structure transformation. As the core equipment for capturing wind energy, the long-term stable and efficient operation of wind turbines is key to ensuring the economic benefits of wind farms. Wind turbine blades are the first line of defense in energy capture and are also among the most expensive and vulnerable components of the entire unit. They are constantly exposed to complex, variable, and even harsh natural environments, enduring enormous alternating aerodynamic, inertial, and gravitational loads, making them highly susceptible to various surface defects such as leading-edge corrosion, lightning strike damage, coating peeling, cracks, and perforations.

[0003] If these surface defects are not detected and addressed in a timely manner, they will rapidly expand during operation, leading not only to decreased aerodynamic performance, power generation loss, and increased noise, but also potentially catastrophic accidents such as blade breakage and tower collapse, causing enormous economic losses and safety hazards. Therefore, real-time monitoring and early warning of wind turbine blade status are core requirements of modern wind farm intelligent operation and maintenance (OM).

[0004] Currently, the monitoring and detection technologies for the condition of wind turbine blades mainly fall into the following categories: 1. Traditional periodic manual inspection; Operators use telescopes, high-powered cameras, and drones for aerial photography to conduct long-distance observations or close-up inspections of the blade surface. This method is currently the most common, but it has significant drawbacks: First, it relies heavily on the experience and responsibility of the inspectors, resulting in a high rate of missed inspections; second, the inspection cycle is long (usually six months or a year), making it difficult to detect defects occurring between inspections; third, manual tower climbing or close-up inspections using drones are costly, risky, and severely dependent on weather conditions.

[0005] 2. Machine vision-based automatic inspection system; This system continuously acquires images of the blades by installing high-definition cameras on the tower or nacelle, and automatically identifies surface defects such as cracks, dirt, and lightning strikes using image processing algorithms (such as image segmentation, feature extraction, and pattern recognition). This technology represents a significant improvement over manual inspection, but its drawbacks are also quite obvious: First, its monitoring effectiveness is heavily dependent on environmental conditions such as light, weather, and visibility, and it is almost ineffective at night, in rain, fog, snow, or other adverse weather conditions; second, it is difficult to capture defects on the back of the blades and in obscured areas, resulting in monitoring blind spots; finally, this method mainly targets macroscopic defects that are already formed and visible, making it difficult to detect microscopic changes or internal damage in the early stages of defect development, resulting in a significant delay in early warning.

[0006] 3. Condition monitoring based on vibration analysis; This system collects vibration signals during blade operation by installing vibration sensors (such as accelerometers) at the blade root or main structure. It then analyzes changes in vibration frequency, amplitude, and modal characteristics to determine structural health (SHM). While this technology is one of the mainstream methods for condition monitoring, its application to blade monitoring has limitations: First, vibration signals are sensitive to damage at the blade root and other installation locations, but respond weakly to minor surface defects at distant points (such as small-scale corrosion on the leading edge), resulting in a low signal-to-noise ratio and difficulty in achieving early warning. Second, vibration analysis is easily affected by strong interference from normal operating conditions such as turbine start-up and shutdown, pitch changes, and wind speed variations, making feature extraction and fault identification difficult and leading to a high false alarm rate.

[0007] 4. Acoustic / noise-based monitoring methods; In recent years, research has begun to focus on the acoustic signals generated during blade operation. One approach is to indirectly determine the operating status by analyzing abnormal noise emitted by the unit. However, most existing acoustic monitoring methods remain at the level of "abnormal noise detection," meaning they can only determine whether the noise level exceeds the standard or whether there is an anomaly—a reactive alarm mechanism. Its core drawback is that it cannot accurately identify the abnormal sound source (is it from the gearbox, generator, or blades?); moreover, it cannot establish a quantitative correlation model between specific "abnormal sound" characteristics (such as acoustic signatures, spectrum, modulation sidebands, etc.) and the specific type, severity, and geographical location of defects on the blade surface. This means that even when maintenance personnel receive a noise alarm, they still need to spend a significant amount of time and resources investigating, failing to achieve accurate fault location and diagnosis, and limiting the guiding significance of early warnings.

[0008] In summary, existing technologies share the following core drawbacks: Early warning lag: Most methods only issue an alarm when the defect has developed to the point of being macroscopically visible or has a significant impact on the structural strength, thus missing the optimal maintenance window.

[0009] Poor anti-interference ability: It is easily affected by environmental and operating conditions, resulting in high false alarm and missed alarm rates and insufficient reliability.

