Methods and related devices for early warning of abnormal acoustic signatures of wind turbine blade defects

Abstract

This invention discloses a method and related device for early warning of acoustic anomalies in wind turbine blade defects, comprising: collecting raw acoustic signals from inside the nacelle during turbine operation; preprocessing the raw acoustic signals; segmenting and extracting features from the preprocessed acoustic signals to obtain acoustic feature vectors; inputting the acoustic feature vectors into an acoustic baseline model to obtain an anomaly score; and determining whether there are defects in the wind turbine blades based on the anomaly score. This method and related device achieve low-cost, non-contact, highly anti-interference, and highly sensitive online intelligent early warning of wind turbine blade defects.

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 early warning of abnormal acoustic signatures of wind turbine blade defects. Background Technology

[0002] Wind power generation, as a clean and renewable energy source, has experienced rapid development globally. As the core power generation equipment, the safety and reliability of wind turbines directly impact the economic benefits of the entire wind farm. Wind turbine blades are key components for capturing wind energy, constantly exposed to a complex and variable atmospheric environment, enduring enormous alternating aerodynamic, inertial, and gravitational loads. This makes them highly susceptible to defects such as cracks, corrosion, leading-edge erosion, coating peeling, structural cracking, and even breakage. Serious blade failures can not only cause significant economic losses and prolonged downtime but may even lead to catastrophic accidents. Therefore, real-time monitoring of wind turbine blade condition and early defect warning are of paramount importance.

[0003] Currently, monitoring and diagnostic technologies for wind turbine blades can be mainly categorized as follows: Visual inspection methods include manual inspection and drones equipped with high-definition cameras or thermal imagers for photography. These methods rely on the experience of the inspectors, are greatly affected by environmental factors such as light and weather, and are typically conducted on a periodic basis, making it impossible to achieve 24 / 7, real-time online monitoring and difficult to detect early defects inside the blades.

[0004] Vibration monitoring methods: By installing accelerometers at the blade root or on the main structure, changes in the blade's vibration characteristics (such as modal frequencies and damping ratios) are analyzed to determine its structural health. This technology is relatively mature, but it has significant drawbacks: First, the sensors usually need to be installed inside the blade, making retrofitting and wiring difficult and costly for existing units; second, the blade vibration signals are easily affected by strong interference from unit operating conditions (such as speed and power) and external wind conditions, resulting in a low signal-to-noise ratio and difficulty in extracting early, subtle fault characteristics; finally, the vibration sensors are not sensitive enough to aerodynamic defects on the blade surface (such as leading-edge erosion).

[0005] Acoustic and ultrasonic testing methods: Traditional ultrasonic testing requires the testing equipment to contact the blade surface, which is inefficient and unsuitable for online monitoring. Acoustic-based monitoring methods typically involve installing microphone arrays around the unit to collect aerodynamic noise generated by blade rotation, and then using techniques such as beamforming to locate and identify the sound source. However, existing acoustic methods mostly focus on the measurement and analysis of external sound radiation. Their signals are also severely contaminated by background noise such as environmental wind noise, nacelle noise, and gearbox noise. Effective signals are often lost in the strong noise, making it difficult to guarantee the accuracy and reliability of early warnings.

[0006] Monitoring methods based on strain gauges or fiber Bragg grating sensors assess the structural integrity of a blade by measuring strain changes on its surface. While offering high measurement accuracy, these methods suffer from inherent sensor fragility, complex installation processes, poor long-term stability, high maintenance costs, and wiring challenges, making large-scale deployment on wind turbines difficult.

[0007] In summary, the existing technology has the following main drawbacks and limitations: Poor real-time performance: Methods such as visual inspection and traditional ultrasonic inspection cannot achieve 24 / 7 uninterrupted online monitoring, making it difficult to detect sudden defects in a timely manner.

