A method and device for monitoring a wind power plant, an electronic device and a storage medium

Abstract

The application provides a wind power equipment monitoring method and device, electronic equipment and storage medium. The method comprises the following steps: acquiring a plurality of spherical microphone array monitoring first original sound signals of the wind power equipment; performing time-frequency conversion processing on each first original sound signal respectively to obtain a spherical harmonic domain signal of each spherical microphone array; constructing an acoustic sensing network beamformer according to the spherical harmonic domain signals of all spherical microphone arrays; performing spatial grid division processing on a region where the wind power equipment is located by using the acoustic sensing network beamformer to obtain a spatial spectrum; performing beam locking processing by using a target direction of the wind power equipment determined according to the spatial spectrum; performing enhancement processing on the locked beam by using the acoustic sensing network beamformer to obtain a target sound signal; and analyzing the target sound signal according to a pre-designed abnormality detection model to obtain a monitoring result of the wind power equipment. In this way, the maintenance cost can be effectively reduced and the accuracy of the monitoring result can be improved by using the present application.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of device monitoring, in particular to a wind power device monitoring method and device, electronic equipment and storage medium. BACKGROUND

[0002] With the deterioration of the environment and the intensification of energy shortages, wind energy as a green energy is increasingly valued. As of 2023, China's wind power equipment quantity leads the world, and wind power accounts for more than 30% of the energy structure, becoming an indispensable force. However, wind turbines often operate in harsh environments, with complex working conditions and high failure risks. If problems cannot be promptly investigated, not only is the maintenance cost high, but it can also cause greater economic losses. Common faults of wind turbines include unit failure, blade damage and tower problems, of which unit and blade failure accounts for 90%. Unit failure is often caused by poor lubrication of the gearbox, overloading operation, material fatigue, manufacturing assembly errors, vibration impact or intrusion of external pollutants, while blades are easily eroded by wind and rain, hail and lightning, leading to leading edge corrosion and cracks.

[0003] Traditional detection methods, such as installing vibration, acoustic emission or stress-strain sensors, can monitor but may affect the structure and increase maintenance costs. Visual detection technology, although tracking and shooting through a pan-tilt or a drone to identify defects, is difficult to accurately detect small cracks in the blade in the case of insufficient light and target movement. SUMMARY

[0004] Therefore, the present application aims to provide a wind power device monitoring method and device, electronic equipment and storage medium, which can effectively reduce maintenance costs and improve the accuracy of monitoring results by analyzing acoustic signals to determine the monitoring results of wind power devices.

[0005] The present application provides a wind power device monitoring method, which comprises:

[0006] Obtaining a plurality of spherical microphone arrays to monitor the first original acoustic signals of the wind power device;

[0007] Respectively performing time-frequency transformation processing on each first original acoustic signal to obtain the spherical harmonic domain signal of each spherical microphone array;

[0008] According to the spherical harmonic domain signals of all spherical microphone arrays, constructing an acoustic sensing net beamformer;

[0009] Using the acoustic sensing net beamformer to perform spatial grid division processing on the area where the wind power device is located to obtain a spatial spectrum;

[0010] Using the target direction of the wind power device determined according to the spatial spectrum to perform beam locking processing;

[0011] The target sound signal is obtained by enhancing the locked beam using the acoustic sensor network beamformer;

[0012] The monitoring result of the wind power equipment is obtained by analyzing the target sound signal according to the pre-designed abnormality detection model.

[0013] Optionally, the first original sound signal of the wind power equipment monitored by each spherical microphone array is obtained by:

[0014] For each spherical microphone array, the second original sound signal of the wind power equipment monitored by each unidirectional microphone on the spherical microphone array is obtained;

[0015] All the second original sound signals of the spherical microphone array are combined to obtain the first original sound signal of the spherical microphone array.

[0016] Optionally, the time-frequency transformation processing is performed on each first original sound signal to obtain the spherical harmonic domain signal of each spherical microphone array, including:

[0017] For each spherical microphone array, the short-time Fourier transform processing is performed on the first original sound signal of the spherical microphone array to obtain the time-frequency spectrum of the spherical microphone array;

[0018] The spherical harmonic domain transformation processing is performed on the time-frequency spectrum of the spherical microphone array to obtain the spherical harmonic domain signal of the spherical microphone array.

[0019] Optionally, the acoustic sensor network beamformer is constructed according to the spherical harmonic domain signals of all the spherical microphone arrays, including:

[0020] The relative distance and azimuth angle between each spherical microphone array and the wind power equipment are obtained;

[0021] The steering vector matrix of the acoustic sensor network is constructed according to the relative distance and azimuth angle between all the spherical microphone arrays and the wind power equipment;

[0022] The spherical harmonic domain signal of the acoustic sensor network is constructed according to the spherical harmonic domain signals of all the spherical microphone arrays;

[0023] The steering vector matrix of the acoustic sensor network and the spherical harmonic domain signal of the acoustic sensor network are combined to obtain the acoustic sensor network beamformer.

