Wind turbine generator blade vibration detection method based on optical fiber echo wall mode singular point microcavity model

Abstract

The invention discloses a wind turbine generator blade vibration detection method based on an optical fiber echo wall mode singular point micro-cavity model, and relates to the field of wind turbine generator detection, and the method comprises the steps: carrying out the external vibration response of a detection position of a wind turbine generator blade, obtaining a micro-cavity response signal, measuring the mode splitting value of a micro-cavity based on the micro-cavity response signal, and calculating the mode splitting value of the micro-cavity. Determining vibration data of the wind turbine generator blades; constructing a three-dimensional model of the wind turbine generator blade, extracting modal parameters from the vibration data by using a random space method, and obtaining a vibration damage result of the wind turbine generator blade based on the modal parameters and the three-dimensional model; and performing fusion processing on the vibration data according to the combination of the on-site environment data of the wind turbine generator and the dynamic correction, determining a state estimation value of the wind turbine generator blade, and generating a vibration detection result of the wind turbine generator blade based on the state estimation value. The method achieves the accurate measurement of the weak vibration in a complex environment, and plays a key role in the operation detection and fault prediction of a wind driven generator.

Description

Technical Field

[0001] The present invention relates to the field of wind turbine detection, and more specifically, to a method for detecting the vibration of wind turbine blades based on a fiber optic whispering gallery mode singularity microcavity model. Background Art

[0002] As an important renewable energy device, the health status of the blades of a wind turbine directly affects the performance and safety of the unit. During operation, the blades are subjected to complex aerodynamic forces, gravity, and inertia, resulting in vibrations. The blades become one of the key and most vulnerable components in a wind turbine. If abnormal vibrations are not detected and processed in a timely manner, it may lead to fatigue damage and fracture of the blades, triggering serious accidents.

[0003] Although existing vibration detection methods have been widely used in the wind power industry, they still face a series of limitations. For example, traditional vibration detection methods, such as sensors based on accelerometers, although reliable in some environments, are usually sensitive to environmental noise and may require complex signal processing to distinguish background noise from real vibration signals. In addition, traditional vibration detection methods may be difficult to meet the detection requirements for high-frequency and micro-vibrations in wind turbines.

[0004] No effective solution has been proposed for the problems in the related art. Summary of the Invention

[0005] In view of the problems in the related art, the present invention proposes a method for detecting the vibration of wind turbine blades based on a fiber optic whispering gallery mode singularity microcavity model to overcome the above-mentioned technical problems existing in the existing related technologies.

[0006] To this end, the specific technical solution adopted by the present invention is as follows:

[0007] A method for detecting the vibration of wind turbine blades based on a fiber optic whispering gallery mode singularity microcavity model, the method comprising:

[0008] Performing an external vibration response on the detection position of the wind turbine blade to obtain a microcavity response signal, measuring the mode splitting value of the microcavity based on the microcavity response signal, and determining the vibration data of the wind turbine blade;

[0009] Constructing a three-dimensional model of the wind turbine blade, extracting modal parameters from the vibration data using the random space method, and obtaining the vibration damage result of the wind turbine blade based on the modal parameters and the three-dimensional model;

[0010] Generating an environmental compensation mechanism according to the environmental data on the wind turbine site, performing fusion processing on the vibration data in combination with dynamic correction to determine the state estimation value of the wind turbine blade, and generating the vibration detection result of the wind turbine blade based on the state estimation value and the vibration damage result.

[0011] Preferably, an external vibration response is performed on the detection position of the wind turbine blade to obtain a microcavity response signal, and based on the microcavity response signal, the mode splitting value of the microcavity is measured, and the vibration data of the wind turbine blade is determined, including:

[0012] A scattering source and a laser are set, the spatial characteristics between the scattering source and the microcavity are adjusted based on the control system, the microcavity is adjusted to the singular point state, and the fiber optic guiding technology is used to couple the guiding state of the microcavity;

[0013] According to the test results, the microcavity in the singular point state is set at the detection position of the wind turbine blade, and the microcavity is used as a vibration sensing detector to perform an external response on the detection position and output a response signal;

[0014] The response signal is converted into an electrical signal, the resonance frequency of the microcavity is obtained, and a singular point microcavity model is constructed based on the resonance frequency. Combining the response process, the mode splitting value is determined to obtain a vibration measurement model;

[0015] The vibration measurement model is used to analyze the vibration value of the wind turbine blade, determine the vibration data of the wind turbine blade, and preliminarily verify the vibration state of the wind turbine blade according to the vibration data.

[0016] Preferably, setting a scattering source and a laser, adjusting the spatial characteristics between the scattering source and the microcavity based on the control system, adjusting the microcavity to the singular point state, and using the fiber optic guiding technology to couple the guiding state of the microcavity includes:

[0017] A microcavity is selected and the corresponding transmission spectrum of the microcavity is analyzed. According to the transmission spectrum, the resonance mode of the microcavity is found. At the same time, two groups of scattering sources are selected, and the initial spatial position between the microcavity and the scattering source is determined;

[0018] According to the initial spatial position, the control system is used to adjust the distance and phase angle between the scattering source and the edge of the microcavity. Based on the adjustment results, any one group of scattering sources is moved, and the movement is stopped after the resonance mode of the microcavity disappears;

[0019] Analyze the splitting phenomenon of the resonance mode of the microcavity after movement, and when the splitting phenomenon completely disappears, a microcavity in the singular point state is obtained, and the laser is turned on to emit a laser light source;

[0020] The fiber optic guiding technology is used to guide the laser light source to the microcavity in the singular point state, perform the light source-microcavity coupling process, and after the coupling is completed, the microcavity in the singular point state is encapsulated in the mounting piece for standby.

[0021] Preferably, converting the response signal into an electrical signal, obtaining the resonance frequency of the microcavity, and constructing a singular point microcavity model based on the resonance frequency, and combining the response process to determine the mode splitting value to obtain a vibration measurement model includes:

[0022] Analyze the resonant frequency of the singular point state microcavity according to the perturbation degree value of the scattering source, and at the same time judge the emission light intensity of the propagation mode of the singular point state microcavity based on the angular position of the scattering source, and combine it with the emission light intensity to generate a singular point microcavity model;

[0023] Amplify the response signal based on the signal amplifier and convert it into discrete digital data, and obtain the disturbance amount and angle during the vibration of the wind turbine blade according to the discrete digital data;

[0024] Generate an induction model after the singular point microcavity responds to vibration according to the disturbance amount and angle, and fuse the induction model with the singular point microcavity model to obtain a vibration response model;

[0025] Analyze the relationship between the splitting value caused by vibration and the disturbance amount based on the vibration measurement model and the singular point microcavity model, determine the mode splitting value, and generate a vibration measurement model according to the mode splitting value and the scale factor.