[0008] Weak anti-interference capability: The signals from vibration and external acoustic methods are easily drowned out by unit operating noise and environmental noise, resulting in a low signal-to-noise ratio. Especially under variable operating conditions and complex wind field conditions, it is difficult to extract fault features, leading to a high false alarm and missed alarm rate.

[0009] High deployment costs and poor feasibility: Vibration sensors, strain sensors, etc. require intrusive installation, and retrofitting existing wind turbines involves a large amount of work and high costs. The maintenance and replacement of the sensors themselves are also very difficult.

[0010] Insufficient early warning capability: Most existing methods are effective for macroscopic defects that have reached a certain scale, but they are not sensitive to early and weak defects such as material fatigue, microcrack initiation, and slight coating peeling inside the blade. They lack effective early warning means and cannot achieve "prevention before the problem occurs".

[0011] Low level of intelligence: Many methods still rely on expert experience and threshold judgment, making it difficult to adapt to changes in different units and operating conditions, and lacking intelligent diagnosis and early warning capabilities based on artificial intelligence and self-learning.

[0012] Therefore, there is an urgent need in this field for a new technology that can overcome the above-mentioned shortcomings and achieve low-cost, non-contact, highly anti-interference, and highly sensitive online intelligent early warning of wind turbine blade defects. 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 early warning of abnormal acoustic signatures of wind turbine blade defects. This method and related device achieve low-cost, non-contact, highly anti-interference, and highly sensitive online intelligent early warning of wind turbine blade defects.

[0014] To achieve the above objectives, this invention discloses a method for early warning of abnormal acoustic signatures in wind turbine blade defects, comprising: The raw acoustic signals inside the engine room during unit operation are collected and preprocessed. The preprocessed acoustic signal is segmented and its features are extracted to obtain acoustic feature vectors; The acoustic feature vector is input into the voiceprint baseline model to obtain the anomaly score; The presence of defects in the wind turbine blades is determined based on the anomaly score.

[0015] Furthermore, the preprocessing process for the original acoustic signal is as follows: The original acoustic signal is subjected to noise reduction, filtering, and signal enhancement.

[0016] Furthermore, the process of segmenting and extracting features from the preprocessed acoustic signal to obtain acoustic feature vectors is as follows: the rotor rotation period is determined based on the wind turbine's rotation speed signal, and the continuous acoustic signal is segmented into several periodic signals with the rotor rotation period as the unit; acoustic feature vectors that can characterize the blade state are extracted from each periodic signal.

[0017] Furthermore, this also includes: constructing a voiceprint baseline model.

[0018] Furthermore, the specific steps for constructing the voiceprint baseline model are as follows: When the wind turbine blades are confirmed to be in a healthy state without defects, acoustic feature vectors in the healthy state are collected. Using the acoustic feature vectors in the healthy state, an acoustic baseline model is trained through an unsupervised learning algorithm. The acoustic baseline model is a Gaussian mixture model or a support vector machine. The acoustic baseline model is used to learn and represent the normal pattern of sound of healthy blades.

[0019] Furthermore, if the anomaly score continues to exceed the threshold, it is determined that the wind turbine blades have defects.

[0020] Furthermore, after determining whether the wind turbine blades have defects based on the anomaly score, the process also includes: Based on the judgment results, multi-dimensional early warnings and feedback will be provided.

[0021] This invention discloses a wind turbine blade defect acoustic signature anomaly early warning system, comprising: The acoustic signal acquisition and preprocessing module is used to acquire the raw acoustic signals inside the cabin during unit operation and to preprocess the raw acoustic signals. The speaker signal segmentation and feature extraction module is used to segment and extract features from the preprocessed acoustic signal to obtain acoustic feature vectors. The real-time anomaly calculation and defect identification module is used to input the acoustic feature vector into the acoustic baseline model to obtain anomaly scores; and to determine whether there are defects in the wind turbine blades based on the anomaly scores.

[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 wind turbine blade defect acoustic anomaly early warning method.

[0023] 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 defect acoustic anomaly early warning method.

