Offshore wind turbine blade pitch control method and device, computer equipment and medium

Abstract

This invention discloses a method, device, computer equipment, and medium for pitch control of offshore wind turbine blades. The method includes: receiving pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine end; the multi-source heterogeneous monitoring data is acquired through a high-frequency multi-dimensional monitoring network that integrates mechanical state and marine environmental characteristics; constructing a digital twin model by integrating physical mechanisms and data compensation; continuously purifying the multi-source heterogeneous monitoring data through a potential denoising autoencoder model and calibrating the digital twin model; establishing a multi-dimensional health assessment index system based on the digital twin model to calculate health assessment indicators for the health status of key components of the offshore wind turbine; calculating the safe operating boundary of the offshore wind turbine under the current health state based on the real-time health assessment results; and converting the dynamic safe operating boundary and real-time health state into pitch control commands for each blade. This invention can improve the safety and power generation economy of offshore wind turbines.

Description

Technical Field

[0001] This invention relates to a method, device, equipment, and medium for controlling the pitch of offshore wind turbine blades, belonging to the field of wind power generation technology and marine engineering equipment. Background Technology

[0002] With the accelerated global energy structure transformation, offshore wind power, as an important component of clean energy, has seen continuous and rapid growth in installed capacity. However, offshore wind turbines operate for extended periods in environments characterized by high salt spray, high humidity, strong turbulence, and complex sea conditions with alternating loads. This makes the blades susceptible to corrosion and icing, the drivetrain subjected to alternating stress and wear, and the tower and foundation structures facing cumulative fatigue damage. Consequently, the degradation rate of components is significantly higher than that of onshore wind turbines, leading to a substantial increase in unplanned downtime costs and maintenance risks. Traditional wind turbine pitch control systems are generally based on preset fixed design parameters (such as rated power, ultimate load, and cut-off wind speed) for logic control, failing to dynamically adjust operating boundaries according to the actual health status of components. This results in a structural defect: a lack of a closed-loop "health status-control decision" mechanism.

[0003] Existing pitch control technologies mainly include Collective Pitch Control (CPC) and Independent Pitch Control (IPC). CPC achieves power regulation and load control by synchronously adjusting the pitch angle of the three blades, but ignores individual differences between blades and aerodynamic imbalances caused by local damage. While IPC can perform differentiated adjustments for individual blades, its triggering mechanism is mostly based on feedback from single physical quantities such as blade load or tower vibration, lacking a systematic assessment of the overall health status of components. Furthermore, its compensation strategy is fixed and difficult to adapt to the dynamic control requirements during damage evolution. In addition, existing IPC applications mostly focus on reducing fatigue loads and do not incorporate health status as a core control variable into the decision-making system.

[0004] In terms of condition monitoring, current offshore wind turbines commonly deploy sensors for vibration, temperature, and pressure. However, the monitoring system exhibits "data silos": sensor types are limited, sampling frequencies are heterogeneous, time synchronization accuracy is insufficient (typically at the millisecond level), and there is a lack of deep integration of marine environmental parameters (waves, currents, icing, corrosion) with mechanical condition parameters. At the data processing level, simple threshold alarms and trend analyses are insufficient to effectively eliminate marine environmental noise and sensor drift interference, leading to a high false alarm rate. While digital twin technology has seen initial applications in the wind power sector, it is mostly offline simulation or one-way data mapping, lacking online model calibration and uncertainty quantification mechanisms based on real-time monitoring data, making it difficult to guarantee the consistency between the virtual image and the physical entity.

[0005] At the health assessment level, traditional methods rely on single indicators (such as vibration amplitude, number of metal particles in oil) or empirical rules, failing to establish a comprehensive, quantifiable health assessment system covering blade aerodynamic efficiency, structural strength, surface integrity, and components such as gearboxes and generators. Assessment results are mostly qualitative and cannot accurately support the quantitative adjustment of control parameters. More critically, existing technologies have failed to translate health assessment results into real-time, executable dynamic adjustment strategies for operating boundaries, causing wind turbines to continue operating at full load with rated parameters even in sub-healthy or early-stage component failure states, accelerating damage evolution and potentially leading to catastrophic failure.