[0010] Unable to correlate and accurately diagnose: There is a lack of models and methods to correlate the early sensitive indicator of "acoustic characteristics" with the specific fault mode of "surface defects", making it impossible to achieve the leap from "alarm" to "diagnosis".

[0011] Low level of intelligence: It lacks the ability to conduct in-depth analysis based on multi-source data fusion and artificial intelligence, and the system's intelligent early warning and decision support functions are weak.

[0012] Therefore, there is an urgent need in this field for an innovative solution that can overcome the above-mentioned shortcomings, utilize the high sensitivity of acoustic signals to achieve very early detection of blade defects, establish a precise correlation between abnormal sounds and defects, and ultimately form a system that can provide accurate, early, and intelligent early warning. Summary of the Invention

[0013] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and related device for analyzing the correlation between abnormal noise and surface defects in wind turbine blades. This method and related device can solve the problems of delayed early warning, poor anti-interference ability, inability to accurately correlate acoustic features with surface defects, and low level of intelligence in the prior art.

[0014] To achieve the above objectives, this invention discloses a system for analyzing the correlation between abnormal noise and surface defects in wind turbine blades, comprising: The multi-source sensor acquisition module is used to simultaneously acquire abnormal acoustic signals, surface images, and unit operating parameters of the wind turbine blades. The signal preprocessing and feature extraction module is used to preprocess the abnormal acoustic signal and surface image, and extract acoustic features and visual defect features. The multimodal data fusion and correlation analysis module is used to input the acoustic features and visual defect features into a pre-trained correlation model for analysis, and obtain analysis results representing the correlation between abnormal sounds and defects; The intelligent diagnosis and early warning module is used to assess health status and generate graded early warning information based on the analysis results. The human-computer interaction and visualization module is used to display the analysis results and early warning information.

[0015] Furthermore, the multi-source sensing acquisition module includes: An acoustic sensor array, located on the top of the nacelle, is used to collect aerodynamic noise signals from the blades. A high-speed vision acquisition unit, located at the bottom of the nacelle, is used to trigger the capture of image sequences on the blade surface; A multi-source synchronous acquisition unit is used to provide unified time synchronization calibration for abnormal acoustic signals and image data; The operating condition parameter acquisition unit is used to obtain real-time operating condition parameters from the wind turbine SCADA system.

[0016] Furthermore, acoustic features include spectral features, time-frequency features, and modulation features.

[0017] Furthermore, the visual defect features include the geometric and textural features of the defect.

[0018] Furthermore, the association model is a deep learning network model based on the attention mechanism. It learns the nonlinear mapping relationship between acoustic features and visual defect features through training on historical data, and outputs an association probability matrix, where each element represents the association probability between a certain acoustic feature and a certain type of defect.

[0019] Furthermore, the hierarchical early warning strategy of the intelligent diagnosis and early warning module includes: The Comprehensive Health Index (CHI) is calculated based on the defect probability output by the correlation model. Based on the CHI value range, it is divided into four warning levels: normal, attention, warning, and alarm. Early warning information, including defect type, location, and handling suggestions, is generated for different levels.

[0020] Furthermore, the human-computer interaction and visualization module provides sound-visual correlation maps, multimodal data dashboards, and mobile terminal early warning push functions.

[0021] This invention discloses a method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades, including: Abnormal acoustic signals, surface images, and unit operating parameters of the wind turbine blades are collected synchronously through a multi-source sensor acquisition module. The abnormal acoustic signal and surface image are preprocessed by the signal preprocessing and feature extraction module, and acoustic features and visual defect features are extracted. The acoustic features and visual defect features are input into a pre-trained association model through a multimodal data fusion and association analysis module for analysis, and analysis results representing the relationship between abnormal sounds and defects are obtained. The intelligent diagnosis and early warning module assesses health status based on the analysis results and generates graded early warning information. The analysis results and early warning information are displayed through human-computer interaction and visualization modules.

[0022] 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 method for analyzing the correlation between abnormal noise and surface defects of wind turbine blades.

[0023] This invention discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades.