[0024] The present invention has the following beneficial effects: The wind turbine blade defect acoustic anomaly early warning method and related device of the present invention preprocesses the original acoustic signal, segments and extracts features from the preprocessed acoustic signal to obtain an acoustic feature vector, inputs the acoustic feature vector into the acoustic baseline model to obtain an anomaly score, and determines whether there is a defect in the wind turbine blade based on the anomaly score. By using sensor technology, signal acquisition and processing, feature extraction and selection, machine learning / deep learning models, and early warning design, the method performs online health status monitoring, early fault diagnosis, and intelligent early warning of wind turbine blades, thereby improving the reliability, safety, and power generation efficiency of wind turbine operation and achieving low-cost, non-contact, highly anti-interference, and highly sensitive online intelligent early warning of wind turbine blade defects. 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 Schematic diagram of acoustic sensor deployment Figure 2 This is a flowchart of the method of the present invention; Figure 3 This is a structural block diagram for early warning of abnormal acoustic signatures in wind turbine blade defects. 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 , Figure 2 and Figure 3 The wind turbine blade defect acoustic signature anomaly early warning method of the present invention includes the following steps: 1) Acquire acoustic signals and perform preprocessing Inside the wind turbine nacelle, one or more high-temperature resistant and electromagnetic interference-resistant acoustic sensors are deployed near the hub to collect the sound inside the nacelle during turbine operation. The collected raw sound signals are preprocessed, including noise reduction, filtering, and signal enhancement, to initially suppress steady-state noise interference generated by equipment such as gearboxes, generators, and cooling fans inside the nacelle.

[0036] 2) Segment and extract features from the voiceprint signal; The preprocessed acoustic signal is analyzed. Based on the wind turbine's rotational speed signal or the periodic characteristics calculated from the acoustic signal, the continuous acoustic signal is divided into several periodic signals with the rotor rotation period as the unit. Multi-dimensional acoustic feature vectors that can characterize the blade state are extracted from each periodic signal. The acoustic feature vectors include, but are not limited to: time-domain features (such as root mean square, amplitude, kurtosis), frequency-domain features (such as spectral centroid, spectral width, Mel frequency cepstral coefficients MFCC), and time-frequency-domain features (such as wavelet packet energy spectrum).

[0037] 3) Construct a voiceprint baseline model; When the wind turbine blades are confirmed to be in a healthy state without defects, the system is run for a period of time to collect a sufficient amount of acoustic feature vectors in the healthy state. Using the acoustic feature vectors in the healthy state, an acoustic baseline model is trained by an unsupervised learning algorithm. The acoustic baseline model is used to learn and represent the "normal pattern" of sound of healthy blades. The acoustic baseline model is preferably a Gaussian mixture model (GMM) or a one-class support vector machine (SVM).

[0038] 4) Calculate real-time anomaly degree and identify defects; During the online monitoring phase, steps 1) and 2) are performed on the real-time acquired acoustic signals to obtain real-time acoustic feature vectors. The real-time acoustic feature vectors are then input into the acoustic baseline model constructed in step 3), and the degree of deviation from the healthy "normal mode" is calculated to output an anomaly score. Based on a preset dynamic threshold or adaptive threshold algorithm, the anomaly score is evaluated. If the anomaly score continuously exceeds the threshold, it is determined that the blade has a defect.

[0039] 5) Multi-dimensional early warning and feedback; When a defect is detected in the blade, the system generates early warning information of different levels and notifies the operation and maintenance personnel through the human-machine interface, audible and visual alarm, or remote communication module. The early warning information includes the defect confidence level, abnormal trend spectrum, possible defect type (preliminary judgment by comparison with historical fault database), and suggested handling measures.

[0040] Example 2 Taking a 2.0MW doubly-fed induction generator (DFIG) wind turbine in a wind farm as an example, the implementation steps of the method of the present invention will be described in detail: 1) Acoustic signal acquisition and preprocessing; Acoustic sensors continuously collect mixed sound signals inside the cabin. .