[0006] In summary, existing offshore wind turbine pitch control technology suffers from the following core problems: 1) insufficient dimensions and limited accuracy in health status perception; 2) lack of high-fidelity, real-time calibrable digital twin models; 3) severe disconnect between health assessment and pitch control decision-making; and 4) statically fixed operating boundaries, unable to adapt to the evolution of component health status. These problems severely restrict the improvement of intelligent operation and maintenance levels and the overall lifecycle power generation efficiency of offshore wind turbines, necessitating the development of a new intelligent pitch control method capable of achieving a closed loop of "monitoring-assessment-decision-control". Summary of the Invention

[0007] In view of this, the present invention provides a method, device, computer equipment and storage medium for controlling the pitch of offshore wind turbine blades. It integrates multi-physics field sensing and monitoring, high-fidelity digital twin modeling, deep learning data purification and health status assessment technology to achieve real-time quantitative assessment of the health status of key components of offshore wind turbine generators, and dynamically optimizes the blade pitch control strategy based on the assessment results, thereby improving the safety and power generation economy of offshore wind turbines.

[0008] The first objective of this invention is to provide a method for controlling the pitch of offshore wind turbine blades.

[0009] The second objective of this invention is to provide a pitch control device for offshore wind turbine blades.

[0010] A third objective of this invention is to provide a computer device.

[0011] A fourth objective of this invention is to provide a storage medium.

[0012] The first objective of this invention can be achieved by adopting the following technical solution:

[0013] A method for controlling the pitch of offshore wind turbine blades, the method comprising:

[0014] Receive pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine end. The multi-source heterogeneous monitoring data is acquired through a high-frequency multi-dimensional monitoring network that integrates mechanical status and marine environmental characteristics.

[0015] By integrating the physical mechanisms and data compensation of offshore wind turbines, a digital twin model is constructed.

[0016] The potential denoising autoencoder model is used to continuously purify multi-source heterogeneous monitoring data and calibrate the digital twin model.

[0017] Based on purification data and combined with digital twin models, a multi-dimensional health assessment index system is established to calculate health assessment indicators for the health status of key components of offshore wind turbines.

[0018] Based on the real-time health assessment results, calculate the safe operating boundary of the offshore wind turbine under the current health status;

[0019] The dynamic safety operating boundary and real-time health status are converted into pitch control commands for each blade, thereby enabling pitch control of the blades.

[0020] Furthermore, the high-frequency multidimensional monitoring network includes a triaxial vibration sensor, an online oil monitoring unit, a fiber optic strain sensor, a triaxial acceleration sensor, a biaxial tilt sensor, an optical icing detector, an electrochemical corrosion rate sensor, a lidar, an acoustic Doppler wave velocity profiler, a preload sensor, an acoustic sensor array, and a data acquisition and monitoring control (SCADA) system.

[0021] The triaxial vibration sensors are deployed at the meshing points of gears in each stage of the gearbox, at each stage of bearing housings, at the front and rear bearings of the generator, and on the main shaft bearing; the online oil monitoring unit is connected in series in the gearbox lubrication circuit; the fiber optic strain sensors are attached along the principal stress direction at the blade root section, blade mid-section, and at the bottom and middle of the tower; the triaxial acceleration sensor and the biaxial tilt sensor are installed at the top of the tower and the center of mass of the nacelle; the optical icing detector and the electrochemical corrosion rate sensor are embedded on the leading edge surface of the blade; the lidar is installed at the center of the top of the nacelle; the acoustic Doppler wave velocity profiler is deployed near the monopile foundation or jacket foundation platform; the preload sensor is installed on the bolts connecting the pitch bearing to the hub and on the T-bolts at the blade root; the acoustic sensor array is arranged inside the nacelle, inside the hub, and in the electrical cabinet at the bottom of the tower.