[0024] Optionally, the monitoring result of the wind power equipment is obtained by analyzing the target sound signal according to the pre-designed abnormality detection model, including:

[0025] The multi-dimensional voiceprint feature is obtained by performing the voiceprint feature extraction processing on the target sound signal through the convolution layer and the activation function in the abnormality detection model.

[0026] The multi-dimensional voiceprint features are analyzed by a multi-layer perception network classifier in an anomaly detection model to generate a monitoring result of the wind power equipment.

[0027] Optionally, the sound sensing network is constructed by the multiple spherical microphone arrays in a preset arrangement manner, and the sound sensing network is deployed below the wind power equipment.

[0028] Optionally, the monitoring result of the wind power equipment is one of normal wind power equipment, abnormal wind turbine generator, and abnormal wind turbine blade.

[0029] Embodiments of the present application also provide a monitoring device of wind power equipment, which comprises:

[0030] An acquisition module is configured to acquire first original sound signals of multiple spherical microphone arrays monitoring wind power equipment;

[0031] A processing module is configured to perform time-frequency conversion processing on each first original sound signal to obtain a spherical harmonic domain signal of each spherical microphone array;

[0032] A construction module is configured to construct a sound sensing network beamformer according to the spherical harmonic domain signals of all spherical microphone arrays;

[0033] A division module is configured to perform spatial grid division processing on a region where the wind power equipment is located by using the sound sensing network beamformer to obtain a spatial spectrum;

[0034] A locking module is configured to perform beam locking processing by using a target direction of the wind power equipment determined according to the spatial spectrum;

[0035] An enhancement module is configured to perform enhancement processing on the locked beam by using the sound sensing network beamformer to obtain a target sound signal;

[0036] An analysis module is configured to analyze the target sound signal according to a pre-designed anomaly detection model to obtain a monitoring result of the wind power equipment.

[0037] Optionally, when the acquisition module is used to acquire first original sound signals of multiple spherical microphone arrays monitoring wind power equipment, the acquisition module is configured to:

[0038] For each spherical microphone array, the acquisition module is configured to acquire second original sound signals of each unidirectional microphone on the spherical microphone array monitoring wind power equipment;

[0039] The acquisition module is configured to combine all the second original sound signals of the spherical microphone array to obtain the first original sound signal of the spherical microphone array.

[0040] Optionally, when the processing module is configured to perform time-frequency transformation on each first original sound signal respectively to obtain a spherical harmonic domain signal of each spherical microphone array, the processing module is configured to:

[0041] perform short-time Fourier transformation on the first original sound signal of each spherical microphone array to obtain a time-frequency spectrum of the spherical microphone array;

[0042] perform spherical harmonic domain transformation on the time-frequency spectrum of the spherical microphone array to obtain a spherical harmonic domain signal of the spherical microphone array.

[0043] Optionally, when the constructing module is configured to construct the acoustic sensor network beamformer according to the spherical harmonic domain signals of all spherical microphone arrays, the constructing module is configured to:

[0044] obtain a relative distance and an azimuth angle between each spherical microphone array and the wind power equipment;

[0045] construct a steering vector matrix of the acoustic sensor network according to the relative distances and the azimuth angles between all spherical microphone arrays and the wind power equipment;

[0046] construct a spherical harmonic domain signal of the acoustic sensor network according to the spherical harmonic domain signals of all spherical microphone arrays;

[0047] combine the steering vector matrix of the acoustic sensor network and the spherical harmonic domain signal of the acoustic sensor network to obtain the acoustic sensor network beamformer.

[0048] Optionally, when the analyzing module is configured to analyze the target sound signal according to the pre-designed anomaly detection model to obtain the monitoring result of the wind power equipment, the analyzing module is configured to:

[0049] extract a voiceprint feature from the target sound signal through a convolution layer and an activation function in the anomaly detection model to obtain a multi-dimensional voiceprint feature;

[0050] analyze the multi-dimensional voiceprint feature through a multi-layer perception network classifier in the anomaly detection model to generate the monitoring result of the wind power equipment.

[0051] Optionally, the acoustic sensor network is constructed by the plurality of spherical microphone arrays in a preset arrangement manner, and the acoustic sensor network is deployed below the wind power equipment.

[0052] Optionally, the monitoring result of the wind power equipment is one of normal, abnormal wind turbine generator, and abnormal wind turbine blade.

[0053] This application also provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the monitoring method described above are performed.

[0054] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the monitoring method described above.

[0055] This application provides a method, apparatus, electronic device, and storage medium for monitoring wind power equipment. The method includes: acquiring first raw acoustic signals from multiple spherical microphone arrays to monitor wind power equipment; performing time-frequency transformation processing on each first raw acoustic signal to obtain spherical harmonic domain signals for each spherical microphone array; constructing an acoustic sensor network beamformer based on the spherical harmonic domain signals of all spherical microphone arrays; performing spatial grid division processing on the area where the wind power equipment is located using the acoustic sensor network beamformer to obtain a spatial spectrum; performing beam locking processing using the target orientation of the wind power equipment determined based on the spatial spectrum; enhancing the locked beam using the acoustic sensor network beamformer to obtain a target acoustic signal; and analyzing the target acoustic signal according to a pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment.