[0041] Preprocessing: First, bandpass filtering is performed with a cutoff frequency set to 100Hz - 10kHz to preserve the aeroacoustic characteristics of the blades and filter out low-frequency vibrations and high-frequency noise.

[0042] Subsequently, spectral subtraction is used for noise reduction. The specific steps are as follows: During a specific period when the wind turbine is idle (the impeller is idling and not connected to the grid), a segment of pure background noise was collected. ), calculate its power spectrum .

[0043] For the acquired real-time signals Perform a short-time Fourier transform (STFT) to obtain its spectrum. .

[0044] Perform spectral subtraction calculation:

[0045] in, This is an over-reduction factor (with a value of 1.2) used to further suppress residual noise.

[0046] Processed spectrum Perform inverse STFT transformation to obtain the enhanced acoustic signal. .

[0047] 2) Voiceprint signal segmentation and feature extraction; Real-time speed signals are obtained through a SCADA system. Assume the current speed is... (rpm), then one impeller rotation cycle (seconds) is:

[0048] The sampling rate of the data acquisition card Hz, therefore the number of sampling points in one period for:

[0049] The system is based on Each sampling point constitutes one frame for the preprocessed audio signal. Perform precise periodic segmentation.

[0050] For each frame of signal (representing the first) (each rotation cycle) to extract the following multi-dimensional acoustic feature vectors: Temporal characteristics: Calculate the kurtosis of the signal in this frame. Kurtosis is exceptionally sensitive to the impact component in the signal, making it ideal for early damage detection.

[0051]

[0052] in, This is the mean value of the signal in that frame. The standard deviation is denoted as .

[0053] Frequency domain characteristics: Calculate Mel-frequency cepstral coefficients (MFCCs). MFCCs conform to the characteristics of human hearing and can well describe the texture features of sound. The specific process is as follows: right Perform pre-emphasis, frame segmentation, and Hamming windowing.

[0054] The power spectrum of each frame is calculated and passed through a 40-channel Mel filter bank.

[0055] Take the logarithm of the filter bank output and then perform a discrete cosine transform (DCT).

[0056] The first 13 coefficients (including the 0th energy term) are used to form a 13-dimensional MFCC eigenvector.

[0057] Time-frequency domain features: Wavelet packet transform (WPT) is performed, with a decomposition layer of 4. Five sub-band nodes with the highest energy concentration from low frequency to high frequency are selected from the 4th layer (e.g., nodes (4,0), (4,1), (4,2), (4,3), (4,4)). The energy of each node is calculated. And normalize to obtain the energy ratio characteristics:

[0058] Finally, 13-dimensional MFCC and 1-dimensional kurtosis will be used. The 5-dimensional wavelet packet energy ratio feature is combined to form a 19-dimensional comprehensive acoustic feature vector. This is used for subsequent modeling.

[0059] 3) Construct a voiceprint baseline model; After the wind turbine is installed, commissioned, and running stably, the blades are confirmed to be in good condition. Under these conditions, acoustic data is continuously collected for more than 72 hours, covering various typical operating conditions such as different wind speeds (e.g., 4m / s - 15m / s) and different power outputs.

[0060] Steps 1) and 2) are performed on the collected massive amounts of data to obtain 19-dimensional health feature vectors for tens of thousands of health states. .

[0061] Using these health feature vectors, a Gaussian Mixture Model (GMM) is trained as the voiceprint baseline model. The voiceprint baseline model is the set of parameters for the trained GMM model, and the probability density function of the GMM is:

[0062] in, The model parameters represent the first, second, and third parts, respectively. The weights, mean vectors, and covariance matrices of each Gaussian component. The number of Gaussian components is taken as in this embodiment. =32. Parameter estimation was performed using the Expectation-Maximization (EM) algorithm. After training, the GMM model defines the probability distribution of healthy voiceprints in the feature space.