[0026] Preferably, construct a three-dimensional model of the wind turbine blade, extract modal parameters from the vibration data using the random space method, and obtain the vibration damage results of the wind turbine blade based on the modal parameters and the three-dimensional model, including:

[0027] Denoise and normalize the vibration data, set the processed vibration data as the output data sequence, and arrange it according to the preset requirements to generate a Hankel matrix;

[0028] Perform decomposition processing on the Hankel matrix, calculate the state space model corresponding to the vibration data according to the processing result, and use the state space model to output the extraction result of the test modal parameters of the vibration data;

[0029] Scan the wind turbine blade using a three-dimensional laser scanner to obtain the point cloud data on the surface of the wind turbine blade, and perform polygonization and surface fitting processing on the point cloud data;

[0030] Establish a three-dimensional model of the wind turbine based on the processed point cloud data, and obtain the finite analysis modal parameters of the wind turbine by combining the three-dimensional model of the wind turbine with the finite element analysis technology results;

[0031] Evaluate the loss probability of the wind turbine blade according to the test modal parameters and the finite analysis modal parameters, construct a damage prediction model in combination with the vibration signal, and obtain the vibration damage results of the wind turbine blade.

[0032] Preferably, perform decomposition processing on the Hankel matrix, calculate the state space model corresponding to the vibration data according to the processing result, and the extraction result of the test modal parameters of the vibration data output by using the state space model includes:

[0033] Perform singular value decomposition on the Hankel matrix based on singular value decomposition technology, analyze the rank information of the Hankel matrix according to the magnitude of the singular values during the processing, and obtain the noise information in the vibration data;

[0034] Compare the magnitude of the singular values with a relative threshold to obtain the order of the Hankel matrix, and combine the order result with the singular value decomposition processing result to estimate the state sequence of the vibration data;

[0035] Combine the state sequence with the noise information to construct a state space model, use the state space model to output the output vector of the vibration data, and take the output vector as the extraction result of the test modal parameters of the vibration data.

[0036] Preferably, evaluate the loss probability of the wind turbine blade according to the test modal parameters and the finite analysis modal parameters, combine the vibration signals to construct a damage prediction model, and obtain the vibration damage results of the wind turbine blade, including:

[0037] Take the test modal parameters and the finite analysis modal parameters as the test state modal vector and the finite element analysis modal vector respectively, and use the modal confidence criterion to quantify the matching degree between the test state modal vector and the finite element analysis modal vector;

[0038] Verify the accuracy of the combination of the three-dimensional model of the wind turbine and the finite element analysis technology according to the matching degree, and evaluate the loss probability of the wind turbine blade based on the accuracy result and the matching degree;

[0039] Obtain the corresponding vibration signals when various damages occur to the wind turbine blade, extract the characteristic states based on the vibration signals, and combine the characteristic states with the optimizer to construct a damage prediction model;

[0040] When the loss probability of the wind turbine blade reaches the threshold, use the damage prediction model to output the probability distribution result of the damage type, and select the category with the largest probability result as the vibration damage category result of the wind turbine blade.

[0041] Preferably, the expression of the modal confidence criterion is:

[0042]

[0043] In the formula, MAC ij represents the modal confidence criterion between the test state modal vector i and the finite element analysis modal vector j, represents the modal vector in the test state, represents the modal vector in the finite element analysis state, represents the transpose of the modal vector in the test state, represents the transpose of the modal vector in the finite element analysis state.

[0044] Preferably, vibration signals corresponding to various damages of the wind turbine blade are obtained, and characteristic states are extracted based on the vibration signals. Constructing a damage prediction model by combining the characteristic states with an optimizer includes:

[0045] Simulate the state information of the wind turbine blade in various damaged conditions, collect vibration signals corresponding to various damages according to the state information, convert the vibration signals into two-dimensional time-frequency diagrams using the short-time Fourier transform, and mark the damage types for the two-dimensional time-frequency diagrams at the same time;

[0046] Use a deep learning framework library to construct a basic model including a convolutional layer, a pooling layer and a fully connected layer, extract local features of the two-dimensional time-frequency diagram as convolutional layer features, and output classification probabilities using an activation function at the same time;

[0047] Define a cross-loss function, use the cross-loss function to measure the difference between the predicted classification probability and the true label, and select an optimizer to update the basic model parameters according to the difference result to obtain the final damage prediction model.

[0048] Preferably, an environmental compensation mechanism is generated according to the environmental data at the wind turbine site, and combined with dynamic correction to perform fusion processing on the vibration data to determine the state estimation value of the wind turbine blade. Generating a vibration detection result of the wind turbine blade based on the state estimation value and the vibration damage result includes:

[0049] Generate an initial state covariance matrix according to the static vibration displacement and velocity estimation values of the wind turbine blade, and predict the subsequent predicted state of the wind turbine blade based on the initial state covariance matrix;

[0050] Obtain the vibration displacement and vibration velocity of the wind turbine blade according to the vibration data, set a state vector, and use the response period of the external vibration response as the acquisition time period to obtain the environmental data at the wind turbine site during the acquisition time period;

[0051] Combine the state vector with the environmental data to determine an observation vector, and use the observation vector to correct the subsequent predicted state to obtain an updated state, and obtain the state estimation value of the wind turbine blade based on the updated state;

[0052] Fuse the state estimation value with the vibration damage result to generate a vibration detection result of the wind turbine blade, and judge the fault state of the vibration. If the state is normal, it runs normally. If the state is abnormal, a reminder is sent to the user side to adjust the wind turbine blade.

[0053] The beneficial effects of the present invention are:

[0054] 1. The present invention introduces a fiber-optic whispering gallery mode exceptional point microcavity model that can capture optical signals within a very narrow frequency range, making it a highly sensitive vibration detection tool. As a result, it can detect tiny vibrations far below the threshold of traditional sensors. When the exceptional point microcavity model is subjected to small vibrations, a large splitting value can be obtained, greatly improving the sensitivity when measuring small vibrations, thus enabling accurate measurement of weak vibrations, broadening the measurement range of the vibration values to be measured, and playing a key role in the operation detection and fault prediction of wind turbines.

[0055] 2. A method for detecting the vibration of wind turbine blades based on a fiber-optic whispering gallery mode exceptional point microcavity model proposed by the present invention can detect and analyze the tiny vibrations of wind turbine blades with high precision and high sensitivity. At the same time, it has the characteristics of anti-interference ability and adaptability to harsh environments, and combines advanced optical and signal analysis technologies to achieve high-precision and real-time detection of the vibration of wind turbine blades, providing guarantee for the safe and stable operation of wind turbines.

