An online monitoring method for offshore wind turbine blades based on voiceprint recognition

Abstract

The present invention discloses an online monitoring method for offshore wind turbine blades based on voiceprint recognition, which belongs to the field of acoustic signal analysis of wind turbine blades. The present invention obtains wind turbine voiceprint data, weather condition data and wind turbine power generation data, constructs a signal processing model, denoises, amplifies and extracts features of the collected voiceprint signal through a signal amplifier, a filter and a feature extraction unit, constructs a deep learning model, imports voiceprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks, constructs a crack assessment model, imports crack sound source position, crack depth, crack length and estimated damage score into the crack assessment model to evaluate the degree of wind turbine blade cracks under future weather conditions, classifies wind turbine cracks according to the crack damage score and gives an alarm, and realizes early warning of future wind turbine blade cracks by combining the voiceprint signal, weather condition and power generation of offshore wind turbine blades.

Description

Technical Field

[0001] The present invention belongs to the field of wind turbine blade acoustic signal analysis, and specifically relates to an online monitoring method for offshore wind turbine blades based on soundprint recognition. Background Art

[0002] With the rapid development of the wind power industry, offshore wind power has gradually become an important part of future clean energy. As an important key component of wind turbines, wind turbine blades are affected by various environmental factors in long-term operation, affecting power generation efficiency and unit safety.

[0003] Traditional wind turbine blade monitoring methods mainly include visual inspection, infrared detection, and acoustic emission technology. However, these methods have limitations. Visual inspection relies on manpower, is inefficient, and cannot be monitored in real time. Infrared detection and acoustic emission technology require additional sensors to be installed, and have limited effects on blade damage monitoring during service. Based on the shortcomings of these existing technologies, it is particularly important to develop a method that can accurately monitor blade status in real time.

[0004] In recent years, voiceprint recognition technology has been widely used in wind turbine blade fault monitoring. By collecting sound signals when the wind turbine is running, the condition of the blade can be judged. The voiceprint-based monitoring system uses deep learning and signal processing algorithms to extract and analyze the characteristics of the soundprint of the wind turbine blade, combined with power generation and future weather conditions, to achieve early warning of future cracks in the wind turbine blade. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention proposes an online monitoring system for offshore wind turbine blades based on voiceprint recognition.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An online monitoring method for offshore wind turbine blades based on voiceprint recognition comprises the following specific steps:

[0008] Obtain wind turbine soundprint data, weather condition data, and wind turbine power generation data;

[0009] Build a signal processing model to denoise, amplify and extract features of the collected voiceprint signal through signal amplifiers, filters and feature extraction units;

[0010] Build a deep learning model and import the soundprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks;

[0011] Construct a crack assessment model, import the crack source location, crack depth, crack length and estimated damage score into the crack assessment model to assess the crack degree of wind turbine blades under future weather conditions;

[0012] Fan cracks are graded and alarms are issued based on the crack damage score.

[0013] Preferably, the obtaining of wind turbine soundprint data, weather condition data and wind turbine power generation data comprises the following specific steps:

[0014] S11. Acquire the soundprint data of the fan blades in operation through a soundprint collection terminal arranged at a monitoring position, and acquire the historical data of the fan cracks;

[0015] S12, obtaining offshore weather data through monitoring stations, and obtaining wind turbine power generation data through power stations;

[0016] S13, constructing the historical data of wind turbine blade cracks, historical data of weather conditions and historical data of power generation into a training set.

[0017] Preferably, the construction of the signal processing model, denoising, amplifying and feature extracting the collected voiceprint signal through a signal amplifier, a filter and a feature extraction unit comprises the following specific steps:

[0018] S21, introducing the collected voiceprint signal into the signal amplifier to amplify the primary collected voiceprint signal, introducing the amplified signal into the filter to filter the noise, introducing the processed signal into the feature extraction unit to extract the feature, and extracting the frequency, amplitude, and frequency domain amplitude parameters of the voiceprint signal;

[0019] S22. Substitute the voiceprint signal parameters into the kurtosis value calculation formula to calculate the kurtosis value, where the kurtosis value calculation formula is: Among them, x i is the amplitude of the i-th instantaneous signal, x′ is the mean signal amplitude, and N is the total number of signals;

[0020] S23, substituting the voiceprint signal parameters into the frequency band energy calculation formula to calculate the frequency band energy, wherein the frequency band energy calculation formula is: Where M(f) is the frequency domain amplitude of the voiceprint signal, and [f1,f2] is the frequency range of the voiceprint signal;

[0021] S24. Substitute the voiceprint signal parameters into the high-frequency harmonic calculation formula to calculate the high-frequency harmonics, where the high-frequency harmonic calculation formula is: Among them, f b is the high frequency cut-off frequency, is the total energy in the high frequency region, ∑[M(f)] 2 is the total energy in the entire frequency range.

