Offshore wind turbine blade frequency domain fatigue evaluation method and system based on MIMO nonlinear dynamic system

Abstract

The invention relates to the technical field of fatigue evaluation, and discloses an offshore wind turbine blade frequency domain fatigue evaluation method and system based on an MIMO nonlinear dynamic system, and the method comprises the steps: collecting the load data of an offshore wind turbine blade; preprocessing the load data; an MIMO nonlinear dynamic system is established; establishing a dynamic mapping relation between the preprocessed load data and the frequency domain response by using an MIMO nonlinear dynamic system; decomposing the time domain dynamic response data by using a frequency domain decomposition method, and quantifying the influence weight of each frequency band load on the blade stress based on a frequency domain energy distribution matrix to obtain a frequency domain energy distribution function of the total stress; and calculating the fatigue damage of the blade through a frequency domain fitting optimization method based on the frequency domain energy distribution function of the total stress. The method solves the problems that fatigue damage evaluation confidence is insufficient under the multi-load coupling effect and actual engineering requirements are difficult to meet in the prior art, and has the characteristic that the coupling effect of the wind load and the wave load can be accurately processed.

Description

Technical Field

[0001] The present invention relates to the technical field of fatigue assessment, and more specifically, to a method and system for frequency-domain fatigue assessment of an offshore wind turbine blade based on a MIMO non-linear dynamic system. Background Art

[0002] Existing fatigue damage assessment methods for offshore wind turbine blades usually adopt a serial development mode, that is, first conduct load analysis, then perform structural modeling, and finally carry out fatigue assessment. In this mode, design problems often emerge only in the later integration stage, resulting in a long assessment cycle and high costs. In addition, existing methods are difficult to effectively handle the coupling effect of wind loads and wave loads, resulting in insufficient accuracy of fatigue damage assessment and inability to accurately predict the actual life of the blade. The application of existing frequency-domain fatigue analysis methods in a multi-input multi-output non-linear dynamic system (such as the wind-wave coupling effect) has limitations and cannot accurately calculate the frequency-domain energy distribution function of local maximum stress.

[0003] Fatigue assessment is a key link in the design, manufacture, operation, and maintenance of offshore wind turbine blades, aiming to quantify the damage accumulation of the blade under cyclic loads, predict its remaining life, and thus avoid major accidents caused by fatigue failure.

[0004] There is a prior art method and system for predicting the fatigue life of a wind turbine blade, which relates to the technical field of data processing. The method includes: analyzing historical data and real-time data, identifying key factors affecting the fatigue life of the blade, and determining a dynamic adjustment factor according to the key factors; performing weighted assessment on each area of the blade according to the stress distribution map and the dynamic adjustment factor to determine the key fatigue areas of the wind turbine blade; processing the key fatigue areas to obtain the stress cycle times and stress ranges of the key fatigue areas; predicting the fatigue life of the blade root according to the stress cycle times and stress ranges of the key fatigue areas to obtain a prediction result.

[0005] However, the prior art has the problem of insufficient confidence and is difficult to meet the actual engineering requirements. Therefore, how to invent a method for frequency-domain fatigue assessment of offshore wind turbine blades with high confidence is a technical problem that urgently needs to be solved in this technical field. Summary of the Invention

[0006] In order to solve the problem that the existing technology has insufficient confidence in fatigue damage assessment under the coupling action of multiple loads and is difficult to meet the actual engineering requirements, the present invention provides a method and system for frequency-domain fatigue assessment of an offshore wind turbine blade based on a MIMO non-linear dynamic system, which has the characteristic of being able to accurately handle the coupling effect of wind loads and wave loads.

[0007] To achieve the above object of the present invention, the following technical solutions are adopted:

[0008] A method for frequency-domain fatigue assessment of an offshore wind turbine blade based on a MIMO non-linear dynamic system, comprising the following specific steps:

[0009] Collect the load data of the offshore wind turbine blade;

[0010] Preprocess the load data;

[0011] Establish a MIMO non-linear dynamic system; use the MIMO non-linear dynamic system to establish a dynamic mapping relationship between the preprocessed load data and the frequency-domain response, and obtain the time-domain dynamic response data;

[0012] Decompose the time-domain dynamic response data by using the frequency-domain decomposition method to obtain a frequency-domain energy distribution matrix, and quantify the influence weight of the load in each frequency band on the blade stress based on the frequency-domain energy distribution matrix to obtain the frequency-domain energy distribution function of the total stress;

[0013] Based on the frequency-domain energy distribution function of the total stress, calculate the fatigue damage of the blade by using the frequency-domain fitting optimization method.