[0022] The preprocessing process of the multi-source heterogeneous monitoring data includes: in the edge computing unit near the sensor, the vibration data and acoustic data are processed in three levels, namely signal filtering, outlier removal and feature extraction; the sensor time synchronization is achieved by using a precise time protocol, a timestamp is embedded in each data point, and spline interpolation is used to resample asynchronous data.

[0023] Furthermore, the digital twin model constructed by integrating the physical mechanisms and data compensation of the offshore wind turbine includes:

[0024] A blade aerodynamic load model is established based on the blade element momentum theory. The blade is divided into multiple blade element segments along the span. The local angle of attack, lift coefficient and drag coefficient are calculated for each blade element segment. Tip loss, tower shadow effect, wake effect, wind shear and blade elastic deformation based on Euler-Bernoulli beam theory are also integrated.

[0025] A dynamic model of the tower and foundation structure was established based on finite element analysis. Beam elements were used to simulate the tower, and spring elements were used to simulate the soil-structure interaction.

[0026] A dual-mass transmission chain model is adopted, in which the low-speed shaft is equivalent to a damped torsion spring and the high-speed shaft is equivalent to a rigid body;

[0027] Establish an electromagnetic transient model of the generator and a control model of the converter, and reproduce the torque control and pitch control logic of the main control system of the offshore wind turbine.

[0028] Long short-term memory network model is used to identify aerodynamic efficiency and transmission efficiency. Historical SCADA data of offshore wind turbine operation is collected and the long short-term memory network model is trained to predict the deviation between actual power and theoretical power as a correction term for the physical model.

[0029] The aerodynamic load model of the blade, the dynamic model of the tower and foundation structure, the dual-mass transmission chain model, the electromagnetic transient model of the generator, and the converter control model are integrated to form a digital twin model.

[0030] Furthermore, the continuous purification of multi-source heterogeneous monitoring data through a potential denoising autoencoder model and the calibration of the digital twin model include:

[0031] A latent denoising autoencoder model is embedded in the digital twin model, and the latent denoising autoencoder model is trained offline using historical data of offshore wind turbines.

[0032] After multi-source heterogeneous monitoring data enters the digital twin model, it is compressed into low-dimensional latent variables through a latent denoising autoencoder model, and then reconstructed into clean data by a decoder.

[0033] The unscented Kalman filter algorithm is used to recursively estimate and correct the deformation of offshore wind turbine blades, transmission chain torque, and tower displacement by using the purified data as observations. The process noise covariance and observation noise covariance of the unscented Kalman filter are dynamically adjusted according to the uncertainty of the output of the potential denoising autoencoder model to achieve adaptive filtering.

[0034] Furthermore, based on purification data and combined with a digital twin model, a multi-dimensional health assessment index system is established to calculate health assessment indicators for the health status of key components of offshore wind turbines, including:

[0035] Establish wind turbine structural health indicators: aerodynamic efficiency health based on aerodynamic efficiency value deviation calculated by digital twin model, structural strength health based on fatigue damage accumulation based on blade strain monitoring, and surface condition health based on blade icing monitoring data and blade corrosion monitoring data.

[0036] Establish gearbox health indicators: vibration health based on the deviation rate of gear meshing frequency amplitude in the engine compartment, bearing health based on the characteristic frequency energy of the outer / inner ring of the power generation system, oil health based on the concentration of wear particles in the gearbox, and temperature health based on the deviation of the gearbox lubrication system oil temperature from the theoretical value.

[0037] Establish generator health indicators: winding insulation health based on generator partial discharge amplitude, bearing health based on generator high-frequency vibration kurtosis value, and cooling system health based on generator coolant temperature gradient;

[0038] The digital twin model collects multiple sets of steady-state data and multiple sets of dynamic data in each preset wind speed range from the cut-in wind speed to the rated wind speed range, records the theoretical parameters of each component, forms a health baseline, and stores the data according to season and environment to form a benchmark health database.

[0039] During the operation of the offshore wind turbine, at the first preset time interval, the theoretical state values ​​of the components output by the digital twin model and the measured values ​​after purification by the potential denoising autoencoder model are obtained, and the deviation rate is calculated; the deviation rate is mapped to the health index using the membership function;

[0040] The entropy weight method is used to determine the weight of each indicator and calculate the overall health of the component.