[0056] In this way, this application constructs an acoustic sensor network using multiple spherical microphone arrays, enabling the constructed acoustic sensor network to possess high directivity characteristics in the high-order spherical harmonic domain. Furthermore, the multi-point layout forms a larger aperture acoustic sensing plane, which not only improves the spatial sampling rate but also enhances the spatial resolution, thereby enabling more accurate capture of acoustic signals from wind turbine units and blades. In addition, the use of a beamformer based on a high-order Ambisonics (HOA) sensor network to acquire target signals significantly improves signal enhancement and anti-interference capabilities. Moreover, combining convolutional neural networks for feature extraction and classification of the spatially enhanced signals significantly improves the accuracy and generalization ability of turbine and blade fault detection.

[0057] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 A flowchart illustrating a method for monitoring wind power equipment provided in an embodiment of this application;

[0060] Figure 2 This application provides a schematic diagram of an acoustic sensor network deployment method.

[0061] Figure 3 This is a schematic diagram of the structure of a monitoring device for wind power equipment provided in an embodiment of this application;

[0062] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application 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 this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0064] With environmental degradation and escalating energy shortages, wind energy, as a green energy source, is gaining increasing attention. As of 2023, my country led the world in the number of wind power installations, and wind power accounted for over 30% of its energy mix, becoming an indispensable force. However, wind turbines often operate in harsh environments with complex operating conditions and a high risk of failure. Failure to promptly identify and address problems not only leads to high maintenance costs but can also result in greater economic losses. Common wind turbine failures include turbine malfunctions, blade damage, and tower issues, with turbine and blade failures accounting for up to 90%. Turbine malfunctions are often caused by poor gearbox lubrication, overload operation, material fatigue, manufacturing and assembly errors, vibration and shock, or the intrusion of external contaminants. Blades, on the other hand, are susceptible to corrosion from wind, rain, hail, and lightning, leading to leading-edge corrosion and cracks.

[0065] Traditional detection methods, such as installing vibration, acoustic emission, or stress-strain sensors, can monitor defects but may affect the structure and increase maintenance costs. Vision-based detection technologies, while using gimbals or drones to track and capture images to identify defects, struggle to accurately detect minute cracks in blades under low light conditions or when the target is moving. Furthermore, for blade defects, there is an acoustic signature detection method that assesses the degree of damage by analyzing the acoustic signal characteristics of defects on the blade surface. This method has attracted considerable attention due to its non-contact, high-efficiency, easy-to-install, and low-cost characteristics.

[0066] For example, the existing technology "A Wind Turbine Fault Detection Method and System Based on AI Auscultation, and a Wind Turbine Safety System" (CN115163426 A) uses a single-channel microphone to collect noise data from wind turbine blades and compares it with existing knowledge bases using various features such as time domain, frequency domain, acoustic signature, and frequency band energy to identify potential faults. However, the omnidirectional acoustic signal acquisition method used in this approach is prone to mixing in background noise from multiple directions, thereby reducing the quality of the acoustic signal. Furthermore, the limitations of similarity-based algorithms in generalization ability may adversely affect the accuracy of fault diagnosis.

[0067] The existing technology, "Remote Auscultation Method for Wind Turbine Blades Based on Acoustic Diagnosis" (CN 112067701 A), captures the sweeping sound of the blades using a single-channel approach. It mainly relies on adaptive techniques to extract octave band features and uses the SVDD algorithm for blade condition monitoring. Although this method has made some progress in signal processing, it still struggles to eliminate strong environmental noise. Moreover, the SVDD detection model is computationally complex when processing high-dimensional data, has limitations in defining spherical boundaries, and lacks multi-class recognition capabilities, making it difficult to meet the generalization and expansion requirements for blade fault detection.

[0068] The existing technology, "A Non-Contact Monitoring System and Method for Offshore Wind Turbine Blade Faults" (CN109763944A), collects aeroacoustic signals through marine meteorological and hydrological monitoring nodes and uses neural networks to analyze the blade sweep signals to determine anomalies. However, this method has shortcomings in acoustic signal acquisition and requires improvement in acoustic signal enhancement algorithms, which directly affects the accuracy of fault detection.

[0069] Another existing technology, "An Acoustic Detection Method and System for Wind Turbine Blade Faults Based on Deep Learning" (CN116631442 A), uses a microphone array to capture acoustic signals, performs spatial filtering and enhancement processing on the target signal using beamforming technology, and then uses deep learning to classify Mel-spectrum features to achieve anomaly detection. Although this method improves noise extraction capability through linear array signal processing, the inconsistent spatial directivity of the linear array increases deployment difficulty, affects the effect of spatial filtering enhancement, and thus reduces the overall performance of blade fault detection.