[0014] Preferably, the load data specifically includes wind load data, wave load data, and mechanical load data.

[0015] Furthermore, the specific steps for collecting the load data of the offshore wind turbine blade are as follows:

[0016] Real-time monitor the wind speed, wind direction, and turbulence intensity through sensors to obtain continuous time-series data, and construct a wind speed spectrum; use the spectrum analysis method to convert the time-domain signal of the wind speed into a frequency-domain energy distribution to clarify the dynamic influence of the wind energy in different frequency bands on the blade;

[0017] Combine either the actual measurement of a wave buoy or the hydrodynamic simulation method to extract the wave load data including wave height, period, and direction characteristics, generate a wave spectrum, and quantify the frequency distribution law of wave energy;

[0018] Deploy strain gauges and acceleration sensors at the blade root to collect the wave load data including the time history of the blade rotation speed and pitch angle.

[0019] Even further, the specific steps for preprocessing the load data are as follows:

[0020] Synchronize multi-source data using GPS timestamps to eliminate the timing deviation in the load data caused by differences in sensor positions or sampling frequencies;

[0021] For the high-frequency vibration signals in the load data, use the wavelet threshold denoising technique to eliminate noise interference; apply a band-pass filter to the low-frequency wave data to retain the effective frequency band and avoid redundant information.

[0022] Furthermore, establish a MIMO non-linear dynamic system. The specific steps are as follows:

[0023] Define the system inputs as wind load data, wave load data, and mechanical load data;

[0024] Define the system outputs as blade stress, displacement, and vibration frequency;

[0025] Introduce either a non-linear autoregressive exogenous model or a time series neural network model to establish the dynamic mapping relationship between the input and output;

[0026] Introduce a feedforward-feedback composite structure; the feedforward path is used to directly map the input to the output, and the feedback path is used to dynamically adjust the gain change;

[0027] Introduce an optimization algorithm for the system.

[0028] Furthermore, use the MIMO non-linear dynamic system to establish the dynamic mapping relationship between the preprocessed load data and the frequency domain response. The specific steps are as follows:

[0029] Input the wind load data, wave load data, and mechanical load data into the MIMO non-linear dynamic system;

[0030] Through either a non-linear autoregressive exogenous model or a time series neural network, establish the dynamic mapping relationship between the input and output based on the historical load data and the input data. In the mapping, the feedforward path directly maps the input to the output based on the non-linear series model, and the feedback path dynamically adjusts the gain change through the vibration frequency to enhance the adaptability to extreme working conditions;

[0031] Introduce either a genetic algorithm or a particle swarm optimization algorithm. Using the RMSE as the objective function, iteratively adjust the neural network weights and bias parameters, and dynamically update the aerodynamic damping ratio and material damping ratio;

[0032] Output the optimized dynamic mapping relationship.

[0033] Furthermore, use the frequency domain decomposition dynamic mapping relationship to quantify the influence weight of the loads in each frequency band on the blade stress based on the frequency domain energy distribution matrix, and obtain the frequency domain energy distribution function of the total stress. The specific steps are as follows:

[0034] Decompose the time domain dynamic response data into sine wave components of different frequencies, and extract the frequency domain energy distribution;

[0035] Quantify the influence weight of the loads in different frequency bands in the frequency domain energy distribution on the blade stress, and construct the frequency domain energy distribution matrix;

[0036] Based on the frequency domain energy distribution matrix, solve the transfer function between the input load and the output stress, and quantify the contribution of each load frequency band to the blade dynamic response;

[0037] Construct a load coupling model based on the contributions of each load frequency band to the blade dynamic response and historical data, quantify the interaction mechanism between high-frequency turbulence and low-frequency waves, and generate the total stress frequency-domain energy distribution function through spectrum synthesis.