[0056] 3. The present invention introduces a fiber-optic whispering gallery mode exceptional point microcavity model that can capture optical signals within a very narrow frequency range, making it a highly sensitive vibration detection tool. And by using the stochastic subspace method to process the vibration data for damage identification and combining the Kalman filter algorithm to remove the interference of environmental factors on the vibration data, it can achieve accurate measurement of weak vibrations in complex environments and play a key role in the operation detection and fault prediction of wind turbines. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0058] Figure 1 is a flowchart of a method for detecting the vibration of wind turbine blades based on a fiber-optic whispering gallery mode exceptional point microcavity model according to an embodiment of the present invention;

[0059] Figure 2 is a working flowchart of a method for detecting the vibration of wind turbine blades based on a fiber-optic whispering gallery mode exceptional point microcavity model according to an embodiment of the present invention;

[0060] Figure 3 is a schematic diagram of the relative positions of a whispering gallery micro-ring cavity, a nano-probe, and a tiny perturbation in a method for detecting the vibration of wind turbine blades based on a fiber-optic whispering gallery mode exceptional point microcavity model according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0061] To further illustrate each embodiment, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be combined with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, those of ordinary skill in the art should be able to understand other possible implementation manners and the advantages of the present invention.

[0062] According to an embodiment of the present invention, a method for detecting the vibration of a wind turbine blade based on a fiber optic whispering gallery mode singularity microcavity model is provided.

[0063] Now, the present invention will be further described in conjunction with the accompanying drawings and specific implementation manners. As Figure 1 shown in Figure 2 the method for detecting the vibration of a wind turbine blade based on a fiber optic whispering gallery mode singularity microcavity model according to an embodiment of the present invention includes:

[0064] Step S1, perform an external vibration response on the detection position of the wind turbine blade to obtain a microcavity response signal, measure the mode splitting value of the microcavity based on the microcavity response signal, and determine the vibration data of the wind turbine blade.

[0065] In one embodiment, when performing an external vibration response on the detection position of the wind turbine blade to obtain a microcavity response signal, measuring the mode splitting value of the microcavity based on the microcavity response signal, and determining the vibration data of the wind turbine blade, a scattering source and a laser can be set, the spatial characteristics between the scattering source and the microcavity can be adjusted based on a control system, the microcavity can be adjusted to the singularity state, and the fiber optic guiding technology can be used to couple the guiding state of the microcavity; according to the test results, the microcavity in the singularity state is set at the detection position of the wind turbine blade, the microcavity is used as a vibration sensing detector to perform an external response on the detection position, and a response signal is output; the response signal is converted into an electrical signal, the resonant frequency of the microcavity is obtained, and a singularity microcavity model is constructed based on the resonant frequency. The mode splitting value is determined by combining the response process to obtain a vibration measurement model; the vibration value of the wind turbine blade is analyzed using the vibration measurement model to determine the vibration data of the wind turbine blade, and the vibration state of the wind turbine blade is preliminarily verified based on the vibration data.

[0066] In one embodiment, when setting the scattering source and the laser, adjusting the spatial characteristics between the scattering source and the microcavity based on the control system, adjusting the microcavity to the singular point state, and using the optical fiber guiding technology to couple the guiding state of the microcavity, the microcavity can be selected and the corresponding transmission spectrum of the microcavity can be analyzed. The resonance mode of the microcavity can be found according to the transmission spectrum. At the same time, two sets of scattering sources are selected, and the initial spatial position between the microcavity and the scattering source is determined; according to the initial spatial position, the control system is used to adjust the distance and phase angle between the scattering source and the edge of the microcavity, and any one set of scattering sources is moved based on the adjustment result, and the movement is stopped after the resonance mode of the microcavity disappears; analyze the splitting phenomenon of the resonance mode of the microcavity after movement, and when the splitting phenomenon completely disappears, obtain the microcavity in the singular point state, turn on the laser to emit a laser light source; use the optical fiber guiding technology to guide the laser light source to the microcavity in the singular point state, perform the light source-microcavity coupling process, and after the coupling is completed, encapsulate the microcavity in the singular point state in the mounting piece for standby.

[0067] In one embodiment, when converting the response signal into an electrical signal, obtaining the resonant frequency of the microcavity, and constructing a singular point microcavity model based on the resonant frequency, and determining the mode splitting value in combination with the response process to obtain the vibration measurement model, the resonant frequency of the microcavity in the singular point state can be analyzed according to the perturbation degree value of the scattering source. At the same time, based on the angular position of the scattering source, judge the emission light intensity of the propagation mode of the microcavity in the singular point state, and generate a singular point microcavity model in combination with the emission light intensity; amplify the response signal based on the signal amplifier and convert it into discrete digital data, and obtain the perturbation amount and angle during the vibration of the wind turbine blade according to the discrete digital data; generate an induction model after the singular point microcavity responds to vibration according to the perturbation amount and angle, and fuse the induction model with the singular point microcavity model to obtain a vibration response model; analyze the relationship between the splitting value caused by vibration and the perturbation amount based on the vibration measurement model and the singular point microcavity model, determine the mode splitting value, and generate a vibration measurement model according to the mode splitting value and the scale factor.

[0068] It should be noted that the fiber optic whispering gallery mode singular point microcavity model mainly includes a laser, scattering particles (scattering source), a fiber optic whispering gallery mode microcavity, a fiber optic taper coupling device (fiber optic guiding technology), a photodetector, an oscilloscope, a three-dimensional laser scanner, and a computer.

[0069] At the same time, two finely processed silica nanoprobes are used as Rayleigh scattering particles for the silica nanoprobes, with sizes between 30 and 50 nanometers. Using a high-precision nano-manipulation platform, the distance between the nanoprobes and the edge of the microring cavity is adjusted, and the phase angle is determined. By monitoring the transmission spectrum (transmission spectrum) of the microring cavity, a clear resonance mode is found when there are no nanoprobes. After introducing the nanoprobes, the initial distance between the nanoprobes and the edge of the microring cavity is fixed, and one of the probes is precisely moved. When the splitting phenomenon of this resonance mode on the transmission spectrum disappears, it indicates that the microring cavity system has been adjusted to the singular point state, and then the movement is stopped.

[0070] The laser is a distributed feedback laser (DFB), which has strong tunability and can adjust the output wavelength through current and temperature control. The output power is 10 mW and is used to provide a stable light source to excite the microcavity.