[0041] Furthermore, the calculation of the safe operating boundary of the offshore wind turbine under the current health state based on the real-time health assessment results includes:

[0042] Based on the structural health of the blades and tower, the limit state method is used to calculate in real time the maximum aerodynamic load that the structure can withstand under normal service limit state.

[0043] Assess the current maximum safe output power based on the gearbox health and generator health.

[0044] The maximum aerodynamic load and maximum safe output power that the structure can withstand under normal service limit conditions are taken as the safe operating boundaries.

[0045] Furthermore, the process of converting dynamic safety operating boundaries and real-time health status into pitch commands for each blade to achieve pitch control of the blades includes:

[0046] Below the rated wind speed, the maximum power point tracking algorithm is used to calculate the optimal pitch angle every second preset time interval;

[0047] Above the rated wind speed, constant power control is adopted, and the pitch angle adjustment adopts an adaptive proportional-integral-derivative algorithm.

[0048] When an abnormality in the health of a blade is detected, the system automatically switches to independent pitch control mode and applies independent pitch angle compensation to each blade.

[0049] When the health of the gearbox or generator drops below the preset health level, increase the pitch angle to the preset angle range, control the load reduction rate, and reduce the power accordingly.

[0050] When the predicted wind speed exceeds the dynamic cut-out wind speed or a typhoon warning is received, the emergency feathering procedure is initiated before the third preset time.

[0051] The second objective of this invention can be achieved by adopting the following technical solution:

[0052] A pitch control device for offshore wind turbine blades, the device comprising:

[0053] The data receiving module is used to receive pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine. The multi-source heterogeneous monitoring data is acquired through a high-frequency multi-dimensional monitoring network that integrates mechanical status and marine environmental characteristics.

[0054] The model building module is used to integrate the physical mechanisms and data compensation of offshore wind turbines to build a digital twin model;

[0055] The data cleansing module is used to continuously clean multi-source heterogeneous monitoring data through a potential denoising autoencoder model and calibrate the digital twin model.

[0056] The assessment system establishment module is used to establish a multi-dimensional health assessment indicator system based on purification data and combined with a digital twin model, so as to realize the calculation of health assessment indicators for the health status of key components of offshore wind turbines.

[0057] The calculation module is used to calculate the safe operating boundary of the offshore wind turbine under the current health status based on the real-time health assessment results;

[0058] The pitch control module is used to convert dynamic safe operating boundaries and real-time health status into pitch commands for each blade, thereby enabling pitch control of the blades.

[0059] The third objective of this invention can be achieved by adopting the following technical solution:

[0060] A computer device includes a processor and a memory for storing processor-executable programs, wherein when the processor executes the program stored in the memory, it implements the above-described method for controlling the pitch of offshore wind turbine blades.

[0061] The fourth objective of this invention can be achieved by adopting the following technical solution:

[0062] A storage medium storing a program that, when executed by a processor, implements the above-described method for controlling the pitch of offshore wind turbine blades.

[0063] The present invention has the following advantages over the prior art:

[0064] This invention addresses the shortcomings of traditional control systems that cannot dynamically adjust operating boundaries based on component health status. It constructs a closed-loop architecture of "multi-dimensional monitoring - digital twin - health assessment - intelligent control": deploying heterogeneous sensors for vibration, strain, oil, icing, and corrosion, and synchronizing them; transmitting data after three-stage filtering and feature compression at the edge; establishing a high-fidelity digital twin model by integrating blade element momentum theory and finite element analysis, embedding a potential denoising autoencoder to purify data in real time, and calibrating online through unscented Kalman filtering; establishing a multi-dimensional health assessment index system for blades, gearboxes, and generators to quantify overall health; dynamically calculating and cutting off wind speed and power limits based on health status, automatically reducing load when component health falls below a preset level; and converting health status into refined pitch commands to achieve unified / independent pitch switching driven by health status and emergency feathering control during typhoon warnings, thereby improving wind turbine safety and power generation economy. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0066] Figure 1 This is a flowchart of the offshore wind turbine blade pitch control method according to Embodiment 1 of the present invention.