[0038] Furthermore, based on the total stress frequency-domain energy distribution function, calculate the fatigue damage of the blade through a frequency-domain fitting optimization method. The specific steps are as follows:

[0039] Based on the total stress frequency-domain energy distribution function, divide the frequency-domain stress spectrum into different frequency bands according to the frequency range, and distinguish the contributions of different loads to fatigue damage.

[0040] Convert the continuous stress spectrum into discrete stress amplitudes and cycle numbers, assume that the energy of the stress spectrum is concentrated in a specific frequency band, extract the key stress amplitudes, estimate the occurrence frequencies of different stress amplitudes based on the probability density distribution of the stress spectrum, and calculate the cumulative amount of fatigue damage of the blade.

[0041] Furthermore, after calculating the cumulative amount of fatigue damage of the blade, introduce a nonlinear correction coefficient on the basis of the traditional Miner criterion to correct the linear damage accumulation, and generate a fatigue damage distribution map of each part of the blade to identify high-risk areas.

[0042] A frequency-domain fatigue assessment system for offshore wind turbine blades based on a MIMO nonlinear dynamic system, including a data acquisition module, a preprocessing module, a MIMO nonlinear dynamic mapping module, a frequency-domain decomposition module, and a frequency-domain fitting module;

[0043] The data acquisition module is used to collect the load data of the offshore wind turbine blade;

[0044] The preprocessing module is used to preprocess the load data;

[0045] The MIMO nonlinear dynamic mapping module is used to establish a MIMO nonlinear dynamic system; use the MIMO nonlinear dynamic system to establish the dynamic mapping relationship between the preprocessed load data and the frequency-domain response, and obtain the time-domain dynamic response data;

[0046] The frequency-domain decomposition module is used to decompose the time-domain dynamic response data by the frequency-domain decomposition method to obtain the frequency-domain energy distribution matrix, and quantify the influence weight of the load in each frequency band on the blade stress based on the frequency-domain energy distribution matrix to obtain the total stress frequency-domain energy distribution function;

[0047] The frequency-domain fitting module is used to calculate the fatigue damage of the blade through a frequency-domain fitting optimization method based on the total stress frequency-domain energy distribution function.

[0048] The beneficial effects of the present invention are as follows:

[0049] The present invention uses a frequency-domain analysis method to achieve efficient evaluation of fatigue damage, avoiding the problems of large computational amount and long time consumption of traditional time-domain methods. Through the theory of multi-input multi-output nonlinear dynamic systems, the multi-load coupling effect is incorporated into a unified frequency-domain analysis framework to achieve accurate prediction of fatigue damage under multi-load coupling, enabling early detection and solution of problems, so as to solve the problems of insufficient multi-load coupling processing and limited applicability of frequency-domain methods in current blade fatigue evaluation, shorten the fatigue evaluation cycle, reduce the development cost, and improve the accuracy and reliability of evaluation results at the same time. Description of the Drawings

[0050] Figure 1 It is a schematic flow chart of a frequency-domain fatigue evaluation method for offshore wind turbine blades based on a MIMO nonlinear dynamic system of the present invention.

[0051] Figure 2 It is a specific flow chart of a frequency-domain fatigue evaluation method for offshore wind turbine blades based on a MIMO nonlinear dynamic system in Embodiment 2 of the present invention.

[0052] Figure 3 It is a schematic system diagram of a frequency-domain fatigue evaluation system for offshore wind turbine blades based on a MIMO nonlinear dynamic system of the present invention. Detailed Embodiments

[0053] The present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0054] Embodiment 1

[0055] As Figure 1 shown, a frequency-domain fatigue evaluation method for offshore wind turbine blades based on a MIMO nonlinear dynamic system includes the following specific steps:

[0056] Collect the load data of the offshore wind turbine blades;

[0057] Preprocess the load data;

[0058] Establish a MIMO nonlinear dynamic system; use the MIMO nonlinear dynamic system to establish a dynamic mapping relationship between the preprocessed load data and the frequency-domain response to obtain time-domain dynamic response data;

[0059] Use the frequency-domain decomposition method to decompose the time-domain dynamic response data to obtain a frequency-domain energy distribution matrix, and quantify the influence weight of the load in each frequency band on the blade stress based on the frequency-domain energy distribution matrix to obtain the frequency-domain energy distribution function of the total stress;

[0060] Based on the frequency-domain energy distribution function of the total stress, calculate the fatigue damage of the blade through the frequency-domain fitting optimization method.