[0070] Based on this, embodiments of this application provide a method, device, electronic device, and storage medium for monitoring wind power equipment. By designing an acoustic sensor network and analyzing the acoustic signals collected by the acoustic sensor network to determine the monitoring results of the wind power equipment, maintenance costs can be effectively reduced and the accuracy of monitoring results can be improved.

[0071] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for monitoring wind power equipment provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the monitoring method includes:

[0072] S101. Acquire the first raw acoustic signal of the wind power equipment by multiple spherical microphone arrays.

[0073] S102. Perform time-frequency transformation processing on each of the first original acoustic signals to obtain the spherical harmonic domain signals of each spherical microphone array.

[0074] S103. Construct an acoustic sensor beamformer based on the spherical harmonic domain signals of all spherical microphone arrays.

[0075] S104. Use the acoustic sensor network beamformer to perform spatial grid division processing on the area where the wind power equipment is located to obtain the spatial spectrum.

[0076] S105. Beam locking is performed using the target azimuth of the wind power equipment determined based on the spatial spectrum.

[0077] S106. The locked beam is enhanced using a beamformer of an acoustic sensor network to obtain the target acoustic signal.

[0078] S107. Analyze the target acoustic signal according to the pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment.

[0079] For step S101, the first raw acoustic signal obtained by each spherical microphone array in the high-order surround sound sensor network for monitoring the wind power equipment is acquired.

[0080] Here, the Higher Order Ambisonics (HOA) sound sensor network includes multiple spherical microphone arrays, and each spherical microphone array includes multiple unidirectional microphones.

[0081] The acoustic sensor network can monitor wind power equipment in real time or periodically. Real-time monitoring is generally preferred.

[0082] For an example, please refer to Figure 2 , Figure 2This application provides a schematic diagram of a deployment method for an acoustic sensor network, as shown below. Figure 2 As shown, a high-order surround sound sensor is obtained by constructing an acoustic sensor network using multiple spherical microphone arrays arranged in a preset manner; the acoustic sensor network is deployed below the wind power equipment. Here, the spherical microphone array is a high-order surround spherical acoustic array, thereby achieving a larger aperture sound field sensing surface and higher spatial resolution.

[0083] Continuing with step S101, in one embodiment provided in this application, acquiring the first raw acoustic signal of a wind power equipment monitored by a plurality of spherical microphone arrays includes:

[0084] S1011. For each spherical microphone array, acquire the second original acoustic signal of each unidirectional microphone on the spherical microphone array that monitors the wind power equipment.

[0085] S1012. Combine all the second original sound signals using the spherical microphone array to obtain the first original sound signal of the spherical microphone array.

[0086] For step S1011, for each spherical microphone array, the original sound signals collected by each unidirectional microphone distributed at different positions in the spherical array are acquired, resulting in multiple first original sound signals. In this way, the sound signals emitted by the wind power equipment can be recorded from multiple directions.

[0087] For step S1012, for each spherical microphone array, the "second original sound signal" collected by all unidirectional microphones in the spherical microphone array is combined in a predetermined combination method to form a composite signal (i.e., the "first original sound signal").

[0088] For example, the predetermined combination method includes the following methods: signal superposition and averaging, weighted averaging, time-frequency domain analysis, etc.

[0089] Regarding step S102, in one embodiment provided in this application, the step of performing time-frequency transformation processing on each first original acoustic signal to obtain the spherical harmonic domain signal of each spherical microphone array includes:

[0090] S1021. For each spherical microphone array, perform a short-time Fourier transform on the first original acoustic signal of the spherical microphone array to obtain the time-frequency spectrum of the spherical microphone array.

[0091] S1022. Perform spherical harmonic domain transformation on the time-frequency spectrum of the spherical microphone array to obtain the spherical harmonic domain signal of the spherical microphone array.

[0092] In step S1021, a short-time Fourier transform is performed on each first original acoustic signal to determine the frequency characteristics of the signal as it changes over time, thereby helping to capture the dynamic frequency information in the acoustic signal of the wind power equipment.

[0093] Regarding step S1022, the time-frequency spectrum of each spherical microphone array is further processed by spherical harmonic domain transformation, which decomposes the sound signals in different directions into multiple spherical harmonic components, which helps to identify and locate the sound source direction and frequency characteristics of wind power equipment.

[0094] Therefore, according to steps S1021 and S1022, signal conversion from the time-frequency domain to the spatial domain can be realized, enabling each spherical microphone array to accurately analyze the acoustic characteristics of wind power equipment in time, frequency, and space.