[0022] Preferably, the construction of a deep learning model, importing the soundprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks includes the following specific steps:

[0023] S31. Input the frequency and amplitude of the voiceprint signal into the CNN convolutional neural network to extract the voiceprint spatial features, output the voiceprint feature vector S, and substitute the voiceprint feature vector, weather data and wind turbine power generation into the time series input matrix, where the time series input matrix is: Among them, W j is the wind speed at the jth moment, D j is the wind direction at the jth moment, T j is the temperature at the jth moment, R j is the humidity at the jth moment, P j is the power generation at the jth moment, S j is the voiceprint feature vector at the jth moment, and j is any item from 1 to n;

[0024] S32, input the time series feature matrix into the LSTM long short-term memory network, and output the estimated damage score, where the estimated damage score calculation formula is: Y = f LSTM (X), where f LSTM Modeling function Y for the time of the LSTM network.

[0025] Preferably, the construction of the crack assessment model, importing the crack sound source position, crack depth, crack length and estimated damage score into the crack assessment model to assess the crack degree of the wind turbine blade under future weather conditions includes the following specific steps:

[0026] S41, calculating the crack sound source position by the time difference of the signal arrival received by the voiceprint collection terminal, wherein the calculation formula of the crack sound source position is: d=v×t, wherein v is the sound wave propagation speed, and t is the time difference of the voiceprint signal arrival;

[0027] S42, substituting the voiceprint signal parameters into the crack depth calculation formula to calculate the crack depth, wherein the crack depth calculation formula is: Wherein, v is the propagation speed of sound waves, t is the time difference of the arrival of the soundprint signal, Kr is the kurtosis value of the fan in normal state, Er is the frequency band energy of the fan in normal state, K is the current kurtosis value, and E is the current frequency band energy;

[0028] S43, substituting the voiceprint signal parameters into the crack length calculation formula to calculate the crack length, wherein the crack length calculation formula is: Wherein, A is the currently detected signal amplitude, A0 is the initial amplitude under normal condition of the fan, α is the attenuation coefficient of the sound wave in the material, C is the amplitude unit conversion factor, that is, the coefficient of converting the amplitude ratio into the sound pressure level, C=20, Kr is the kurtosis value under normal condition of the fan, K is the current kurtosis value, Hr is the high-frequency harmonic under normal condition of the fan, and H is the current high-frequency harmonic;

[0029] S44. Substitute the crack sound source location, crack depth, crack length and predicted damage score into the crack damage score formula to evaluate the crack degree of the wind turbine blade under future weather conditions, wherein the crack damage score formula is: Where Y is the estimated damage score, d is the crack source location, ξ is the crack source location weight, h is the crack depth, L is the crack length, and h m is the maximum crack depth that the fan blade can withstand, L m is the maximum crack length of the fan blade, λ is the degree of crack aggregation, and the calculation formula for the degree of crack aggregation is: Where ρ is the length of the wind turbine blade and u is the average distance between existing cracks.

[0030] Preferably, the fan cracks are graded and an alarm is issued according to the crack damage score, and an alarm signal is issued according to different grades when a blade crack is detected, and a remote alarm unit transmits the alarm information to a remote monitoring center through a network.

[0031] An online monitoring system for offshore wind turbine blades based on voiceprint recognition is implemented based on the above-mentioned online monitoring method for offshore wind turbine blades based on voiceprint recognition, and specifically includes:

[0032] A data acquisition module is used to acquire wind turbine soundprint data, weather condition data and wind turbine power generation data;

[0033] The signal processing module is used to denoise, amplify and extract features of the collected voiceprint signal;

[0034] A deep learning module, which is used to import soundprint signal parameters, weather data and wind turbine power generation into a deep learning model to evaluate future wind turbine blade cracks;

[0035] A crack assessment module is used to assess the extent of cracks in wind turbine blades under future weather conditions;

[0036] The alarm module is used to classify fan cracks through crack damage scoring and generate an alarm.