[0071] The fiber optic whispering gallery mode microcavity is a special optical resonator with dimensions in the micrometer or sub-micrometer range. Here, it is made of high-purity quartz material, ring-shaped, with an outer diameter of 350 micrometers and an inner diameter of 200 micrometers, fabricated by femtosecond laser etching technology with an accuracy of ±1 micrometer, and a quality factor (Q value) of 10 6 , to ensure high sensitivity and low loss, and is used to receive vibrations and generate optical responses. By continuously total internally reflecting light within the microcavity interface to form a stable propagation mode that meets certain conditions, the resonant optical field is confined within the cavity. The optical whispering gallery mode microring cavity has the remarkable characteristics of a high quality factor, a small mode volume, and convenient integration. When the microcavity is disturbed by external vibrations, it will cause changes in physical properties such as the geometric shape and refractive index of the microcavity, thereby leading to a change in the resonant frequency of the microcavity. This sensor is caused by the vibration of the wind turbine blade to generate a vibration disturbance in the fiber optic whispering gallery mode microcavity. By measuring the mode splitting value of the microcavity and calculating the vibration value of the blade based on a specific corresponding relationship.

[0072] The fiber optic taper coupling device effectively couples the laser into the microcavity and collects the output signal of the microcavity. The fiber optic taper is formed by processing ordinary optical fiber using the heating and tapering technique. After heating and softening the optical fiber and stretching it, the diameter of the optical fiber gradually becomes thinner. The standard single-mode optical fiber is placed on a three-dimensional microfabrication platform, and a waveguide with a fixed length is written in the fiber core using femtosecond laser. Then, the optical fiber with the waveguide is placed in a fusion splicer for discharging to form a spindle-shaped air cavity; at the same time, a femtosecond laser is used to scribe a mark on the upper surface of the air cavity on the laser microfabrication platform, and the optical fiber is manually broken to obtain an open air cavity; then, the microring is pushed into the air cavity with the help of a fusion splicer and a tapered optical fiber with a small amount of UV glue; finally, the position where the microring contacts the air cavity wall is adjusted with the tapered optical fiber to meet the phase matching condition to excite the WGM (optical whispering gallery mode), and the device is stabilized by curing with UV light.

[0073] The photodetector includes a photodiode and a signal amplifier. The photodiode detects the optical signal output by the microcavity and converts the optical signal into an electrical signal, and the signal amplifier amplifies the weak electrical signal received from the photodiode.

[0074] The oscilloscope has high-speed sampling and analysis functions. It performs high-speed acquisition on the input electrical signal and converts the continuous electrical signal into discrete digital data. Through the analysis of these data, the oscilloscope can obtain the frequency information of the signal. Since the microcavity mode splitting will cause the frequency of the optical signal to change, and the frequency change will be reflected in the electrical signal.

[0075] The specific detection process is as follows: First, turn on the power supply of the laser to provide a stable current or voltage. The control system adjusts the power and stability of the laser to ensure that the laser emitter body generates a single-wavelength and stable laser light source. Place two silica nanoprobes (scattering sources) as Rayleigh scattering particles. Through the nano-manipulation device, precisely adjust the relative position and effective size (spatial characteristics) of the nanoprobes and the microcavity to adjust the microcavity system to the singularity state. Guide the light emitted by the laser into the whispering gallery mode micro-ring cavity through an optical fiber taper (optical fiber guiding technology), and use a fixing device to fix the microcavity sensor to the corresponding position of the blade for detection.

[0076] At the same time, the fixing device encapsulates the fiber optic microcavity sensor in a thin sheet made of the same or similar composite material as the blade. Pay attention to protecting the optical and electrical connection parts of the sensor during encapsulation, and use an adhesive, such as epoxy resin, to fix the encapsulation result to the wind turbine blade, which can enhance the stability of the sensor and also provide it with a certain degree of protection against harsh environments.

[0077] The installation positions are as follows: The installation position of the sensor at the blade root, specifically near the connection with the hub at the blade root, to detect the vibration of the connection part between the blade and the hub; The installation position of the sensor in the middle of the blade, specifically near 1 / 3 of the distance from the root, to detect the vibration of the middle part of the blade under stress; The installation position of the sensor at the blade tip, specifically at the flat position at the blade tip, to detect the vibration of the blade tip under high-speed rotation.

[0078] Then, use the microcavity in the singularity state as a system vibration sensor to respond to external vibrations. The response signal of the microcavity is transmitted out through the optical fiber taper coupling device. The optical fiber taper transmits the light output by the microcavity to the photodetector. The photodiode converts the optical signal into an electrical signal, and the signal amplifier amplifies the weak electrical signal for subsequent processing. The oscilloscope receives the signal, analyzes and obtains the frequency information of the signal, and calculates the vibration situation after determining the splitting value according to the measurement principle.

[0079] Such as Figure 3As shown, in actual operation, when establishing the sensing model of the exceptional point microcavity, the resonant frequency of the exceptional point microcavity after introducing scattering particles is: f′ = f + ε1 + ε2, where f represents the resonant frequency of the microcavity without introducing the exceptional point, and ε1 and ε2 respectively represent the perturbation amounts caused by the two scattering particles. The Hamiltonian operator is used as the model (exceptional point microcavity model) to describe the exceptional point microcavity:

[0080]

[0081] The exceptional point microcavity system is the system when the optical whispering gallery mode microresonator system is tuned to the exceptional point state, and the exceptional point microcavity is the microcavity that reaches the exceptional point state.

[0082] Among them, a s and a n respectively represent the reflected light intensities of the microcavity's inherent clockwise-to-counterclockwise and counterclockwise-to-clockwise propagation modes. Their values are approximately equal, and there is:

[0083]

[0084] Among them, e represents the natural constant, i represents the imaginary unit, and its value is The imaginary exponential form in the formula is a way to describe physical quantities with periodic, oscillatory and other characteristics in complex form. θ1 represents the angular position of the first scattering particle, that is, the first nanosensor, and θ2 represents the angular position of the second scattering particle, that is, the second nanosensor;

[0085] Establish the sensing model after the exceptional point microcavity responds to vibration: When the blade vibration is transmitted to the microcavity, calculate the first Hamiltonian operator (vibration measurement model) of the exceptional point microcavity after being perturbed by vibration:

[0086]

[0087] Among them, k represents the perturbation amount caused by vibration, θ represents the azimuth angle of the perturbation amount, and then the sensing model after the exceptional point microcavity responds to vibration is obtained: H = H0 + H1;

[0088] Determine the vibration measurement equation: When the vibration to be measured is extremely weak and satisfies |a n | >> |k|, the vibration measurement equation is determined according to the above corresponding relationship expression as:

[0089]

[0090] Among them, y represents the vibration value to be measured, m represents the scale factor, which is only related to the structural characteristics of the microcavity itself and remains unchanged during vibration measurement. Δ represents the splitting value generated by the vibration perturbation of the singular point microcavity. When the microcavity is not vibrating, there is an initial resonant frequency. Under the action of vibration, the microcavity generates mode splitting, and two new resonant frequencies f1 and f2 appear, which are reflected on the oscilloscope. The differences between these two new resonant frequencies and the initial frequency are obtained, and the absolute values of these two difference values are the splitting values, that is

[0091] Δ = |f1 - f2|;

[0092] In actual operation, by reading the frequency data displayed on the oscilloscope and calculating the difference between the two new resonant frequencies, the splitting value Δ caused by vibration can be obtained. Substituting Δ into the vibration measurement equation, the vibration value y of the blade can be calculated. According to the preset vibration threshold, it can be judged whether the blade vibration is normal.