[0067] Figure 2 This is an architecture diagram of the high-frequency multidimensional monitoring network of Embodiment 1 of the present invention.

[0068] Figure 3 This is a schematic diagram of multi-source heterogeneous monitoring data acquisition and preprocessing in Embodiment 1 of the present invention.

[0069] Figure 4 This is a structural block diagram of the offshore wind turbine blade pitch control device according to Embodiment 2 of the present invention.

[0070] Figure 5 This is a structural block diagram of the computer device according to Embodiment 3 of the present invention. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0072] Example 1:

[0073] like Figure 1 As shown, this embodiment provides a method for controlling the pitch of offshore wind turbine blades. This method is mainly implemented through an onshore control center server and includes the following steps:

[0074] S101 Receives pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine.

[0075] This embodiment first constructs a high-frequency, multi-dimensional monitoring network that integrates mechanical condition and marine environmental characteristics, such as... Figure 2 As shown, the high-frequency multidimensional monitoring network includes a triaxial vibration sensor, an online oil monitoring unit, a fiber optic strain sensor, a triaxial accelerometer, a biaxial tilt sensor, an optical icing detector, an electrochemical corrosion rate sensor, a lidar (LiDAR), an acoustic Doppler current profiler (ADCP), a preload sensor, an acoustic sensor array, and a SCADA (Supervisory Control and Data Acquisition) system. It then acquires multi-source heterogeneous monitoring data through this high-frequency multidimensional monitoring network, such as... Figure 3 As shown, the specific process includes:

[0076] 1) Deploy triaxial vibration sensors on key rotating components such as gear meshing points at each stage of the gearbox, bearing housings at each stage, front and rear bearings of the generator, and main shaft bearings, with a sampling frequency of not less than 10kHz, to fully capture the high-frequency vibration characteristics triggered by early faults such as gear pitting, inner and outer ring spalling of bearings, and rolling element damage; connect an online oil monitoring unit in series in the gearbox lubrication circuit to acquire key oil health indicators such as metal wear particle concentration, particle size distribution, viscosity change rate, and moisture content in real time.

[0077] 2) Fiber optic strain sensors (FBGs) are attached along the principal stress direction at key load-bearing locations such as the blade root section, blade midsection, and the bottom and middle of the tower, with a sampling frequency of not less than 100Hz, to monitor the dynamic strain amplitude of the structure; triaxial accelerometers and biaxial tilt sensors are installed at the top of the tower and the center of mass of the nacelle to monitor the first and second order vibration mode frequencies of the tower and the yaw alignment accuracy of the nacelle; optical icing detectors and electrochemical corrosion rate sensors are embedded on the leading edge surface of the blades to assess the cumulative impact of the marine environment on the aerodynamic performance of the blades.

[0078] 3) Install a nacelle-type lidar at the center of the top of the nacelle, configured in PPI scanning mode, to measure the incoming wind speed, wind direction, wind shear index and turbulence intensity in front of the wind turbine, with a data update frequency of 1Hz; deploy an acoustic Doppler wave velocity profiler near the monopile foundation or jacket foundation platform to monitor marine environmental parameters such as ocean current velocity, wave height and wave period within 20 meters below sea level.

[0079] 4) Install ultrasonic preload sensors or strain gauge preload sensors on key fasteners such as high-strength bolts connecting the pitch bearing to the hub and T-bolts at the blade roots to monitor the bolt preload loss rate in real time; arrange acoustic sensor arrays inside the nacelle, hub, and electrical cabinet at the bottom of the tower to collect 20Hz-20kHz acoustic signals and capture abnormal acoustic features such as bearing noise, arc discharge, and structural impact.