[0037] An electronic device comprises: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;

[0038] The processor executes the above-mentioned method for online monitoring of offshore wind turbine blades based on voiceprint recognition by calling the computer program stored in the memory.

[0039] A computer-readable storage medium, characterized in that it stores instructions, and when the instructions are executed on a computer, the computer executes the above-mentioned online monitoring method for offshore wind turbine blades based on voiceprint recognition.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention discloses an online monitoring method for offshore wind turbine blades based on voiceprint recognition, which belongs to the field of acoustic signal analysis of wind turbine blades. The present invention obtains wind turbine voiceprint data, weather condition data and wind turbine power generation data, constructs a signal processing model, denoises, amplifies and extracts features of the collected voiceprint signal through a signal amplifier, a filter and a feature extraction unit, constructs a deep learning model, imports voiceprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks, constructs a crack assessment model, imports crack sound source position, crack depth, crack length and estimated damage score into the crack assessment model to evaluate the degree of wind turbine blade cracks under future weather conditions, classifies wind turbine cracks according to the crack damage score and gives an alarm, and realizes early warning of future wind turbine blade cracks by combining the voiceprint signal, weather condition and power generation of offshore wind turbine blades. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the overall process of an online monitoring method for offshore wind turbine blades based on voiceprint recognition according to the present invention;

[0043] Figure 2 It is a schematic diagram of the overall framework of an online monitoring system for offshore wind turbine blades based on voiceprint recognition according to the present invention. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0045] Example 1

[0046] See also Figure 1 , an embodiment provided by the present invention: Figure 1 As shown, an online monitoring method for offshore wind turbine blades based on voiceprint recognition includes the following specific steps:

[0047] Obtain wind turbine soundprint data, weather condition data, and wind turbine power generation data;

[0048] Build a signal processing model to denoise, amplify and extract features of the collected voiceprint signal through signal amplifiers, filters and feature extraction units;

[0049] Build a deep learning model and import the soundprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks;

[0050] Construct a crack assessment model, import the crack source location, crack depth, crack length and estimated damage score into the crack assessment model to assess the crack degree of wind turbine blades under future weather conditions;

[0051] Fan cracks are graded and alarms are issued based on the crack damage score.

[0052] In this embodiment, it should be specifically explained that obtaining wind turbine soundprint data, weather condition data and wind turbine power generation data includes the following specific steps:

[0053] S11. Acquire the soundprint data of the fan blades in operation through a soundprint collection terminal arranged at a monitoring position, and acquire the historical data of the fan cracks;

[0054] S12. Obtain offshore weather data, including wind speed, wind direction, temperature and humidity, through monitoring stations, and obtain wind turbine power generation data through power stations;

[0055] S13, constructing the historical data of wind turbine blade cracks, historical data of weather conditions and historical data of power generation into a training set.

[0056] It should be specifically explained here that the voiceprint collection terminals are evenly arranged around the wind turbine tower base and in multiple locations in the cabin. Four voiceprint collection terminals are arranged around the wind turbine tower base, and each terminal is spaced at the same angle to ensure effective collection of voiceprint signals in all directions around the wind turbine tower base. Two voiceprint collection terminals are arranged in the cabin to capture the voiceprint signals generated during the operation of key components such as the transmission chain. Through multi-point uniform arrangement, the operating area of ​​the wind turbine is covered to the maximum extent, and the voiceprint data of the wind turbine blades under different operating conditions are obtained.

[0057] It should be specifically explained in this embodiment that constructing a signal processing model, denoising, amplifying and extracting features from the collected voiceprint signal through a signal amplifier, a filter and a feature extraction unit includes the following specific steps:

[0058] S21, introducing the collected voiceprint signal into the signal amplifier to amplify the primary collected voiceprint signal, introducing the amplified signal into the filter to filter the noise, introducing the processed signal into the feature extraction unit to extract the feature, and extracting the frequency, amplitude, and frequency domain amplitude parameters of the voiceprint signal;