[0024] The present invention has the following beneficial effects: The wind turbine blade abnormal noise and surface defect correlation analysis method and related device described in this invention utilize the high sensitivity of acoustic signals to changes in the microscopic aerodynamic characteristics of the blade surface. This enables very early detection before defects become visible to the naked eye or cause significant vibrations, greatly extending the warning window. Through the established correlation analysis model, a leap from "abnormal noise" to "specific defects" is achieved, accurately diagnosing the type (e.g., cracks, corrosion), severity (size), and approximate location of defects, significantly improving the accuracy of fault diagnosis and guiding maintenance personnel in efficient troubleshooting. By integrating SCADA operating parameters, interference from normal operating conditions such as wind speed and load on acoustic signals can be eliminated during analysis. Combined with the directional analysis of the acoustic sensor array, blade abnormal noise can be effectively distinguished from noise from other sources, significantly reducing false alarm and false negative rates. It should be noted that the entire process of this invention, from data acquisition, feature extraction, intelligent analysis to early warning generation, is fully automated, reducing reliance on human labor, improving the intelligence level of wind farm operation and maintenance, and providing a solid data foundation and decision support for predictive maintenance. This invention, through early and accurate warning, avoids serious damage and catastrophic accidents caused by the expansion of defects, reduces unplanned downtime and expensive maintenance costs, and at the same time ensures the safe operation of wind turbine units, thereby improving power generation and economic benefits. Attached Figure Description

[0025] 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.

[0026] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] 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.

[0032] 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)."

[0033] 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.

[0034] 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.

[0035] Example 1 refer to Figure 1 and Figure 2 The wind turbine blade abnormal noise and surface defect correlation analysis system of the present invention includes a multi-source sensor acquisition module, a signal preprocessing and feature extraction module, a multi-modal data fusion and correlation analysis module, an intelligent diagnosis and early warning module, and a human-computer interaction and visualization module.

[0036] A multi-source sensor acquisition module is used to simultaneously acquire acoustic signals and surface visual images of the blades during operation, including: Acoustic sensor array: used to acquire aerodynamic noise propagating through the air (collected by the acoustic sensor array), which can be used for abnormal sound feature extraction.

[0037] High-speed vision acquisition unit: Employing a high-definition industrial camera (resolution ≥2048×2048, frame rate ≥60fps) with a telephoto lens, mounted at the optimal observation position at the bottom of the nacelle, it is used to capture high-definition images of the blade surface. The camera is equipped with an autofocus mechanism and a protective housing (IP67 rating) to withstand harsh environments such as wind and rain.

[0038] Multi-source synchronous acquisition unit: It adopts a GPS / BeiDou dual-mode synchronous clock module to provide precise time synchronization for acoustic signals and image data (synchronization accuracy ≤1ms), ensuring the time consistency of multimodal data.

[0039] Operating Parameter Acquisition Unit: Interconnected with the wind turbine SCADA system via an industrial Ethernet interface to acquire operating parameters such as wind speed, wind direction, speed, power, and blade pitch angle in real time.

[0040] The signal preprocessing and feature extraction module, deployed in an embedded industrial computer within the cabin, includes: Signal conditioning submodule: Includes a multi-channel preamplifier (60dB gain) and a 4th-order Butterworth anti-aliasing filter (20kHz cutoff frequency) to perform preliminary conditioning of acoustic signals.

[0041] High-speed data acquisition card: It adopts a 24-bit high-precision ADC to simultaneously acquire multiple acoustic signals at a sampling rate of 48kHz, with a dynamic range of ≥110dB.

[0042] Real-time processing unit: Adopting an FPGA+ARM dual-core architecture, it realizes real-time wavelet denoising, frequency domain analysis and feature extraction of acoustic signals, and extracts multi-dimensional feature vectors including time domain features (root mean square, kurtosis, peak factor, etc.), frequency domain features (spectral centroid, frequency band energy, etc.) and time-frequency features (wavelet packet energy entropy).

[0043] Image processing unit: Employs an embedded GPU to perform real-time preprocessing on the acquired images, including image enhancement, distortion correction, defect segmentation, etc., and extracts the geometric features (area, perimeter, aspect ratio) and texture features (contrast, entropy value) of the defects.

[0044] The multimodal data fusion and correlation analysis module, deployed in the wind farm control center server cluster, is the core innovative part of the system and includes: Multimodal feature database: A time-series database is used to store acoustic features, image features, and operating condition data to establish a multimodal dataset.

[0045] Association Analysis Model Library: The association model employs a cross-modal attention mechanism and is trained under supervision using a large amount of labeled data. Training data includes acoustic feature vectors, visual defect feature vectors, and their corresponding defect type, size, and location labels. The loss function is cross-entropy loss, and the optimizer is Adam.