[0093] Step S2, construct a three-dimensional model of the wind turbine blade, extract modal parameters from the vibration data using the random space method, and obtain the vibration damage knot of the wind turbine blade based on the modal parameters and the three-dimensional model.

[0094] In one embodiment, during the process of constructing a three-dimensional model of the wind turbine blade, extracting modal parameters from the vibration data using the random space method, and obtaining the vibration damage result of the wind turbine blade based on the modal parameters and the three-dimensional model, the vibration data can be denoised and normalized, the processed vibration data is set as the output data sequence, and a Hankel matrix is generated according to the preset requirements; the Hankel matrix is decomposed, and the state space model corresponding to the vibration data is calculated according to the processing result, and the test modal parameter extraction result of the vibration data is output using the state space model; the wind turbine blade is scanned using a three-dimensional laser scanner to obtain the point cloud data on the surface of the wind turbine blade, and the point cloud data is polygonized and surface-fitted; a three-dimensional model of the wind turbine is established based on the processed point cloud data, and the finite analysis modal parameters of the wind turbine are obtained by combining the three-dimensional model of the wind turbine and the finite element analysis technology result; the loss probability of the wind turbine blade is evaluated according to the test modal parameters and the finite analysis modal parameters, and a damage prediction model is constructed in combination with the vibration signal to obtain the vibration damage result of the wind turbine blade.

[0095] In one embodiment, when performing decomposition processing on a Hankel matrix, calculating a state-space model corresponding to vibration data based on the processing result, and extracting the test modal parameter extraction result of the vibration data by using the state-space model, the singular value decomposition processing can be performed on the Hankel matrix based on the singular value decomposition technique, the rank information of the Hankel matrix can be analyzed according to the magnitudes of the singular values during the processing, and the noise information in the vibration data can be obtained; the magnitude of the singular value is compared with a relative threshold to obtain the order of the Hankel matrix, and the order result is combined with the singular value decomposition processing result to estimate the state sequence of the vibration data; the state sequence and the noise information are combined to construct a state-space model, the output vector of the vibration data is output by using the state-space model, and the output vector is used as the test modal parameter extraction result of the vibration data.

[0096] In one embodiment, when evaluating the loss probability of a wind turbine blade according to the test modal parameters and the finite analysis modal parameters, constructing a damage prediction model by combining vibration signals, and obtaining the vibration damage result of the wind turbine blade, the test modal parameters and the finite analysis modal parameters can be used as the test state modal vector and the finite element analysis modal vector respectively, and the modal confidence criterion can be used to quantify the matching degree between the test state modal vector and the finite element analysis modal vector; the accuracy of the combination of the three-dimensional model of the wind turbine and the finite element analysis technology is verified according to the matching degree, and the loss probability of the wind turbine blade is evaluated based on the accuracy result and the matching degree; the vibration signals corresponding to various damages of the wind turbine blade are obtained, the characteristic states are extracted based on the vibration signals, and the characteristic states are combined with an optimizer to construct a damage prediction model; when the loss probability of the wind turbine blade reaches the threshold, the probability distribution result of the damage type is output by using the damage prediction model, and the category with the largest probability result is selected as the vibration damage category result of the wind turbine blade.

[0097] In one embodiment, when obtaining the vibration signals corresponding to various damages of the wind turbine blade, extracting the characteristic states based on the vibration signals, and combining the characteristic states with an optimizer to construct a damage prediction model, the state information of the wind turbine blade in various damaged situations can be simulated, the vibration signals corresponding to various damages are collected according to the state information, and the vibration signals are converted into a two-dimensional time-frequency diagram by using the short-time Fourier transform. At the same time, the damage type is marked for the two-dimensional time-frequency diagram; a basic model including a convolutional layer, a pooling layer and a fully connected layer is constructed by using a deep learning framework library, the local features of the two-dimensional time-frequency diagram are extracted as the convolutional layer features, and the classification probability is output by using an activation function; a cross-loss function is defined, and the cross-loss function is used to measure the difference between the predicted classification probability and the true label. According to the difference result, an optimizer is selected to update the parameters of the basic model to obtain the final damage prediction model.

[0098] It should be noted that in the process of modal analysis and finite element verification, after obtaining the vibration data, the computer uses the stochastic subspace method to process the vibration data, so as to extract the modal parameters of the blade, including the vibration frequency and the modal vibration mode.

[0099] The stochastic subspace method is a modal parameter identification method based on the system state space model. It can effectively separate different modal components from complex vibration response data (vibration data). The specific steps are as follows:

[0100] (1) Data preprocessing: Denoise and normalize the collected data (vibration data). Specifically, filtering can be used for denoising and min-max normalization for normalization when applying.

[0101] (2) Construct the Hankel matrix: Let the collected output data sequence be y(1), y(2)…, y(n), and arrange it into the Hankel matrix Y. Use N to represent the length of the output data sequence, that is, the number of data points. Assume that the number of forward block rows is i Y , the number of backward block rows is j Y , the number of columns is k Y , and satisfy i Y + j Y + k Y - 1 is less than or equal to N, then the output Hankel matrix Y can be expressed as:

[0102]

[0103] (3) Singular value decomposition: Perform singular value decomposition on the output Hankel matrix Y:

[0104] Y = U∑V T ;

[0105] Among them, U represents the left singular matrix, Σ represents the diagonal singular value matrix, and the diagonal elements are singular values σ1≥σ2≥…≥σ n , and V represents the right singular matrix.

[0106] (4) Determine the system order: According to the magnitude of the singular values, select an appropriate threshold to truncate the singular values, set a relative threshold (1% of the largest singular value), set the singular values less than this threshold to zero, and the number of non-zero singular values retained is the system order n.

[0107] (5) State sequence estimation: Based on the singular value decomposition result, estimate the state sequence of the system as:

[0108]

[0109] Among them, U nDenote the matrix formed by the first n columns of U, Σ n is the diagonal matrix formed by the first n rows and n columns of Σ.