[0080] like Figure 3 As shown, the preprocessing of multi-source heterogeneous monitoring data at the offshore wind turbine end includes the following specific steps:

[0081] 1) In the edge computing unit near the sensor, high-frequency raw data such as vibration and acoustic data are processed in three levels: The first level is signal filtering, which uses a Butterworth low-pass filter with a cutoff frequency of 8kHz to filter out high-frequency electromagnetic noise and a high-pass filter with a cutoff frequency of 5Hz to eliminate sensor temperature drift; The second level is outlier removal, which uses the isolated forest algorithm to identify abnormal data points caused by sensor failure or instantaneous impact, with the removal rate controlled at 0.1%-0.5%; The third level is feature extraction, which calculates RMS value, peak factor and kurtosis in the time domain, and calculates the 1, 2 and 3 times the turnaround amplitude and the sideband energy of the gear meshing frequency through FFT in the frequency domain, compressing the 10kHz raw data into a 10-dimensional feature vector with a data compression ratio of 1000:1.

[0082] 2) Due to the sampling frequency differences of 10kHz for vibration sensors, 100Hz for strain sensors, and 1Hz for SCADA data—a difference of four orders of magnitude—a precise time protocol is used to achieve time synchronization of all sensors in the network, with a synchronization accuracy better than 1 microsecond. A Unix timestamp is embedded in each data point, and spline interpolation is used to resample asynchronous data to unify it to a 1Hz reference frequency.

[0083] 3) The edge computing unit publishes the preprocessed feature data and the original key data to the land-based control center server via industrial Ethernet or 5G wireless private network using the MQTT protocol.

[0084] S102. Integrate the physical mechanism and data compensation of offshore wind turbines to construct a digital twin model.

[0085] This embodiment establishes a high-fidelity digital twin model with over 2000 state variables by integrating leaf element momentum theory and finite element analysis. The construction process includes:

[0086] S1021. Based on the blade element momentum theory, establish a blade aerodynamic load model, divide the blade along the span into 20-30 blade element segments, calculate the local angle of attack, lift coefficient and drag coefficient for each segment, and integrate correction terms such as tip loss, tower shadow effect, wake effect, wind shear and blade elastic deformation based on Euler-Bernoulli beam theory.

[0087] S1022. Based on finite element analysis (FEA), a dynamic model of the tower and foundation structure is established. Beam elements are used to simulate the tower, and spring elements are used to simulate the soil-structure interaction, accurately simulating the first four vibration modes of the tower.

[0088] S1023. A dual-mass transmission chain model is adopted, in which the low-speed shaft is equivalent to a damped torsion spring and the high-speed shaft is equivalent to a rigid body.

[0089] S1024. Establish the electromagnetic transient model of the generator and the control model of the converter, and reproduce the torque control and pitch control logic of the main control system of the offshore wind turbine.

[0090] S1025. To overcome the parameter bias and unmodeled dynamics of the pure physical model (the model established in steps S1021-S2024), a Long Short-Term Memory (LSTM) network model is used to identify aerodynamic efficiency and transmission efficiency online. Historical SCADA data of offshore wind turbine operation is collected, specifically a massive amount of SCADA data (wind speed, pitch angle, speed, and power) over 6 months. The LSTM network model is trained to predict the deviation ΔP between actual power and theoretical power as a correction term for the physical model.

[0091] S1026. The blade aerodynamic load model, tower and foundation structure dynamic model, dual-mass transmission chain model, generator electromagnetic transient model and converter control model are integrated through the MATLAB / Simulink platform to form a high-fidelity digital twin model containing more than 2000 state variables. In order to meet the requirements of real-time calculation, the intrinsic orthogonal decomposition (POD) method is used to reduce the order of the finite element model and extract the first 20 modes to construct the reduced-order model.

[0092] S103. Continuously purify multi-source heterogeneous monitoring data through a potential denoising autoencoder model and calibrate the digital twin model.

[0093] This embodiment cleanses data in real time by embedding a latent denoising autoencoder (L-DAE) model within a digital twin model and calibrates it online using an unscented Kalman filter. The specific process includes:

[0094] S1031. Embed a latent denoising autoencoder model within the digital twin model. Implemented using the TensorFlow framework, the latent denoising autoencoder model is trained offline using historical data of offshore wind turbines (6 months, covering spring, summer, autumn, and winter operating conditions).