[0046] Real-time analysis engine: Adopting a stream processing architecture, it receives multimodal data in real time, calls the correlation model for analysis, and outputs defect type identification results, severity assessment and location information.

[0047] The intelligent diagnosis and early warning module makes intelligent decisions based on the results of correlation analysis, including: Health status assessment submodule: Calculates the comprehensive health index of the blades based on the defect probability distribution output by the correlation model. The evaluation formula is: ,in, For the first Weights of class defects (satisfying) This is determined based on historical fault data statistics; This represents the probability of this type of defect occurring, as output by the model. This represents the total number of defect types that the system can identify.

[0048] Multi-level early warning submodule: Establishing a four-level early warning mechanism: Level 1 (Normal): >85, only record data; Level 2 (Note): 70< ≤ 85, mark as followable; Level 3 (Warning): 50< If the value is ≤ 70, generate maintenance suggestions; Level 4 (Alarm): If the value is ≤ 50, stop the machine immediately for inspection.

[0049] Early warning information generation unit: Automatically generates early warning reports that include defect type, location, severity, and handling suggestions.

[0050] The human-computer interaction and visualization module provides a complete visual interface, including: Multimodal data dashboard: Simultaneously displays acoustic spectrum, blade surface image, defect markers, and trend curves; Correlation analysis visualization: Displaying the correlation strength between acoustic features and surface defects in the form of a heatmap; Mobile App: Pushes early warning information and on-site inspection guidelines to maintenance personnel.

[0051] The method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades as described in this invention includes the following steps: Acoustic signals, surface images, and operating condition data are acquired synchronously through a multi-source sensor acquisition module. The signal preprocessing and feature extraction module processes the data and extracts features in real time. Multimodal data is transmitted to the central control center via an industrial fiber optic network; Deep correlation analysis is performed using the multimodal data fusion and correlation analysis module; Early warning decisions are generated through the intelligent diagnosis and early warning module; The analysis results and early warning information are displayed through human-computer interaction and visualization modules.

[0052] Example 2 Reference Figure 2(Method Flowchart) This embodiment provides a method for monitoring and early warning of wind turbine blade condition based on the correlation analysis of abnormal noise and surface defects. By establishing a deep correlation model between acoustic features and visual defects, intelligent diagnosis and early warning of blade condition are achieved. Specifically, it includes the following steps: 1) Deployment and synchronous calibration of multi-source sensor acquisition modules.

[0053] Multi-source sensor acquisition modules are deployed on the wind turbine: a high-precision acoustic sensor array (sampling frequency ≥48kHz) is arranged on the top of the nacelle to collect aeroacoustic signals generated during blade rotation from all directions; a high-speed, high-definition camera unit (frame rate ≥60fps) is installed at the bottom of the nacelle to capture images of the blade surface; and the system is connected to the SCADA system to obtain real-time operating parameters of the unit. A GPS synchronization module is used to ensure the time synchronization of acoustic, visual, and operational data, establishing a unified spatiotemporal coordinate system.

[0054] 2) Abnormal acoustic signals and surface conditions are acquired synchronously.

[0055] During normal operation of the wind turbine, the following multimodal data are collected simultaneously: Abnormal acoustic signal acquisition: The noise of the blade passing through is continuously acquired through an acoustic sensor array, with particular attention to the acoustic characteristics in the 200Hz-8kHz frequency band, and the raw sound pressure signal including timestamps is recorded. Surface image acquisition: The high-speed camera is triggered based on the blade azimuth angle signal to capture a high-resolution image sequence of the blade surface (including close-ups of the windward side, leeward side, and leading edge) at the optimal observation position. Operating condition record: Synchronously record SCADA parameters such as wind speed, engine speed, power, and blade pitch angle to provide operating condition background for subsequent analysis.

[0056] 3) Extraction of abnormal sound features and visual detection of defects.

[0057] Preprocessing and feature extraction of the collected multi-source data: Anomalous acoustic signal processing: The sound pressure signal is framed, windowed, and subjected to Fourier transform to extract the following feature sets: Spectral characteristics: 1 / 3 octave band sound pressure level, spectral centroid, spectral bandwidth; Time-frequency characteristics: wavelet packet energy entropy, Mel frequency cepstral coefficients (MFCC); Modulation characteristics: amplitude modulation spectrum, frequency modulation index.