[0110] (6) System matrix identification: The state-space model of a linear time-invariant system is:

[0111]

[0112] where x(t) represents the state vector, y(t) represents the output vector, u(t) represents the input vector, w(t) and v(t) are the process noise and measurement noise respectively, A represents the state transition matrix, which can be obtained by solving the following least squares problem where X f represents the state sequence at the current moment, X P represents the state sequence at the next moment, ‖·‖ F represents the Frobenius norm, B represents the input matrix, A is known and solved according to the first equation of the state model, C represents the output matrix, and is solved using the known state sequence X and output sequence Y without considering the input u(t) and noise. D represents the direct transmission matrix, and C is known and solved according to the second equation of the state model.

[0113] During the process of model construction, the blade is scanned comprehensively by a 3D laser scanner to obtain accurate point cloud data of the blade surface. These point cloud data are processed for operations such as noise reduction, polygonization, and surface fitting to construct a 3D CAD model of the blade. This CAD model is imported into finite element analysis software to establish a finite element model of the blade.

[0114] Modal matching analysis is to compare the measured modal parameters extracted by the stochastic subspace method with the modal parameters calculated by the finite element model, and the Modal Assurance Criterion (MAC) is used to quantify the matching degree between the two. The calculation formula of the Modal Assurance Criterion is:

[0115]

[0116] In the formula, MAC ij represents the Modal Assurance Criterion between the test state modal vector i and the finite element analysis modal vector j, represents the modal vector in the test state, represents the modal vector in the finite element analysis state, represents the transpose of the modal vector in the test state, represents the transpose of the modal vector in the finite element analysis state.

[0117] For a wind turbine blade structure with multiple degrees of freedom, there are multiple natural vibration frequencies f1, f2, …, f n, at a certain natural frequency f k Under this condition, the vibration displacement distribution of the structure can be represented by a modal vector The MAC value is mainly used to compare the modal vectors obtained from experimental measurements and those from finite element analysis. The experimental modal vector is based on the vibration test data of the actual structure and reflects the true vibration characteristics of the structure. The finite element modal vector is calculated based on the mathematical model of the structure and represents the theoretical vibration situation. By calculating the MAC value, the accuracy of the finite element model can be evaluated. Theoretically, if the MAC value is close to 1, it indicates that the modes calculated by the finite element model are very similar to those of the actual structure, and the model can better reflect the true vibration characteristics of the structure; when the MAC value is less than 0.8, it indicates that the structural state of the blade has changed and there may be damage.

[0118] In the process of constructing a fan blade damage classification model based on a convolutional neural network (CNN), the specific steps are as follows:

[0119] (1) Data preparation: Simulate various damage situations (such as cracks, corrosion, dust accumulation, etc.), collect the corresponding vibration signals, and transform the collected vibration signals into two-dimensional time-frequency diagrams using the short-time Fourier transform. And label the damage type for each time-frequency diagram sample.

[0120] (2) Construct the CNN model: Use the TensorFlow library in Python to construct a model including a convolutional layer, a pooling layer, and a fully connected layer. The feature of the convolutional layer is to extract local features of the time-frequency diagram using a convolutional kernel and introduce non-linearity using an activation function; the feature of the pooling layer is to downsample the output of the convolutional layer to reduce the data dimension; the feature of the fully connected layer is to integrate the feature maps and output the classification probability using Softmax.

[0121] (3) Model training: Define the loss function, use the cross-entropy loss function to measure the difference between the prediction and the true label, select the Adam optimizer to update the model parameters, and determine parameters such as the learning rate, batch size, and number of training epochs. Finally, use the training set data to iteratively update the model parameters.

[0122] (4) Damage identification: When the MAC value is less than 0.8, the CNN model will output the probability distribution of each damage type, and select the category with the highest probability as the predicted damage type.

[0123] In step S3, an environmental compensation mechanism is generated according to the environmental data at the wind turbine site, and combined with dynamic correction to perform fusion processing on the vibration data, determine the state estimation value of the wind turbine blade, and generate the vibration detection result of the wind turbine blade based on the state estimation value and the vibration damage result.

[0124] In one embodiment, when generating an environmental compensation mechanism based on the environmental data of the wind turbine site, combining it with dynamic correction to perform fusion processing on vibration data, determining the state estimation value of the wind turbine blade, and generating the vibration detection result of the wind turbine blade based on the state estimation value and the vibration damage result, an initial state covariance matrix can be generated according to the static vibration displacement and velocity estimation value of the wind turbine blade, and the subsequent predicted state of the wind turbine blade can be predicted based on the initial state covariance matrix; obtain the vibration displacement and vibration velocity of the wind turbine blade according to the vibration data, set the state vector, and use the response period of the external vibration response as the acquisition time period to obtain the environmental data of the wind turbine site within the acquisition time period; combine the state vector and the environmental data to determine the observation vector, and use the observation vector to correct the subsequent predicted state to obtain the updated state, and obtain the state estimation value of the wind turbine blade based on the updated state; fuse the state estimation value and the vibration damage result to generate the vibration detection result of the wind turbine blade, and judge the fault state of the vibration. If the state is normal, it operates normally. If the state is abnormal, a reminder is sent to the user terminal to adjust the wind turbine blade.

[0125] It should be noted that environmental data such as wind speed, temperature, and humidity are collected in real time at the wind turbine site. At the same time, these environmental data and the vibration data collected by the microcavity sensor are synchronously recorded using the data acquisition system, and it is required to ensure the time consistency of the data. The computer algorithm uses the Kalman filter algorithm, which is an efficient recursive filter. It can filter and correct the vibration data from a series of incomplete and noisy measurement data in the presence of noise, remove the interference of environmental factors on the vibration data, and obtain data that more accurately reflects the true vibration condition of the blade. The specific steps are as follows:

[0126] (1) Construct the system equation state equation: Determine the system state variables, such as blade vibration displacement, velocity, etc. Taking the blade vibration model as an example, let the state vector be:

[0127]

[0128] where, represents the vibration displacement at time k t , represents the vibration velocity, and according to the dynamic characteristics, it is established as:

[0129]

[0130] where, A’ represents the state transition matrix, B’ represents the control input matrix (if there is no external control input, B’ = 0), represents the process noise, which follows a Gaussian distribution with a mean of 0 and a covariance of Q. In the case of simple free vibration, A’ is specifically:

[0131]

[0132] Among them, Δt represents the sampling interval.

[0133] (2) Observation equation: Determine the observation vector Z, which includes vibration data and environmental data.