[0095] Furthermore, this application also provides the specific mathematical representation and determination process of the spherical harmonic domain signal of the spherical microphone array, as detailed below:

[0096] Assume a spherical microphone array contains Q unidirectional microphones, uniformly distributed on a rigid sphere of radius r. There are L far-field sound sources in space, and a sound propagation model for the rigid sphere is obtained. The spatial coordinates of the unidirectional microphones are:

[0097] r q =[rcos(θ) q sin(φ) q ),rsin(θ q sin(φ) q ),rcos(φ q )] T ,q=1,2……Q;

[0098] Where θ q The measurement direction is counterclockwise from the x-axis, φ q It is the elevation angle measured downwards from the z-axis.

[0099] Similarly, the coordinates of the L sound sources are:

[0100] r l =[rcos(θ) l sin(φ) l ),rsin(θ l sin(φ) l ),rcos(φ l )] T ,l=1,…,L.

[0101] The received signal model of the spherical microphone array can be represented as:

[0102] p(k)=A(k)S(k)+z(k),

[0103] Where k is the wave number, i.e., k = 2πf / c, f is the frequency, and c is the speed of sound. p(k) is the sound pressure signal vector, A(k) is the steering matrix, S(k) is the sound source sound pressure signal vector, and z(k) is the sensor noise vector. p(k), A(k), and S(k) are respectively expressed as:

[0104] p(k)=[p(k,r1),p(k,r2),…,p(k,r Q )] T ,

[0105] A(k)=[a(k,r1),a(k,r2),…,a(k,r L )] T ,

[0106] S(k) = [s1(k), s2(k), ..., s L (k)] T ,

[0107] Where p(k,r) q Let a(k,r) be the acoustic signal received by the q-th unidirectional microphone. l ) and s l (k) represents the steering vector and sound pressure intensity of the l-th sound source, respectively. l )for:

[0108]

[0109] Where k l for:

[0110] k l =-k[cos(θ) l sin(φ) l ),sin(θ l sin(φ) l ),cos(φ l )] T i 2 =-1.

[0111] According to the spherical harmonic transformation It can be approximately decomposed into a truncated Nth-order spherical harmonic:

[0112]

[0113] Where b n (kr) represents the mode strength for a unidirectional microphone:

[0114] b n (kr)=4πin [j n (kr)-ij′ n (kr)],

[0115] in For spherical harmonic basis functions:

[0116]

[0117] For Legendre function, and j n It is a spherical Bessel function of the first kind. Therefore, the guiding vector can be expressed as:

[0118] A(k)=Y(Ω)B(kr)Y H (Γ),

[0119] Where Y(Ω) is Q×(N+1) 2 The matrix:

[0120] Y(Ω)=[y T (θ1,φ1),…,y T (θ Q ,φ Q )] T ,

[0121]

[0122] From the spherical harmonics of L sound sources, we can obtain L×(N+1). 2 The matrix Y H (Γ):

[0123] Y H (Γ)=[y T (θ1,φ1),…,y T (θ L ,φ L )],

[0124] And (N+1) 2 ×(N+1) 2 The diagonal matrix B(kr):

[0125] B(kr)=diag{b0(kr),b1(kr),b1(kr),…,b N (kr)}.

[0126] Therefore, we can conclude that:

[0127] p(k)=Y(Ω)B(kr)Y H (Γ)S(k)+z(k).

[0128] Due to the orthogonality principle of spherical harmonics, for approximately uniform spherical space sampling, Y H (Ω)Y(Ω)=I, from which we can obtain p nm (k) is:

[0129] p nm (k)=B(kr)Y H (Γ)S(k)+z nm (k),

[0130] Where p nm (k)=Y H (Ω)p(k). For a given spherical array configuration, we can obtain the spherical harmonic domain signal model, whose characteristic coefficients q nm (k):

[0131]

[0132] Where q nm (k)=B -1 (kr)p nm (k).

[0133] In this way, the final spherical harmonic domain signal model after linear transformation is determined.

[0134] For step S103, in this step, the acoustic sensor network beamformer is constructed based on the spherical harmonic domain signals of all spherical microphone arrays on the acoustic sensor network.

[0135] In one embodiment provided in this application, constructing an acoustic sensor network beamformer based on the spherical harmonic domain signals of all spherical microphone arrays includes:

[0136] S1031. Obtain the relative distance and azimuth angle between each spherical microphone array and the wind power equipment.

[0137] S1032. Construct the guiding vector matrix of the acoustic sensor network based on the relative distance and azimuth angle between all spherical microphone arrays and the wind power equipment.

[0138] S1033. Construct the spherical harmonic domain signal of the acoustic sensor network based on the spherical harmonic domain signals of all spherical microphone arrays.

[0139] S1034. The guiding vector matrix of the acoustic sensor network and the spherical harmonic domain signal of the acoustic sensor network are combined and processed to obtain the acoustic sensor network beamformer.