[0059] S22. Substitute the voiceprint signal parameters into the kurtosis value calculation formula to calculate the kurtosis value, where the kurtosis value calculation formula is: Among them, x i is the amplitude of the i-th instantaneous signal, x′ is the mean signal amplitude, and N is the total number of signals;

[0060] S23, substituting the voiceprint signal parameters into the frequency band energy calculation formula to calculate the frequency band energy, wherein the frequency band energy calculation formula is: Where M(f) is the frequency domain amplitude of the voiceprint signal, and [f1,f2] is the frequency range of the voiceprint signal;

[0061] S24. Substitute the voiceprint signal parameters into the high-frequency harmonic calculation formula to calculate the high-frequency harmonics, where the high-frequency harmonic calculation formula is: Among them, f b is the high frequency cut-off frequency, is the total energy in the high frequency region, ∑[M(f)] 2 is the total energy in the entire frequency range.

[0062] It should be specifically explained here that, during signal processing, the filter adopts a finite impulse response (FIR) filter to perform frequency domain processing on the amplified voiceprint signal. The FIR filter adopts a high-order filtering design to filter out low-frequency environmental noise and random noise based on frequency domain processing.

[0063] In this embodiment, it should be specifically explained that building a deep learning model and importing the soundprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks includes the following specific steps:

[0064] S31. Input the frequency and amplitude of the voiceprint signal into the CNN convolutional neural network to extract the voiceprint spatial features, wherein the CNN convolution operation is: Among them, B t+β,f+ε G is the value representing the position at time step t+β and frequency f+ε in the frequency channel. t,f is the weight of the convolution kernel in the frequency channel, B' t+β,f+ε G' is the value of the position at time step t+β and frequency f+ε in the amplitude channel. t,f is the weight of the convolution kernel in the amplitude channel. After pooling, flattening and full connection layer, the voiceprint feature vector S is output. The voiceprint feature vector, weather data and wind turbine power generation are substituted into the time series input matrix. The time series input matrix is: Among them, W j is the wind speed at the jth moment, D j is the wind direction at the jth moment, T j is the temperature at the jth moment, R j is the humidity at the jth moment, Pj is the power generation at the jth moment, S j is the voiceprint feature vector at the jth moment, and j is any item from 1 to n;

[0065] S32, input the time series feature matrix into the LSTM long short-term memory network, and output the estimated damage score through the forget gate, input gate, update unit state and output gate, where the estimated damage score calculation formula is: Y = f LSTM (X), where f LSTM is the time modeling function of the LSTM network, and Y damage score is the impact of future external environment changes on the wind turbine.

[0066] In this embodiment, it should be specifically explained that constructing a crack assessment model, importing the crack sound source position, crack depth, crack length and estimated damage score into the crack assessment model to assess the crack degree of the wind turbine blade under future weather conditions includes the following specific steps:

[0067] S41, calculating the crack sound source position by the time difference of the signal arrival received by the voiceprint collection terminal, wherein the calculation formula of the crack sound source position is: d=v×t, wherein v is the sound wave propagation speed, and t is the time difference of the voiceprint signal arrival;

[0068] S42, substituting the voiceprint signal parameters into the crack depth calculation formula to calculate the crack depth, wherein the crack depth calculation formula is: Wherein, v is the propagation speed of sound waves, t is the time difference of the arrival of the soundprint signal, Kr is the kurtosis value of the fan in normal state, Er is the frequency band energy of the fan in normal state, K is the current kurtosis value, and E is the current frequency band energy;

[0069] S43, substituting the voiceprint signal parameters into the crack length calculation formula to calculate the crack length, wherein the crack length calculation formula is: Among them, A is the currently detected signal amplitude, A0 is the initial amplitude under normal condition of the fan, α is the attenuation coefficient of the sound wave in the material, C is the amplitude unit conversion factor, that is, the coefficient of converting the amplitude ratio into the sound pressure level, Kr is the kurtosis value under normal condition of the fan, K is the current kurtosis value, Hr is the high-frequency harmonic under normal condition of the fan, and H is the current high-frequency harmonic;

[0070] S44. Substitute the crack sound source location, crack depth, crack length and predicted damage score into the crack damage score formula to evaluate the crack degree of the wind turbine blade under future weather conditions, wherein the crack damage score formula is: Where Y is the estimated damage score, d is the crack source location, ξ is the crack source location weight, h is the crack depth, L is the crack length, and h m is the maximum crack depth that the fan blade can withstand, L mis the maximum crack length of the fan blade, λ is the degree of crack aggregation, and the calculation formula for the degree of crack aggregation is: Where ρ is the length of the wind turbine blade and u is the average distance between existing cracks.