[0134] Establish H represents the observation matrix, which maps the system state to the observation space. represents the observation noise, which follows a Gaussian distribution with a mean of 0 and a covariance of R. For example, when only measuring the vibration displacement, H = [1, 0].

[0135] (3) Estimate the initial state: Based on the initial measurement, determine the initial state estimate value Such as the estimated values of vibration displacement and velocity obtained from the static measurement of the blade before startup, estimate the initial state covariance matrix P0, which reflects the uncertainty of the initial state estimate. When there is no more information, it can be set as a large diagonal matrix. The specific P0 is:

[0136]

[0137] Among them, and respectively represent the variances of the initial vibration displacement estimate value and the initial velocity estimate value.

[0138] (4) Perform iterative operations:

[0139] Prediction step: Predict the current state according to the previous state estimate and the state transition equation. The predicted state is:

[0140]

[0141] The predicted state covariance is:

[0142]

[0143] In the update step, after obtaining the new measurement data Calculate the Kalman gain:

[0144]

[0145] Use the Kalman gain to correct the predicted state to obtain the updated state estimate:

[0146]

[0147] And update the state covariance:

[0148]

[0149] In the formula, I represents the identity matrix.

[0150] (5) Apply the fusion result: The fused state estimate obtained through Kalman filter iteration Integrates environmental and vibration data, reducing the influence of noise. It can be used for subsequent analysis, such as more accurately identifying blade damage patterns, improving the accuracy of wind turbine blade problem monitoring, and ensuring the safe and stable operation of the unit.

[0151] In summary, by means of the above technical solutions of the present invention, the present invention introduces a fiber optic whispering gallery mode exceptional point microcavity model that can capture optical signals within a very narrow frequency range, making it a highly sensitive vibration detection tool. Furthermore, it can detect tiny vibrations far below the threshold of traditional sensors. When the exceptional point microcavity model is subjected to small vibrations, a large splitting value can be obtained, greatly improving the sensitivity when measuring small vibrations, thus enabling accurate measurement of weak vibrations, broadening the measurement range of vibration values to be measured, and playing a key role in the operation detection and fault prediction of wind turbines.

[0152] A method for detecting the vibration of wind turbine blades based on a fiber optic whispering gallery mode exceptional point microcavity model proposed by the present invention can detect and analyze the tiny vibrations of wind turbine blades with high precision and high sensitivity. At the same time, it has the characteristics of anti-interference ability and adaptability to harsh environments, and combines advanced optical and signal analysis technologies to achieve high-precision and real-time detection of wind turbine blade vibrations, providing guarantee for the safe and stable operation of wind turbines. The present invention introduces a fiber optic whispering gallery mode exceptional point microcavity model that can capture optical signals within a very narrow frequency range, making it a highly sensitive vibration detection tool. And by using the stochastic subspace method to process vibration data for damage identification and combining the Kalman filter algorithm to remove the interference of environmental factors on vibration data, it can achieve accurate measurement of weak vibrations in complex environments and play a key role in the operation detection and fault prediction of wind turbines.

[0153] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting the vibration of a wind turbine blade based on a fiber optic whispering gallery mode singularity microcavity model, characterized in that The method includes: Performing an external vibration response on the detection position of the wind turbine blade to obtain a microcavity response signal, measuring the mode splitting value of the microcavity based on the microcavity response signal, and determining the vibration data of the wind turbine blade; Constructing a three-dimensional model of the wind turbine blade, extracting modal parameters from the vibration data using the random space method, and obtaining the vibration damage result of the wind turbine blade based on the modal parameters and the three-dimensional model; Generating an environmental compensation mechanism according to the environmental data at the wind turbine site, performing fusion processing on the vibration data in combination with dynamic correction, determining the state estimation value of the wind turbine blade, and generating the vibration detection result of the wind turbine blade based on the state estimation value and the vibration damage result.

2. The method for detecting the vibration of a wind turbine blade based on the fiber optic whispering gallery mode singularity microcavity model according to claim 1, wherein, The performing an external vibration response on the detection position of the wind turbine blade to obtain a microcavity response signal, measuring the mode splitting value of the microcavity based on the microcavity response signal, and determining the vibration data of the wind turbine blade includes: Setting a scattering source and a laser, adjusting the spatial characteristics between the scattering source and the microcavity based on a control system, adjusting the microcavity to a singular point state, and coupling the microcavity guiding state using fiber optic guiding technology; Setting the microcavity in the singular point state at the detection position of the wind turbine blade according to the test result, using the microcavity as a vibration sensing detector to perform an external response on the detection position, and outputting a response signal; Converting the response signal into an electrical signal, obtaining the resonant frequency of the microcavity, constructing a singular point microcavity model based on the resonant frequency, and determining the mode splitting value in combination with the response process to obtain a vibration measurement model; Analyzing the vibration value of the wind turbine blade using the vibration measurement model, determining the vibration data of the wind turbine blade, and preliminarily verifying the vibration state of the wind turbine blade according to the vibration data.

3. The method for detecting the vibration of the wind turbine blade based on the optical fiber whispering gallery mode exceptional point microcavity model according to claim 2, wherein, The setting a scattering source and a laser, adjusting the spatial characteristics between the scattering source and the microcavity based on a control system, adjusting the microcavity to a singular point state, and coupling the microcavity guiding state using fiber optic guiding technology includes: Selecting a microcavity and analyzing the transmission spectrum corresponding to the microcavity, finding the resonance mode of the microcavity according to the transmission spectrum, selecting two groups of scattering sources at the same time, and determining the initial spatial position between the microcavity and the scattering sources; Adjusting the distance and phase angle between the scattering source and the edge of the microcavity based on the initial spatial position using the control system, moving any one group of scattering sources based on the adjustment result, and stopping moving after the resonance mode of the microcavity disappears; Analyzing the splitting phenomenon of the resonance mode of the microcavity after moving, and obtaining the microcavity in the singular point state when the splitting phenomenon completely disappears, and turning on the laser to emit a laser light source; Guiding the laser light source to the microcavity in the singular point state using fiber optic guiding technology, performing light source-microcavity coupling processing, and encapsulating the microcavity in the singular point state in an installation piece for standby after the coupling is completed.