[0061] Different from traditional time-domain analysis methods and frequency-domain analysis methods, the present invention proposes a frequency-domain fatigue assessment method based on the theory of multi-input multi-output nonlinear dynamic systems. By combining frequency-domain analysis with the theory of multi-input multi-output nonlinear dynamic systems, it realizes the efficient and accurate assessment of blade fatigue damage under the coupled action of wind and waves. If the load data has not been fully obtained, a load model can be constructed through a wind-wave scatter diagram and historical data; if some load data has been obtained, it can be directly incorporated into the frequency-domain analysis framework to achieve the fatigue damage assessment under the coupled action of multiple loads. This method optimizes the traditional serial assessment process into parallel processing. As the development progress advances, the confidence level of the assessment results gradually increases, while avoiding the problems of computational waste and insufficient accuracy in traditional methods. This method supports multiple iterative optimizations, can quickly respond to design changes, and makes full use of existing data resources to reduce redundant calculations.

[0062] The present invention aims to solve the problems of insufficient multi-load coupling processing, limited applicability of frequency-domain methods, and low assessment efficiency in the fatigue damage assessment of offshore wind turbine blades. By means of the theory of multi-input multi-output nonlinear dynamic systems and frequency-domain analysis methods, it realizes the high efficiency, accuracy, and parallelization of fatigue damage assessment, shortens the assessment cycle, reduces the development cost, and improves the reliability and engineering practicability of the assessment results.

[0063] Embodiment 2

[0064] As Figure 2 shown, more specifically, the load data specifically includes wind load data, wave load data, and mechanical load data.

[0065] In a specific embodiment, the steps for collecting the load data of an offshore wind turbine blade are as follows:

[0066] Real-time monitor the wind speed, wind direction, and turbulence intensity through sensors to obtain continuous time-series data and construct a wind speed spectrum; use the spectrum analysis method, considering the load time history, convert the time-domain signal of the wind speed into a frequency-domain energy distribution function matrix, and clarify the dynamic influence of wind energy in different frequency bands on the blade;

[0067] Combined with either the actual measurement of a wave buoy or the hydrodynamic simulation method, extract the wave load data including wave height, period, and direction characteristics, generate a wave spectrum, quantify the frequency distribution law of wave energy, and construct a frequency-domain energy distribution function of the stress under the action of wave loads;

[0068] Deploy strain gauges and acceleration sensors at the blade root to collect the wave load data including the time history of blade rotation speed and pitch angle, construct the spectrum analysis of the blade structure, and further construct the stress transfer function of the blade.

[0069] In a specific embodiment, the preprocessing of the load data is carried out with the following specific steps:

[0070] Synchronize multi-source data with GPS timestamps using the IEEE 1588 protocol to eliminate the timing deviation in the load data caused by differences in sensor positions or sampling frequencies;

[0071] For the high-frequency vibration signals in the load data, use the Daubechies wavelet threshold denoising technique to eliminate noise interference; apply a band-pass filter to the low-frequency wave data to retain the effective frequency band and avoid redundant information.

[0072] In a specific embodiment, the establishment of a MIMO non-linear dynamic system is carried out with the following specific steps:

[0073] Define the system inputs as wind load data, wave load data, and mechanical load data;

[0074] Define the system outputs as blade stress, displacement, and vibration frequency;

[0075] Introduce either a non-linear autoregressive exogenous model or a time series neural network model to establish the dynamic mapping relationship between the inputs and outputs;

[0076] Introduce a feedforward-feedback composite structure; the feedforward path is used to directly map the inputs to the outputs, and the feedback path is used to dynamically adjust the gain changes;

[0077] Introduce an optimization algorithm for the system.