[0071] In this embodiment, it should be specifically explained that the fan cracks are graded and alarmed according to the crack damage score, and when blade cracks are detected, alarm signals are sent out according to different levels, and the remote alarm unit transmits the alarm information to the remote monitoring center through the network.

[0072] It should be noted here that fan cracks are graded according to the crack damage score. The standard value selection method for grading is as follows: 1,000 sets of fan crack damage data are obtained, imported into the deep learning model, and the crack damage score is calculated. The crack damage score is imported into the fitting software. The grading standard is determined through expert evaluation and fitting results to divide the risk into three levels. At the first level warning, the monitoring system issues a "minor crack" warning. The alarm information includes the current damage score and the recommended inspection cycle. No shutdown is required, and the monitoring frequency is increased. At the second level warning, the monitoring system issues a "medium crack" warning. The alarm information includes the damage score and the crack development trend. The machine is immediately shut down for inspection and maintenance, and a detailed evaluation of the crack is performed. At the third level warning, the monitoring system issues a "serious crack" warning. The alarm information includes the damage score and the severity of the crack. The remote monitoring center immediately receives the alarm and transmits it to the on-site operator through the network, and controls the fan shutdown through the automation system for immediate repair.

[0073] Example 2

[0074] like Figure 2 As shown, an online monitoring system for offshore wind turbine blades based on voiceprint recognition is implemented based on the above-mentioned online monitoring method for offshore wind turbine blades based on voiceprint recognition, and specifically includes a data acquisition module, a signal processing module, a deep learning module, a crack assessment module and an alarm module. The data acquisition module is used to acquire wind turbine voiceprint data, weather condition data and wind turbine power generation data; the signal processing module is used to denoise, amplify and extract features of the collected voiceprint signals; the deep learning module is used to import voiceprint signal parameters, weather data and wind turbine power generation into a deep learning model to assess future wind turbine blade cracks; the crack assessment module is used to assess the degree of wind turbine blade cracks under future weather conditions; the alarm module is used to classify wind turbine cracks and alarm by crack damage scoring.

[0075] It should be specifically explained in this embodiment that the system also includes a storage unit for storing voiceprint signal characteristics and fault judgment results. The unit not only stores current monitoring data, but also retains historical operation data to support long-term trend analysis.

[0076] What needs to be specifically explained in this embodiment is that the system also includes a data transmission module, which supports multiple wireless communication methods, including 4G, 5G and satellite communication, to meet the communication needs of wind turbines in different offshore environments. The data transmission module transmits the collected voiceprint signals and analysis results to the remote server through the wireless communication network to realize remote storage and monitoring of data.

[0077] It should be specifically explained here that the data transmission module has a high-intensity encryption function and uses the AES256-bit encryption algorithm to encrypt the voiceprint data to ensure that it is not stolen or tampered with during data collection, transmission and storage. In the process of transmitting data from the wind turbine tower base to the remote server, a VPN tunnel is used for secure transmission to further enhance the security of the communication link and prevent third parties from eavesdropping and attacking the data. All data must be processed by the encryption module before transmission and can only be decrypted and read after the transmission is completed.

[0078] What needs to be specifically explained in this embodiment is that the system includes a remote diagnosis module for performing in-depth analysis of the voiceprint data. The remote diagnosis module can present the analysis results to the operation and maintenance personnel through a visual interface. The remote diagnosis module also includes an expert system for providing specific maintenance suggestions based on the degree of cracks. The expert system is based on many years of wind turbine operation and maintenance experience and big data analysis results.

[0079] It should be specifically explained in this embodiment that the system can be integrated with the management system of the wind farm, and data interaction with other monitoring and control systems can be achieved through the interface module to realize intelligent and integrated management of the wind farm.

[0080] It should be specifically noted in this embodiment that each module of the system adopts a modular design to facilitate system expansion and maintenance. The modular design enables each component to be independently upgraded or replaced, thereby enhancing the flexibility and maintainability of the system.