[0140] The data in step S1031 can be directly determined after the acoustic sensor network is deployed. The specific mathematical calculations for steps S1031 to S1034 are as follows:

[0141] Consider a sensor network with C HOA nodes (spherical microphone arrays) randomly deployed in the measurement space, such as... Figure 2 As shown. According to the spherical attenuation model, the acoustic signal p received by the c-th node... c (k)

[0142] p c (k)=α c ⊙W c (k)⊙A c (k)s+z(k)

[0143] in r is the spherical wave attenuation factor. c,l W represents the distance from the l-th sound source to the c-th sensor node. c (k) is the phase shift factor of the c-th node:

[0144] W c (k)=[exp(-jkr c,1 ),…,exp(-jkr c,L )] T

[0145] A c (k) is the steering vector of the c-th sensing node. We can obtain the following variation:

[0146]

[0147] in Substituting the spherical harmonic signal model determined in step S102, we obtain:

[0148]

[0149] Where Γ c Let L be the position coordinate vectors of the L sound sources relative to the c-th node. The spherical harmonic characteristic of the c-th node can be obtained from the following formula:

[0150]

[0151] We can further obtain the received signal of the entire HOA acoustic sensor network.

[0152]

[0153] in It can be obtained through a spherical harmonic signal model of multiple spherical microphone arrays:

[0154]

[0155] HOA network spherical harmonic domain steering vector for:

[0156] Y HOAN (ρ)=[Y T (Γ1),…,Y T (Γ C )] T

[0157] ρ={Γ1,Γ2,…,Γ C}

[0158] Therefore, we can further derive a beamformer based on the HOA sensor network. For broadband sound sources, acoustic imaging under reverberation conditions can be achieved by smoothing in the selected time-frequency domain using the short-time Fourier transform (STFT). The HOA signal model in the spherical harmonic domain can be expressed as:

[0159]

[0160] Where E represents the time-frequency space of the acoustic signal. Furthermore, we can derive a beamformer based on the spherical harmonic domain:

[0161]

[0162] And the spatial spectrum is S p,HOAN :

[0163]

[0164] Where N τ and N v These represent time and frequency frames, respectively.

[0165] In this way, the sound field can be reconstructed based on the determined acoustic sensor beamformer.

[0166] In step S104, the area where the wind power equipment is located is pre-divided into spatial units to form a spatial grid (where each grid represents a spatial region and can be considered a different monitoring point). A beamformer processes the acoustic signals within each grid to generate the acoustic signal intensity or spectral information for each grid, and these information are combined to form the spatial spectrum of the entire monitoring area.

[0167] In this way, acoustic image reconstruction can be performed based on the spatial spectrum, and the location of the turbine and blade noise can be identified through the reconstructed sound field. That is, the location of the wind power equipment can be determined.

[0168] Regarding step S105, in this step, the target azimuth of the wind power equipment is determined based on the spatial spectrum, and then the determined target azimuth is used to lock the beam corresponding to the wind power equipment from multiple beams.

[0169] It should be noted that the spatial spectrum shows the sound energy distribution in the area where the wind power equipment is located. By analyzing the intensity peaks of the spatial spectrum, the location of the strongest sound can be identified, which usually corresponds to the specific location of the wind power equipment.

[0170] A beamformer can generate multiple beams in different directions. After determining the target location, it can select and lock the beam that is aligned with the direction of the wind power equipment from these beams.

[0171] Regarding step S106, the enhancement process in this step may, for example, include techniques such as increasing beam gain, optimizing signal phase consistency, and eliminating background noise.

[0172] Regarding step S107, in one embodiment provided in this application, the step of analyzing the target acoustic signal according to a pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment includes:

[0173] S1071. The target acoustic signal is processed by extracting voiceprint features through the convolutional layer and activation function in the anomaly detection model to obtain multidimensional voiceprint features.

[0174] S1072. The multi-dimensional voiceprint features are analyzed by the multi-layer perceptron classifier in the anomaly detection model to generate monitoring results for wind power equipment.

[0175] Here, the anomaly detection model can be an anomaly classifier based on a convolutional neural network, which is a 2D convolutional network with a kernel size of (N+1). 2 After processing the 2×1 data through N layers of convolution and Rectified Linear Unit (ReLU) activation functions (with N_(layer,1),…,N_(layer,L) convolution channels respectively), N_Feats of voiceprint features are finally obtained. Finally, a Multi-Layer Perception Network (MLP) classifier is used to process the multi-dimensional voiceprint features, generating and outputting monitoring results for wind power equipment.

[0176] Here, the generated monitoring result of the wind power equipment is one of three results: wind power equipment is normal, wind turbine is abnormal, and wind turbine blade is abnormal.

[0177] In this way, this application constructs an acoustic sensor network using multiple spherical microphone arrays, enabling the constructed acoustic sensor network to possess high directivity characteristics in the high-order spherical harmonic domain. Furthermore, the multi-point layout forms a larger aperture acoustic sensing plane, which not only improves the spatial sampling rate but also enhances the spatial resolution, thereby enabling more accurate capture of acoustic signals from wind turbine units and blades. In addition, the use of a beamformer based on a high-order Ambisonics (HOA) sensor network to acquire target signals significantly improves signal enhancement and anti-interference capabilities. Moreover, combining convolutional neural networks for feature extraction and classification of the spatially enhanced signals significantly improves the accuracy and generalization ability of turbine and blade fault detection.