[0078] In a specific embodiment, the establishment of the dynamic mapping relationship between the preprocessed load data and the frequency domain response using the MIMO non-linear dynamic system is carried out with the following specific steps:

[0079] Input the wind load data, wave load data, and mechanical load data into the MIMO non-linear dynamic system;

[0080] Establish the dynamic mapping relationship between the inputs and outputs based on the historical load data and the input data through either a non-linear autoregressive exogenous model or a time series neural network. In the mapping, the feedforward path directly maps the inputs to the outputs based on a non-linear polynomial series model, and the feedback path dynamically adjusts the gain changes through the vibration frequency to enhance the adaptability to extreme working conditions;

[0081] Introduce either a genetic algorithm or a particle swarm optimization algorithm. Using the RMSE as the objective function, iteratively adjust the neural network weights and bias parameters, and dynamically update the aerodynamic damping ratio and the material damping ratio;

[0082] Output the optimized dynamic mapping relationship.

[0083] In a specific embodiment, by using the frequency-domain decomposition dynamic mapping relationship, the influence weight of the load in each frequency band on the blade stress is quantified based on the frequency-domain energy distribution matrix, and the frequency-domain energy distribution function of the total stress is obtained. The specific steps are as follows:

[0084] Convert the time-domain dynamic response data into a frequency-domain signal by either the fast Fourier transform or the wavelet transform, calculate the power spectral density, and obtain the initial frequency-domain energy distribution;

[0085] Divide the frequency-domain signal into at least two different frequency bands according to a preset frequency band; in this embodiment, the quasi-static frequency band is defined as 0 - 1 Hz: quasi-static fatigue caused by wave loads. The dynamic vibration frequency band is defined as 1 - 50 Hz: vibration fatigue caused by wind turbulence.

[0086] Integrate the power spectral density within each frequency band [f k , f k+1 to obtain the energy value of each frequency band:

[0087]

[0088] where X(f) represents the spectral signal, S xx (f) = |X(f)| 2 , representing the energy distribution of each frequency point;

[0089] Determine the contribution ratio of each load to the output stress within the corresponding frequency band by either transfer function analysis or coherence function, and obtain the weight of each load type and frequency band:

[0090] Let H ij (f) be the transfer function from the i-th load to the j-th stress output, and the contribution weight of frequency band k is:

[0091]

[0092] where S ii (f) is the power spectral density of the input load;

[0093] Organize the weights into a matrix according to the load type and frequency band:

[0094]

[0095] Based on the frequency-domain energy distribution matrix, solve the transfer function between the input load and the output stress, and quantify the contribution of each load frequency band to the blade dynamic response;

[0096] Based on the contribution of each load frequency band to the blade dynamic response and historical data, construct a load coupling model, quantify the interaction mechanism between high-frequency turbulence and low-frequency waves, and generate the frequency-domain energy distribution function of the total stress through spectral synthesis.

[0097] In a specific embodiment, based on the frequency-domain energy distribution function of the total stress, the fatigue damage of the blade is calculated by the frequency-domain fitting optimization method. The specific steps are as follows:

[0098] Based on the frequency-domain energy distribution function of the total stress, the frequency-domain stress spectrum is divided into different frequency bands according to the frequency range to distinguish the contributions of different loads to fatigue damage.

[0099] The continuous stress spectrum is converted into discrete stress amplitudes and cycle numbers. Assuming that the energy of the stress spectrum is concentrated in a specific frequency band, the key stress amplitudes are extracted. Based on the probability density distribution of the stress spectrum, the occurrence frequencies of different stress amplitudes are estimated, and the cumulative amount of fatigue damage of the blade is calculated.

[0100] In a specific embodiment, after calculating the cumulative amount of fatigue damage of the blade, a nonlinear correction coefficient is introduced on the basis of the traditional Miner criterion to correct the linear damage accumulation, and a fatigue damage distribution map of each part of the blade is generated to identify high-risk areas such as the root bolt holes and spar joints of the blade.

[0101] Embodiment 3

[0102] As Figure 3 shown, a frequency-domain fatigue assessment system for an offshore wind turbine blade based on a MIMO nonlinear dynamic system includes a data acquisition module, a preprocessing module, a MIMO nonlinear dynamic mapping module, a frequency-domain decomposition module, and a frequency-domain fitting module.

[0103] The data acquisition module is used to collect the load data of the offshore wind turbine blade.

[0104] The preprocessing module is used to preprocess the load data.

[0105] The MIMO nonlinear dynamic mapping module is used to establish a MIMO nonlinear dynamic system; a dynamic mapping relationship between the preprocessed load data and the frequency-domain response is established by using the MIMO nonlinear dynamic system to obtain time-domain dynamic response data.