[0081] Example 3

[0082] This embodiment provides an electronic device, including: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;

[0083] The processor executes the above-mentioned method for online monitoring of offshore wind turbine blades based on voiceprint recognition by calling the computer program stored in the memory.

[0084] The electronic device may have relatively large differences due to different configurations or performances, and may include one or more processors (Central Processing Units, CPU) and one or more memories, wherein the memory stores at least one computer program, which is loaded and executed by the processor to implement a multimedia image quality evaluation method based on scene recognition provided by the above method embodiment. The electronic device may also include other components for implementing the functions of the device, for example, the electronic device may also have components such as a wired or wireless network interface and an input and output interface to input and output data. This embodiment will not be described in detail here.

[0085] Example 4

[0086] This embodiment provides a computer-readable storage medium having a rewritable computer program stored thereon;

[0087] When the computer program runs on a computer device, the computer device executes the above-mentioned method for online monitoring of offshore wind turbine blades based on voiceprint recognition.

[0088] For example, the computer readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, etc.

[0089] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0090] It should be understood that determining B based on A does not mean determining B only based on A. B can also be determined based on A and / or other information.

[0091] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or / and a wireless network. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0092] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0093] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0094] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only one way of division, and there may be other ways of division in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0095] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0096] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0097] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0098] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An online monitoring method for offshore wind turbine blades based on voiceprint recognition, characterized in that: It includes the following specific steps: Obtain wind turbine soundprint data, weather condition data, and wind turbine power generation data; Construct a signal processing model, and use the signal amplifier, filter and feature extraction unit to denoise, amplify and extract features of the collected voiceprint signal to extract the voiceprint signal parameters; Construct a deep learning model, import the soundprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks, including: input the frequency and amplitude of the soundprint signal into the CNN convolutional neural network to output the soundprint feature vector, substitute the soundprint feature vector, weather data and wind turbine power generation into the time series input matrix, input the time series feature matrix into the LSTM long short-term memory network, and output an estimated damage score, which is the impact of future external environmental changes on the wind turbine; Construct a crack assessment model, import the crack source location, crack depth, crack length and estimated damage score into the crack assessment model to assess the crack degree of wind turbine blades under future weather conditions; Fan cracks are graded and alarms are issued based on the crack damage score.

2. The method for online monitoring of offshore wind turbine blades based on voiceprint recognition according to claim 1, characterized in that: The acquisition of wind turbine soundprint data, weather condition data and wind turbine power generation data comprises the following specific steps: S11. Acquire the soundprint data of the fan blades in operation through a soundprint collection terminal arranged at a monitoring position, and acquire the historical data of the fan cracks; S12, obtaining offshore weather data through monitoring stations, and obtaining wind turbine power generation data through power stations; S13, constructing the historical data of wind turbine blade cracks, historical data of weather conditions, and historical data of power generation into a training set.

3. The method for online monitoring of offshore wind turbine blades based on voiceprint recognition according to claim 2, characterized in that: The construction of the signal processing model, denoising, amplifying and feature extracting the collected voiceprint signal through the signal amplifier, filter and feature extraction unit includes the following specific steps: S21, introducing the collected voiceprint signal into the signal amplifier to amplify the primary collected voiceprint signal, introducing the amplified signal into the filter to filter the noise, introducing the processed signal into the feature extraction unit to extract the feature, and extracting the frequency, amplitude, and frequency domain amplitude parameters of the voiceprint signal; S22. Substitute the voiceprint signal parameters into the kurtosis value calculation formula to calculate the kurtosis value, where the kurtosis value calculation formula is: Among them, x i is the amplitude of the i-th instantaneous signal, x′ is the mean signal amplitude, and N is the total number of signals; S23, substituting the voiceprint signal parameters into the frequency band energy calculation formula to calculate the frequency band energy, wherein the frequency band energy calculation formula is: Where M(f) is the frequency domain amplitude of the voiceprint signal, and [f1,f2] is the frequency range of the voiceprint signal; S24. Substitute the voiceprint signal parameters into the high-frequency harmonic calculation formula to calculate the high-frequency harmonics, where the high-frequency harmonic calculation formula is: Among them, f b is the high frequency cut-off frequency, is the total energy in the high frequency region, ∑[M(f)] 2 is the total energy in the entire frequency range.