4. The method for detecting the vibration of the wind turbine blade based on the fiber optic whispering gallery mode exceptional point microcavity model according to claim 3, wherein, The converting the response signal into an electrical signal, obtaining the resonant frequency of the microcavity, constructing a singular point microcavity model based on the resonant frequency, and determining the mode splitting value in combination with the response process to obtain a vibration measurement model includes: Analyzing the resonant frequency of the microcavity in the singular point state according to the perturbation degree value of the scattering source, and at the same time judging the emission light intensity of the propagation mode of the microcavity in the singular point state based on the angular position of the scattering source, and generating a singular point microcavity model in combination with the emission light intensity; The response signal is amplified by a signal amplifier and converted into discrete digital data. The disturbance quantity and angle during the vibration of the wind turbine blade are obtained according to the discrete digital data. An induction model after the response vibration of the singular point microcavity is generated based on the disturbance quantity and angle, and the induction model is fused with the singular point microcavity model to obtain a vibration response model. Based on the vibration measurement model and the singular point microcavity model, the relationship between the splitting value caused by vibration and the disturbance quantity is analyzed to determine the mode splitting value. The vibration measurement model is generated according to the mode splitting value and the scale factor.

5. The method for detecting the vibration of a wind turbine blade based on the optical fiber whispering gallery mode exceptional point microcavity model according to claim 1, characterized in that, The three-dimensional model of the wind turbine blade is constructed, and the modal parameters are extracted from the vibration data by using the random space method. The vibration damage result of the wind turbine blade obtained based on the modal parameters and the three-dimensional model includes: The vibration data is denoised and normalized, the processed vibration data is set as an output data sequence, and a Hankel matrix is generated according to the preset requirements. The Hankel matrix is decomposed, and the state space model corresponding to the vibration data is calculated according to the processing result. The test modal parameter extraction result of the vibration data is output by using the state space model. The wind turbine blade is scanned by a three-dimensional laser scanner to obtain the point cloud data on the surface of the wind turbine blade, and the point cloud data is polygonized and surface-fitted. A three-dimensional model of the wind turbine is established based on the processed point cloud data, and the finite analysis modal parameters of the wind turbine are obtained by combining the three-dimensional model of the wind turbine with the finite element analysis technology result. The loss probability of the wind turbine blade is evaluated according to the test modal parameters and the finite analysis modal parameters, and a damage prediction model is constructed by combining the vibration signal to obtain the vibration damage result of the wind turbine blade.

6. The method for detecting the vibration of a wind turbine blade based on the fiber optic whispering gallery mode exceptional point microcavity model according to claim 5, characterized in that The decomposition process of the Hankel matrix and the calculation of the state space model corresponding to the vibration data according to the processing result, and the output of the test modal parameter extraction result of the vibration data by using the state space model include: The Hankel matrix is subjected to singular value decomposition based on the singular value decomposition technology. The rank information of the Hankel matrix is analyzed according to the size of the singular values during the processing, and the noise information in the vibration data is obtained. The order of the Hankel matrix is obtained by comparing the size of the singular values with the relative threshold, and the order result is combined with the singular value decomposition processing result to estimate the state sequence of the vibration data. The state space model is constructed by combining the state sequence and the noise information. The output vector of the vibration data is output by using the state space model, and the output vector is used as the test modal parameter extraction result of the vibration data.

7. The method for detecting the vibration of a wind turbine blade based on the fiber optic whispering gallery mode exceptional point microcavity model according to claim 6, characterized in that The evaluation of the loss probability of the wind turbine blade according to the test modal parameters and the finite analysis modal parameters, and the construction of a damage prediction model by combining the vibration signal to obtain the vibration damage result of the wind turbine blade includes: The test modal parameters and the finite analysis modal parameters are respectively used as the test state modal vector and the finite element analysis modal vector, and the matching degree between the test state modal vector and the finite element analysis modal vector is quantified by using the modal confidence criterion. The accuracy of the combination of the three-dimensional model of the wind turbine and the finite element analysis technology is verified according to the matching degree, and the loss probability of the wind turbine blade is evaluated based on the accuracy result and the matching degree. Obtain the corresponding vibration signals when various damages occur to the wind turbine blade, extract the characteristic states based on the vibration signals, and combine the characteristic states with an optimizer to construct a damage prediction model; When the loss probability of the wind turbine blade reaches the threshold, use the damage prediction model to output the probability distribution result of the damage type, and select the category with the largest probability result as the vibration damage category result of the wind turbine blade.

8. The method for detecting the vibration of a wind turbine blade based on the fiber optic whispering gallery mode singularity microcavity model according to claim 7, wherein, The expression of the modal confidence criterion is as follows: Wherein, MAC ij represents the Modal Assurance Criterion between the test state modal vector i and the finite element analysis modal vector j, represents the modal vector in the test state, represents the modal vector in the finite element analysis state, represents the transpose of the modal vector in the test state, represents the transpose of the modal vector in the finite element analysis state.

9. The method for detecting the vibration of a wind turbine blade based on the fiber optic whispering gallery mode singularity microcavity model according to claim 8, characterized in that, The obtaining the corresponding vibration signals when various damages occur to the wind turbine blade, extracting the characteristic states based on the vibration signals, and combining the characteristic states with an optimizer to construct a damage prediction model includes: Simulate the state information of the wind turbine blade in various damage situations, collect the corresponding vibration signals for various damages according to the state information, and use the short-time Fourier transform to convert the vibration signals into two-dimensional time-frequency diagrams, and at the same time label the damage types for the two-dimensional time-frequency diagrams; Use the deep learning framework library to construct a basic model including a convolutional layer, a pooling layer and a fully connected layer, extract the local features of the two-dimensional time-frequency diagram as the convolutional layer features, and at the same time use the activation function to output the classification probability; Define the cross-loss function, and use the cross-loss function to measure the difference between the predicted classification probability and the true label. Select the optimizer according to the difference result to update the parameters of the basic model to obtain the final damage prediction model.

10. The method for detecting the vibration of a wind turbine blade based on the optical fiber whispering gallery mode exceptional point microcavity model according to claim 1, wherein, The generating an environmental compensation mechanism according to the environmental data at the wind turbine site, combining with dynamic correction to perform fusion processing on the vibration data, determining the state estimation value of the wind turbine blade, and generating the vibration detection result of the wind turbine blade based on the state estimation value and the vibration damage result includes: Generate an initial state covariance matrix according to the static vibration displacement and velocity estimation value of the wind turbine blade, and predict the subsequent prediction state of the wind turbine blade based on the initial state covariance matrix; Obtain the vibration displacement and vibration velocity of the wind turbine blade according to the vibration data, set the state vector, and use the response period of the external vibration response as the acquisition time period to obtain the environmental data at the wind turbine site during the acquisition time period; Combine the state vector with the environmental data to determine the observation vector, and use the observation vector to correct the subsequent prediction state to obtain the updated state, and obtain the state estimation value of the wind turbine blade based on the updated state; Fuse the state estimation value with the vibration damage result to generate the vibration detection result of the wind turbine blade, and judge the fault state of the vibration. If the state is normal, it operates normally. If the state is abnormal, a reminder is sent to the user terminal to adjust the wind turbine blade.

Images