[0178] Based on the same inventive concept, this application also provides a monitoring device corresponding to the monitoring method. Since the principle of the device in this application is similar to the monitoring method described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0179] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of a monitoring device for wind power equipment provided in an embodiment of this application. Figure 3 As shown, the monitoring device 300 includes:

[0180] The acquisition module 310 is used to acquire the first raw acoustic signal of the wind power equipment monitored by multiple spherical microphone arrays;

[0181] Processing module 320 is used to perform time-frequency transformation processing on each first original acoustic signal to obtain the spherical harmonic domain signal of each spherical microphone array;

[0182] Module 330 is used to construct an acoustic sensor beamformer based on the spherical harmonic domain signals of all spherical microphone arrays.

[0183] The partitioning module 340 is used to perform spatial grid partitioning on the area where the wind power equipment is located using the acoustic sensor network beamformer to obtain a spatial spectrum.

[0184] The locking module 350 is used for beam locking processing based on the target azimuth of the wind power equipment determined according to the spatial spectrum.

[0185] Enhancement module 360 ​​is used to enhance the locked beam using the acoustic sensor network beamformer to obtain the target acoustic signal;

[0186] Analysis module 370 is used to analyze the target acoustic signal according to a pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment.

[0187] Optionally, when acquiring the first raw acoustic signal of a wind power device using a plurality of spherical microphone arrays, the acquisition module 310 is used to:

[0188] For each spherical microphone array, acquire the second raw acoustic signal of each unidirectional microphone on the spherical microphone array that monitors the wind power equipment;

[0189] All the second original sound signals using the spherical microphone array are combined to obtain the first original sound signal of the spherical microphone array.

[0190] Optionally, when the processing module 320 performs time-frequency transformation processing on each first original acoustic signal to obtain the spherical harmonic domain signal of each spherical microphone array, the processing module 320 is used to:

[0191] For each spherical microphone array, a short-time Fourier transform is performed on the first original acoustic signal of the spherical microphone array to obtain the time-frequency spectrum of the spherical microphone array.

[0192] The time-frequency spectrum of the spherical microphone array is processed by spherical harmonic domain transformation to obtain the spherical harmonic domain signal of the spherical microphone array.

[0193] Optionally, when constructing an acoustic sensor network beamformer based on the spherical harmonic domain signals of all spherical microphone arrays, the construction module 330 is used to:

[0194] Obtain the relative distance and azimuth angle between each spherical microphone array and the wind power equipment;

[0195] Based on the relative distances and azimuth angles between all the spherical microphone arrays and the wind power equipment, a steering vector matrix for the acoustic sensor network is constructed.

[0196] Based on the spherical harmonic domain signals of all spherical microphone arrays, construct the spherical harmonic domain signals of the acoustic sensor network;

[0197] By combining the steering vector matrix of the acoustic sensor network and the spherical harmonic domain signal of the acoustic sensor network, an acoustic sensor network beamformer is obtained.

[0198] Optionally, when the analysis module 370 analyzes the target acoustic signal according to a pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment, the analysis module 370 is used to:

[0199] The target acoustic signal is processed by extracting voiceprint features through convolutional layers and activation functions in the anomaly detection model to obtain multidimensional voiceprint features.

[0200] The multi-dimensional voiceprint features are analyzed by a multi-layer perceptron classifier in the anomaly detection model to generate monitoring results for wind power equipment.

[0201] Optionally, an acoustic sensor network can be constructed by arranging multiple spherical microphone arrays in a preset manner, and the acoustic sensor network can be deployed below the wind power equipment.

[0202] Optionally, the monitoring result of the wind power equipment can be one of three results: wind power equipment is normal, wind turbine is abnormal, or wind turbine blade is abnormal.

[0203] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0204] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, they can perform the operations described above. Figure 1 The steps in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0205] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0206] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0207] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0208] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0209] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0210] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0211] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for monitoring wind power equipment, characterized in that, The monitoring method includes: The first raw acoustic signal of a wind power device is acquired by a high-order surround sound sensor network with multiple spherical microphone arrays monitoring the wind power device. The acoustic sensor network is deployed below the wind power device, and the multiple spherical microphone arrays in the acoustic sensor network are arranged in a preset manner, with multiple unidirectional microphones evenly distributed in each spherical microphone array. Each of the first original acoustic signals is subjected to time-frequency transformation to obtain the spherical harmonic domain signal of each spherical microphone array; An acoustic sensor network beamformer is constructed based on the spherical harmonic domain signals of all spherical microphone arrays and the steering vector matrix of the acoustic sensor network; the steering vector matrix of the acoustic sensor network is determined based on the relative distance and azimuth angle between each spherical microphone array and the wind power equipment. A beamformer for an acoustic sensor network is used to perform spatial grid division of the area where the wind power equipment is located, resulting in a spatial spectrum that displays the acoustic energy distribution of the area. Specifically, this involves: dividing the area where the wind power equipment is located into spatial units to form a spatial grid with multiple monitoring points; processing the acoustic signals within each grid using a beamformer to generate the acoustic signal intensity or spectrum information of each grid; and combining the acoustic signal intensity or spectrum information of all the generated grids to form the spatial spectrum of the area where the wind power equipment is located. The target location of the wind power equipment is determined based on the peak intensity of the acoustic energy in the spatial spectrum, and beam locking is performed. The target acoustic signal is obtained by enhancing the locked beam using a beamformer in an acoustic sensor network. The target acoustic signal is analyzed based on a pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment.