[0106] The frequency-domain decomposition module is used to decompose the time-domain dynamic response data by using the frequency-domain decomposition method to obtain a frequency-domain energy distribution matrix, and the influence weight of the load in each frequency band on the blade stress is quantified based on the frequency-domain energy distribution matrix to obtain the frequency-domain energy distribution function of the total stress.

[0107] The frequency-domain fitting module is used to calculate the fatigue damage of the blade by the frequency-domain fitting optimization method based on the frequency-domain energy distribution function of the total stress.

[0108] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limitations on the implementation manners of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A method for frequency-domain fatigue assessment of offshore wind turbine blades based on MIMO nonlinear dynamic systems, characterized in that: It includes the following specific steps: Collect the load data of the offshore wind turbine blade; Preprocess the load data; Establish a MIMO nonlinear dynamic system; use the MIMO nonlinear dynamic system to establish the dynamic mapping relationship between the preprocessed load data and the frequency-domain response, and obtain the time-domain dynamic response data; Decompose the time-domain dynamic response data by the frequency-domain decomposition method to obtain the frequency-domain energy distribution matrix, and quantify the influence weight of the loads in each frequency band on the blade stress based on the frequency-domain energy distribution matrix to obtain the frequency-domain energy distribution function of the total stress; Calculate the fatigue damage of the blade through the frequency-domain fitting optimization method based on the frequency-domain energy distribution function of the total stress.

2. The method for evaluating the frequency-domain fatigue of an offshore wind turbine blade based on an MIMO nonlinear dynamic system according to claim 1, wherein: The load data specifically includes wind load data, wave load data, and mechanical load data.

3. The method for evaluating the frequency-domain fatigue of an offshore wind turbine blade based on an MIMO nonlinear dynamic system according to claim 2, wherein: The specific steps for collecting the load data of the offshore wind turbine blade are as follows: Real-time monitor the wind speed, wind direction, and turbulence intensity through sensors to obtain continuous time-series data and construct the wind speed spectrum; use the spectrum analysis method to convert the time-domain signal of the wind speed into the frequency-domain energy distribution to clarify the dynamic influence of the wind energy in different frequency bands on the blade; Combine either the actual measurement of the wave buoy or the hydrodynamic simulation method to extract the wave load data including the wave height, period, and direction characteristics, generate the wave spectrum, and quantify the frequency distribution law of the wave energy; Deploy strain gauges and acceleration sensors at the blade root to collect the wave load data including the time history of the blade rotation speed and pitch angle.

4. The method for evaluating the frequency-domain fatigue of an offshore wind turbine blade based on an MIMO nonlinear dynamic system according to claim 1, wherein: The specific steps for preprocessing the load data are as follows: Synchronize multi-source data using GPS timestamps to eliminate the timing deviation in the load data caused by differences in sensor positions or sampling frequencies; For the high-frequency vibration signals in the load data, use the wavelet threshold denoising technique to eliminate the noise interference; Apply a band-pass filter to the low-frequency wave data to retain the effective frequency band and avoid redundant information.

5. The method for evaluating the frequency-domain fatigue of the off-shore wind turbine blade based on the MIMO nonlinear dynamic system according to claim 3, wherein: The specific steps for establishing a MIMO nonlinear dynamic system are as follows: Define the system input as wind load data, wave load data, and mechanical load data; Define the system output as blade stress, displacement, and vibration frequency; Introduce either the nonlinear autoregressive exogenous model or the time series neural network model to establish the dynamic mapping relationship between the input and output; Introduce a feedforward-feedback composite structure; the feedforward path is used to directly map the input to the output, and the feedback path is used to dynamically adjust the gain change; Introduce the optimization algorithm of the system.