[0063] 4) Real-time anomaly calculation and defect identification; During online monitoring, the feature vector is calculated in real time for each rotation cycle. The negative log-likelihood relative to the healthy GMM model is calculated as the anomaly score. :

[0064] The larger the value, the greater the deviation of the current voiceprint characteristics from a healthy pattern.

[0065] To dynamically set the threshold, the system calculates the distribution of anomaly scores for all healthy training samples during the initialization phase and uses the 99th percentile as the initial threshold. In practical applications, the threshold can be fine-tuned based on operational experience; for example, it can be set to... .

[0066] The system employs a continuous triggering mechanism to avoid false alarms caused by accidental interference: when 10 consecutive cycles (approximately 1 minute) occur... All values ​​exceeded the threshold. Only then does the system determine that the blade has a defect and initiate an early warning process.

[0067] 5) Multi-dimensional early warning and feedback; After the system identifies a defect, it generates a warning level based on the anomaly score and duration. For example: Attention level (Watch): Continuously exceeding the threshold .

[0068] Alert level: Continuing to exceed 1.5 .

[0069] Critical level: For more than 2 Or it may rise sharply.

[0070] The early warning information is transmitted to the site monitoring center via fiber optic cable through the switch in the nacelle and integrated into the centralized monitoring system of the wind farm group. It displays information such as the defective unit number, early warning level, occurrence time, and historical abnormal trend curve. At the same time, it sends alarm text messages to the mobile terminals of relevant operation and maintenance personnel to guide them to carry out planned inspections and maintenance.

[0071] This invention has the following characteristics: Strong anti-interference capability and high signal-to-noise ratio: This invention innovatively places acoustic sensors inside the nacelle, cleverly utilizing the nacelle shell as a natural barrier to effectively shield against strong external interference such as environmental wind noise and rain noise. Simultaneously, by focusing on analyzing the periodic acoustic signature characteristics that are strictly synchronized with the impeller rotation, the acoustic signals of the blades themselves can be separated from the complex nacelle background noise, greatly improving the signal-to-noise ratio and monitoring sensitivity.

[0072] Achieving true early warning: This invention constructs a baseline acoustic signature model of a healthy state using unsupervised learning, making the system extremely sensitive to any subtle anomalies deviating from the "normal pattern." This enables the timely detection of early defects in blade materials, such as fatigue, microcrack formation, and fine pinholes, before they develop into macroscopic damage, thus realizing a shift from "fault repair" to "predictive maintenance."

[0073] Easy to deploy and low cost: This invention adopts a non-contact acoustic sensing solution, which does not require the installation of any equipment on the rotating blades or the construction of complex wiring. The sensor can be fixed in the nacelle, which greatly reduces the difficulty and cost of implementation. It is particularly suitable for the retrofitting and large-scale deployment of existing wind turbine units.

[0074] Highly intelligent and highly adaptive: The voiceprint baseline model described in this invention can adapt to sound changes under different generator sets and operating conditions (such as different speeds and power). By learning from various operating condition data under healthy conditions, the system establishes a "normal mode" that is a dynamic range rather than a fixed threshold, thereby reducing false alarms caused by changes in operating conditions and improving the accuracy of early warnings.

[0075] Example 3 The wind turbine blade defect acoustic signature anomaly early warning system of the present invention includes: The acoustic signal acquisition and preprocessing module is used to acquire the raw acoustic signals inside the cabin during unit operation and to preprocess the raw acoustic signals. The speaker signal segmentation and feature extraction module is used to segment and extract features from the preprocessed acoustic signal to obtain acoustic feature vectors. The real-time anomaly calculation and defect identification module is used to input the acoustic feature vector into the acoustic baseline model to obtain anomaly scores; and to determine whether there are defects in the wind turbine blades based on the anomaly scores.

[0076] 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.