[0058] Surface defect detection: Deep learning-based visual detection algorithms (object detection algorithms based on Faster R-CNN) are used to analyze blade images. Defect localization: Precisely identifies the pixel-level location of defects such as cracks, corrosion, and peeling; Defect quantification: Calculate geometric features such as defect area, length, width-to-depth ratio, etc.; Severity assessment: Calculate the visual severity index based on the defect type and dimensions.

[0059] 4) Construction and analysis of abnormal sound-defect correlation model.

[0060] Establish a quantitative correlation model between acoustic features and visual defects: Multimodal feature fusion: Perform cross-modal fusion on the acoustic feature vector and the visual defect feature vector, adopting an attention mechanism weighted fusion strategy; Correlation model construction: Construct an end-to-end correlation analysis model based on deep neural networks: Input layer: Multimodal feature vector (acoustic features + visual features + operating condition parameters); Hidden layer: Adopt a cross-modal attention mechanism to learn the nonlinear mapping relationship between acoustic features and visual defects; Output layer: Output the abnormal sound-defect correlation probability matrix, which represents the correlation strength between a specific acoustic feature pattern and various types of defects.

[0061] Correlation degree calculation: Calculate the correlation coefficient between the acoustic anomaly index and the visual defect index, and establish the acoustic-visual correlation model: Strong correlation (r>0.8): Based on historical data statistics, acoustic features can effectively indicate surface defects; Moderate correlation (0.5<r≤0.8): Acoustic features need to be combined with other indicators for joint diagnosis; Weak correlation （r≤0.5） : There is no significant correlation between acoustic features and surface defects.

[0062] 5) Intelligent early warning based on the correlation model.

[0063] According to the output result of the correlation model and the Composite Health Index (CHI), implement the following four-level early warning strategy: Normal (CHI>85): The system continues monitoring and only records data.

[0064] Attention (70<CHI ≤ 85): When the acoustic anomaly index exceeds the threshold but the correlation degree is low (the correlation coefficient r of acoustic-visual features ≤ 0.5), or the defect probability output by the correlation model is low, the system marks the blade state as "Attention", and prompts operation and maintenance personnel to strengthen attention and visual monitoring.

[0065] Early warning (50<CHI ≤ 70): When the correlation probability between the acoustic feature and a specific defect type exceeds a preset threshold (e.g., 0.7) and the correlation degree is moderate or above (r>0.5), the system issues an "Early Warning", clearly points out the possible defect type, estimated position and development trend, and generates maintenance suggestions.

[0066] Alarm (CHI ≤ 50): When the acoustic characteristics indicate that the defect has reached a critical state (such as extremely high correlation probability and worsening trend), or when the CHI value drops to the danger range, the system immediately issues an "alarm" command, requiring shutdown and detailed inspection.

[0067] All levels of early warning systems generate and push detailed reports containing defect type, location, severity, and handling suggestions through the human-computer interaction module.

[0068] 6) Visual feedback and model optimization.

[0069] Correlation visualization: Provides an acoustic-visual correlation map to intuitively display the correspondence between abnormal sound features and surface defects; Early warning dashboard: Displays real-time early warning information, defect probability distribution, and recommended handling measures; Model self-learning: The associated model is continuously optimized based on subsequent visual detection results to improve diagnostic accuracy.

[0070] The innovation of this embodiment lies in establishing a quantitative correlation model between blade noise characteristics and surface defects, achieving a technological leap from "hearing abnormalities" to "seeing defects," and solving the industry problem that traditional acoustic monitoring cannot locate and quantify surface defects. Through multimodal data fusion and deep learning technology, the system can accurately infer the blade surface condition based solely on acoustic signals, providing a reliable technical means for preventive maintenance of wind turbine units.

[0071] Example 3 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 method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades. For example, the method includes: simultaneously acquiring abnormal acoustic signals, surface images, and turbine operating parameters of the wind turbine blades through a multi-source sensing acquisition module; preprocessing the abnormal acoustic signals and surface images using a signal preprocessing and feature extraction module, and extracting acoustic features and visual defect features; inputting the acoustic features and visual defect features into a pre-trained correlation model for analysis using a multi-modal data fusion and correlation analysis module, and obtaining analysis results characterizing the correlation between abnormal noise and defects; performing a health status assessment based on the analysis results and generating graded early warning information using an intelligent diagnosis and early warning module; and displaying the analysis results and early warning information using a human-computer interaction and visualization module. The memory may include main memory, such as high-speed random access memory (RAM), 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, or an extended industry-standard architecture bus. The bus can be categorized as an address bus, data bus, or control bus. The memory stores programs; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0072] Example 4 A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of a method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades. For example, the method includes: simultaneously acquiring abnormal acoustic signals, surface images, and turbine operating parameters of the wind turbine blades through a multi-source sensing acquisition module; preprocessing the abnormal acoustic signals and surface images using a signal preprocessing and feature extraction module, and extracting acoustic features and visual defect features; inputting the acoustic features and visual defect features into a pre-trained correlation model for analysis using a multi-modal data fusion and correlation analysis module, and obtaining analysis results characterizing the correlation between abnormal noise and defects; performing a health status assessment based on the analysis results and generating graded early warning information using an intelligent diagnosis and early warning module; and displaying the analysis results and early warning information using a human-computer interaction and visualization module. 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.