6. The method for evaluating the frequency-domain fatigue of an offshore wind turbine blade based on an MIMO nonlinear dynamic system according to claim 5, wherein: The specific steps for using the MIMO nonlinear dynamic system to establish the dynamic mapping relationship between the preprocessed load data and the frequency-domain response are as follows: Input the wind load data, wave load data, and mechanical load data into the MIMO nonlinear dynamic system; Through either the nonlinear autoregressive exogenous model or the time series neural network, establish the dynamic mapping relationship between the input and output based on the historical load data and the input data. In the mapping, the feedforward path directly maps the input to the output based on the nonlinear series model, and the feedback path dynamically adjusts the gain change through the vibration frequency to enhance the adaptability to extreme working conditions; Introduce either the genetic algorithm or the particle swarm optimization algorithm, use the RMSE as the objective function, iteratively adjust the neural network weight and bias parameters, and dynamically update the aerodynamic damping ratio and material damping ratio; Output the optimized dynamic mapping relationship.

7. The method for evaluating the frequency-domain fatigue of an offshore wind turbine blade based on an MIMO nonlinear dynamic system according to claim 1, wherein: Using the frequency-domain decomposition dynamic mapping relationship, based on the frequency-domain energy distribution matrix, quantify the influence weight of the load in each frequency band on the blade stress, and obtain the frequency-domain energy distribution function of the total stress. The specific steps are as follows: Convert the time-domain dynamic response data into a frequency-domain signal by either the fast Fourier transform or wavelet transform method, calculate the power spectral density, and obtain the initial frequency-domain energy distribution; Divide the frequency-domain signal into at least two different frequency bands according to the preset frequency bands; Integrate the power spectral density within each frequency band to obtain the energy value of each frequency band, and construct the initial frequency-domain energy distribution matrix; Based on the frequency-domain energy distribution matrix, solve the transfer function between the input load and the output stress, and quantify the contribution of each load frequency band to the blade dynamic response; Based on the contribution of each load frequency band to the blade dynamic response and historical data, construct a load coupling model, quantify the interaction mechanism between high-frequency turbulence and low-frequency waves, and generate the frequency-domain energy distribution function of the total stress through spectrum synthesis.

8. The method for evaluating the frequency-domain fatigue of an off-shore wind turbine blade based on an MIMO non-linear dynamic system according to claim 7, wherein: Based on the frequency-domain energy distribution function of the total stress, calculate the fatigue damage of the blade through the frequency-domain fitting optimization method. The specific steps are as follows: Based on the frequency-domain energy distribution function of the total stress, divide the frequency-domain stress spectrum into different frequency bands according to the frequency range, and distinguish the contribution of different loads to the fatigue damage; Convert the continuous stress spectrum into discrete stress amplitudes and cycle numbers. Assume that the energy of the stress spectrum is concentrated in a specific frequency band, extract the key stress amplitudes, and based on the probability density distribution of the stress spectrum, estimate the occurrence frequency of different stress amplitudes. Combine the fatigue characteristics of the material to calculate the cumulative amount of the fatigue damage of the blade.

9. The method for evaluating the frequency-domain fatigue of an offshore wind turbine blade based on an MIMO nonlinear dynamic system according to claim 8, wherein: After calculating the cumulative amount of the fatigue damage of the blade, introduce a non-linear correction coefficient based on the traditional Miner criterion to correct the linear damage accumulation, and generate the fatigue damage distribution map of each part of the blade to identify the high-risk areas.

10. A frequency-domain fatigue assessment system for an offshore wind turbine blade based on a MIMO nonlinear dynamic system, characterized in that: It includes a data acquisition module, a preprocessing module, a MIMO non-linear dynamic mapping module, a frequency-domain decomposition module, and a frequency-domain fitting module; The data acquisition module is used to acquire the load data of the offshore wind turbine blade; The preprocessing module is used to preprocess the load data; The MIMO non-linear dynamic mapping module is used to establish a MIMO non-linear dynamic system; Use the MIMO non-linear dynamic system to establish the dynamic mapping relationship between the preprocessed load data and the frequency-domain response, and obtain the time-domain dynamic response data; The frequency-domain decomposition module is used to decompose the time-domain dynamic response data by the frequency-domain decomposition method to obtain the frequency-domain energy distribution matrix, and based on the frequency-domain energy distribution matrix, quantify the influence weight of the load in each frequency band on the blade stress, and obtain the frequency-domain energy distribution function of the total stress; The frequency-domain fitting module is used to calculate the fatigue damage of the blade through the frequency-domain fitting optimization method based on the frequency-domain energy distribution function of the total stress.

Images