[0077] 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 wind turbine blade defect acoustic anomaly early warning method. For example, the method includes: acquiring raw acoustic signals from inside the nacelle during turbine operation; preprocessing the raw acoustic signals; segmenting and extracting features from the preprocessed acoustic signals to obtain acoustic feature vectors; inputting the acoustic feature vectors into an acoustic signature baseline model to obtain an anomaly score; and determining whether a defect exists on the wind turbine blades based on the anomaly score. The memory may include main memory, such as high-speed random access memory, and may also include 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 address bus, data bus, control bus, etc. The memory stores the program; 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.

[0078] Example 5 A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of a wind turbine blade defect acoustic anomaly early warning method. For example, the method includes: collecting raw acoustic signals from inside the nacelle during turbine operation; preprocessing the raw acoustic signals; segmenting and extracting features from the preprocessed acoustic signals to obtain acoustic feature vectors; inputting the acoustic feature vectors into an acoustic signature baseline model to obtain an anomaly score; and determining whether a defect exists on the wind turbine blade based on the anomaly score. 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.

[0079] 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.

[0080] 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.

[0081] 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.

[0082] 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.

[0083] 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.

[0084] 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.

[0085] 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 early warning of abnormal acoustic signatures in wind turbine blade defects, characterized in that, include: The raw acoustic signals inside the engine room during unit operation are collected and preprocessed. The preprocessed acoustic signal is segmented and its features are extracted to obtain acoustic feature vectors; The acoustic feature vector is input into the voiceprint baseline model to obtain the anomaly score; The presence of defects in the wind turbine blades is determined based on the anomaly score.

2. The method for early warning of abnormal acoustic signatures of wind turbine blade defects according to claim 1, characterized in that, The preprocessing process for the original acoustic signal is as follows: The original acoustic signal is subjected to noise reduction, filtering, and signal enhancement.

3. The method for early warning of abnormal acoustic signatures of wind turbine blade defects according to claim 1, characterized in that, The process of segmenting and extracting features from the preprocessed acoustic signal to obtain acoustic feature vectors is as follows: the rotor rotation period is determined based on the wind turbine's rotation speed signal, and the continuous acoustic signal is segmented into several periodic signals with the rotor rotation period as the unit; acoustic feature vectors that can characterize the blade state are extracted from each periodic signal.

4. The method for early warning of abnormal acoustic signatures of wind turbine blade defects according to claim 1, characterized in that, Also includes: Construct a voiceprint baseline model.

5. The method for early warning of abnormal acoustic signatures of wind turbine blade defects according to claim 4, characterized in that, The specific steps for constructing the voiceprint baseline model are as follows: When the wind turbine blades are confirmed to be in a healthy state without defects, acoustic feature vectors in the healthy state are collected. Using the acoustic feature vectors in the healthy state, an acoustic baseline model is trained through an unsupervised learning algorithm. The acoustic baseline model is a Gaussian mixture model or a support vector machine. The acoustic baseline model is used to learn and represent the normal pattern of sound of healthy blades.

6. The method for early warning of abnormal acoustic signatures of wind turbine blade defects according to claim 1, characterized in that, If the anomaly score continues to exceed the threshold, it is determined that there is a defect in the wind turbine blades.

7. The method for early warning of abnormal acoustic signatures of wind turbine blade defects according to claim 1, characterized in that, After determining whether the wind turbine blades have defects based on the anomaly score, the following steps are also included: Based on the judgment results, multi-dimensional early warnings and feedback will be provided.

8. A wind turbine blade defect acoustic signature anomaly early warning system, characterized in that, include: The acoustic signal acquisition and preprocessing module is used to acquire the raw acoustic signals inside the cabin during unit operation and to preprocess the raw acoustic signals. The speaker signal segmentation and feature extraction module is used to segment and extract features from the preprocessed acoustic signal to obtain acoustic feature vectors. The real-time anomaly calculation and defect identification module is used to input the acoustic feature vector into the acoustic baseline model to obtain anomaly scores; and to determine whether there are defects in the wind turbine blades based on the anomaly scores.

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 defect acoustic signature anomaly early warning 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 wind turbine blade defect acoustic anomaly early warning method as described in any one of claims 1-7.

Images