[0073] 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.

[0074] 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.

[0075] 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.

[0076] 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.

[0077] 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.

[0078] 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.

[0079] 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 system for analyzing the correlation between abnormal noise and surface defects in wind turbine blades, characterized in that, include: The multi-source sensor acquisition module is used to simultaneously acquire abnormal acoustic signals, surface images, and unit operating parameters of the wind turbine blades. The signal preprocessing and feature extraction module is used to preprocess the abnormal acoustic signal and surface image, and extract acoustic features and visual defect features. The multimodal data fusion and correlation analysis module is used to input the acoustic features and visual defect features into a pre-trained correlation model for analysis, and obtain analysis results representing the correlation between abnormal sounds and defects; The intelligent diagnosis and early warning module is used to assess health status and generate graded early warning information based on the analysis results. The human-computer interaction and visualization module is used to display the analysis results and early warning information.

2. The wind turbine blade abnormal noise and surface defect correlation analysis system according to claim 1, characterized in that, The multi-source sensing acquisition module includes: An acoustic sensor array, located on the top of the nacelle, is used to collect aerodynamic noise signals from the blades. A high-speed vision acquisition unit, located at the bottom of the nacelle, is used to trigger the capture of image sequences on the blade surface; A multi-source synchronous acquisition unit is used to provide unified time synchronization calibration for abnormal acoustic signals and image data; The operating condition parameter acquisition unit is used to obtain real-time operating condition parameters from the wind turbine SCADA system.

3. The wind turbine blade abnormal noise and surface defect correlation analysis system according to claim 1, characterized in that, Acoustic characteristics include spectral characteristics, time-frequency characteristics, and modulation characteristics.

4. The wind turbine blade abnormal noise and surface defect correlation analysis system according to claim 1, characterized in that, The visual defect features include the geometric and textural features of the defect.

5. The wind turbine blade abnormal noise and surface defect correlation analysis system according to claim 1, characterized in that, The association model is a deep learning network model based on the attention mechanism. It learns the nonlinear mapping relationship between acoustic features and visual defect features through historical data training and outputs an association probability matrix, where each element represents the association probability between a certain acoustic feature and a certain type of defect.

6. The wind turbine blade abnormal noise and surface defect correlation analysis system according to claim 1, characterized in that, The tiered early warning strategy of the intelligent diagnosis and early warning module includes: The Comprehensive Health Index (CHI) is calculated based on the defect probability output by the correlation model. Based on the CHI value range, it is divided into four warning levels: normal, attention, warning, and alarm. Early warning information, including defect type, location, and handling suggestions, is generated for different levels.

7. The wind turbine blade abnormal noise and surface defect correlation analysis system according to claim 1, characterized in that, The human-computer interaction and visualization module provides sound-visual correlation maps, multimodal data dashboards, and mobile terminal early warning push functions.

8. A method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades, characterized in that, include: Abnormal acoustic signals, surface images, and unit operating parameters of the wind turbine blades are collected synchronously through a multi-source sensor acquisition module. The abnormal acoustic signal and surface image are preprocessed by the signal preprocessing and feature extraction module, and acoustic features and visual defect features are extracted. The acoustic features and visual defect features are input into a pre-trained association model through a multimodal data fusion and association analysis module for analysis, and analysis results representing the relationship between abnormal sounds and defects are obtained. The intelligent diagnosis and early warning module assesses health status based on the analysis results and generates graded early warning information. The analysis results and early warning information are displayed through human-computer interaction and visualization modules.

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 method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades as described in claim 8.

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 method for analyzing the correlation between abnormal noise and surface defects in wind turbine blades as described in claim 8.

Images