4. The method for online monitoring of offshore wind turbine blades based on voiceprint recognition according to claim 3, characterized in that: The construction of the deep learning model and the importation of the soundprint signal parameters, weather data and wind turbine power generation into the deep learning model to evaluate future wind turbine blade cracks include the following specific steps: S31. Input the frequency and amplitude of the voiceprint signal into the CNN convolutional neural network to extract the voiceprint spatial features, output the voiceprint feature vector S, and substitute the voiceprint feature vector, weather data and wind turbine power generation into the time series input matrix, where the time series input matrix is: Among them, W j is the wind speed at the jth moment, D j is the wind direction at the jth moment, T j is the temperature at the jth moment, R j is the humidity at the jth moment, P j is the power generation at the jth moment, S j is the voiceprint feature vector at the jth moment, and j is any item from 1 to n; S32, input the time series feature matrix into the LSTM long short-term memory network, and output the estimated damage score, where the estimated damage score calculation formula is: Y = f LSTM (X), where f LSTM Modeling function Y for the time of the LSTM network.

5. The method for online monitoring of offshore wind turbine blades based on voiceprint recognition according to claim 4, characterized in that: The construction of the crack assessment model, importing the crack sound source position, crack depth, crack length and estimated damage score into the crack assessment model to assess the crack degree of the wind turbine blade under future weather conditions includes the following specific steps: S41, calculating the crack sound source position by the time difference of the signal arrival received by the voiceprint collection terminal, wherein the calculation formula of the crack sound source position is: d=v×t, wherein v is the sound wave propagation speed, and t is the time difference of the voiceprint signal arrival; S42, substituting the voiceprint signal parameters into the crack depth calculation formula to calculate the crack depth, wherein the crack depth calculation formula is: Wherein, v is the propagation speed of sound waves, t is the time difference of the arrival of the soundprint signal, Kr is the kurtosis value of the fan in normal state, Er is the frequency band energy of the fan in normal state, K is the current kurtosis value, and E is the current frequency band energy; S43, substituting the voiceprint signal parameters into the crack length calculation formula to calculate the crack length, wherein the crack length calculation formula is: Wherein, A is the currently detected signal amplitude, A0 is the initial amplitude under normal condition of the fan, α is the attenuation coefficient of the sound wave in the material, C is the amplitude unit conversion factor, Kr is the kurtosis value under normal condition of the fan, K is the current kurtosis value, Hr is the high-frequency harmonic under normal condition of the fan, and H is the current high-frequency harmonic; S44. Substitute the crack sound source location, crack depth, crack length and predicted damage score into the crack damage score formula to evaluate the crack degree of the wind turbine blade under future weather conditions, wherein the crack damage score formula is: Where Y is the estimated damage score, d is the crack source location, ξ is the crack source location weight, h is the crack depth, L is the crack length, and h m is the maximum crack depth that the fan blade can withstand, L m is the maximum crack length of the fan blade, λ is the degree of crack aggregation, and the calculation formula for the degree of crack aggregation is: Where ρ is the length of the wind turbine blade and u is the average distance between existing cracks.

6. The method for online monitoring of offshore wind turbine blades based on voiceprint recognition according to claim 5, characterized in that: The fan cracks are graded and alarmed according to the crack damage score. When blade cracks are detected, alarm signals are issued according to different levels. The remote alarm unit transmits the alarm information to the remote monitoring center through the network.

7. An online monitoring system for offshore wind turbine blades based on voiceprint recognition, which is implemented based on the online monitoring method for offshore wind turbine blades based on voiceprint recognition as claimed in any one of claims 1 to 6, characterized in that: Specifically include: A data acquisition module is used to acquire wind turbine soundprint data, weather condition data and wind turbine power generation data; The signal processing module is used to denoise, amplify and extract features of the collected voiceprint signal; A deep learning module, which is used to import soundprint signal parameters, weather data and wind turbine power generation into a deep learning model to evaluate future wind turbine blade cracks; A crack assessment module is used to assess the extent of cracks in wind turbine blades under future weather conditions; The alarm module is used to classify fan cracks through crack damage scoring and generate an alarm.

8. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; The processor executes the method for online monitoring of offshore wind turbine blades based on voiceprint recognition as described in any one of claims 1 to 6 by calling the computer program stored in the memory.

9. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer executes an online monitoring method for offshore wind turbine blades based on voiceprint recognition as described in any one of claims 1 to 6.

Images