2. The monitoring method according to claim 1, characterized in that, The acquisition of the first raw acoustic signal from multiple spherical microphone arrays monitoring wind power equipment in the high-order surround sound sensor network includes: For each spherical microphone array, acquire the second raw acoustic signal of each unidirectional microphone on the spherical microphone array that monitors the wind power equipment; All the second original sound signals using the spherical microphone array are combined to obtain the first original sound signal of the spherical microphone array.

3. The monitoring method according to claim 1, characterized in that, The step of performing time-frequency transformation processing on each first original acoustic signal to obtain the spherical harmonic domain signal of each spherical microphone array includes: For each spherical microphone array, a short-time Fourier transform is performed on the first original acoustic signal of the spherical microphone array to obtain the time-frequency spectrum of the spherical microphone array. The time-frequency spectrum of the spherical microphone array is processed by spherical harmonic domain transformation to obtain the spherical harmonic domain signal of the spherical microphone array.

4. The monitoring method according to claim 1, characterized in that, The construction of the acoustic sensor network beamformer based on the spherical harmonic domain signals of all spherical microphone arrays and the steering vector matrix of the acoustic sensor network includes: Obtain the relative distance and azimuth angle between each spherical microphone array and the wind power equipment; Based on the relative distances and azimuth angles between all the spherical microphone arrays and the wind power equipment, a steering vector matrix for the acoustic sensor network is constructed. Based on the spherical harmonic domain signals of all spherical microphone arrays, construct the spherical harmonic domain signals of the acoustic sensor network; By combining the steering vector matrix of the acoustic sensor network and the spherical harmonic domain signal of the acoustic sensor network, an acoustic sensor network beamformer is obtained.

5. The monitoring method according to claim 1, characterized in that, The step of analyzing the target acoustic signal according to a pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment includes: The target acoustic signal is processed by extracting voiceprint features through convolutional layers and activation functions in the anomaly detection model to obtain multidimensional voiceprint features. The multi-dimensional voiceprint features are analyzed by a multi-layer perceptron classifier in the anomaly detection model to generate monitoring results for wind power equipment.

6. The monitoring method according to claim 1, characterized in that, The monitoring results of the wind power equipment are one of three results: wind power equipment is normal, wind turbine is abnormal, and wind turbine blade is abnormal.

7. A monitoring device for wind power equipment, characterized in that, The monitoring device includes: The acquisition module is used to acquire the first original acoustic signal of the wind power equipment monitored by multiple spherical microphone arrays in the high-order surround sound sensor network; wherein, the acoustic sensor network is deployed below the wind power equipment, and the multiple spherical microphone arrays in the acoustic sensor network are arranged in a preset manner, with multiple unidirectional microphones evenly distributed in each spherical microphone array. The processing module is used to perform time-frequency transformation processing on each first original acoustic signal to obtain the spherical harmonic domain signal of each spherical microphone array; A construction module is used to construct an acoustic sensor network beamformer based on the spherical harmonic domain signals of all spherical microphone arrays and the steering vector matrix of the acoustic sensor network; the steering vector matrix of the acoustic sensor network is determined based on the relative distance and azimuth angle between each spherical microphone array and the wind power equipment. The segmentation module is used to perform spatial grid segmentation of the area where the wind power equipment is located using a beamformer of an acoustic sensor network, thereby obtaining a spatial spectrum that displays the acoustic energy distribution of the area where the wind power equipment is located. Specifically, it includes: dividing the area where the wind power equipment is located into spatial units to form a spatial grid including multiple monitoring points; processing the acoustic signals in each grid using a beamformer to generate the acoustic signal intensity or spectrum information of each grid; and combining the acoustic signal intensity or spectrum information of all the generated grids to form the spatial spectrum of the area where the wind power equipment is located. The locking module is used to perform beam locking processing on the target orientation of the wind power equipment determined by the intensity peak of the acoustic energy in the spatial spectrum. The enhancement module is used to enhance the locked beam using the acoustic sensor network beamformer to obtain the target acoustic signal; The analysis module is used to analyze the target acoustic signal according to the pre-designed anomaly detection model to obtain the monitoring results of the wind power equipment.

8. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus, and the machine-readable instructions are executed by the processor to perform the steps of the monitoring method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the monitoring method as described in any one of claims 1 to 6.

Images