[0095] S1032. After the multi-source heterogeneous monitoring data (real-time data stream, batch every 1 second) enters the digital twin model, it is compressed into low-dimensional (8-dimensional) latent variables through a latent denoising autoencoder model, and then reconstructed into cleaned data by a decoder, as shown in the following formula:

[0096] ;

[0097] The encoding process automatically filters out uncertainties such as sensor noise and environmental interference. To verify the purification effect, the reconstruction error is calculated in real time, as shown in the following formula:

[0098] ;

[0099] If reconstruction error If the standard deviation continuously exceeds a preset multiple (3 times in this embodiment), a sensor fault alarm will be triggered.

[0100] S1033, Clean up the data The true value is used to calibrate the state variables of the digital twin model.

[0101] This embodiment employs the Unscented Kalman Filter (UKF) algorithm, using purified data as observations to recursively estimate and correct state vectors such as offshore wind turbine blade deformation, transmission chain torque, and tower displacement. Based on the uncertainty of the output of the potential denoising autoencoder model, the process noise covariance Q and observation noise covariance R of the Unscented Kalman Filter are dynamically adjusted to achieve adaptive filtering.

[0102] S104. Based on purification data and combined with a digital twin model, establish a multi-dimensional health assessment indicator system to calculate the health assessment indicators for the health status of key components of offshore wind turbines.

[0103] This embodiment is based on purification data. By combining digital twin models, a multi-dimensional health assessment index system for blades, gearboxes, and generators is established. The Sigmoid function and entropy weight method are used to quantify the overall health. The specific process includes:

[0104] S1041. Define health indicators for each key component, namely blade health indicator, gearbox health indicator, and generator health indicator. Each of the blade health indicator, gearbox health indicator, and generator health indicator has multiple sub-indicators. The specific establishment process is as follows:

[0105] 1) Establish a wind turbine structural health index: aerodynamic efficiency health based on the deviation of aerodynamic efficiency value (Cp) calculated by a digital twin model. Structural strength health based on fatigue damage accumulation from blade and tower strain monitoring Surface health based on blade icing monitoring data and blade corrosion monitoring data .

[0106] 2) Establish gearbox health indicators: vibration health based on the amplitude deviation rate of gear meshing frequency in the engine room. Bearing health based on the characteristic frequency energy of the outer / inner loop of the power generation system Oil health based on gearbox wear particle concentration Temperature health based on the deviation between the gearbox lubrication system oil temperature and the theoretical value. .

[0107] 3) Establish generator health index: winding insulation health based on generator partial discharge amplitude. Bearing health based on generator high-frequency vibration kurtosis value Cooling system health based on generator coolant temperature gradient .

[0108] S1042. Establish a benchmark database after the wind turbine is put into operation or after maintenance is completed and the turbine is confirmed to be in good condition, and run it for 3-6 months.

[0109] In this embodiment, the digital twin model collects 200 sets of steady-state data and 100 sets of dynamic data in each preset wind speed range (0.5 m / s in this embodiment) between the cut-in wind speed of 3 m / s and the rated wind speed of 12 m / s. The theoretical parameters such as theoretical stress, temperature and vibration of each component are recorded to form a health baseline. The data are stored according to season and environment to form a benchmark health database.

[0110] S1043. During the operation of the offshore wind turbine, the theoretical state values ​​of the components output by the digital twin model are acquired every first preset time interval (1 second in this embodiment). And the measured values ​​after purification by the latent denoising autoencoder model Calculate the deviation rate The deviation rate is mapped to a health sub-indicator using a membership function. ,in, This is the sensitivity coefficient (0.3-0.5 in this embodiment). The deviation threshold is set at 10% for vibration and 5% for temperature in this embodiment.

[0111] S1044. Determine the weights of each health sub-indicator using the entropy weight method. First, the monitoring values ​​of each sub-indicator are normalized to eliminate dimensional differences; second, the information entropy corresponding to each indicator is calculated based on its degree of variation. The greater the dispersion of the indicator data, the higher its information entropy. The smaller the value, the richer the fault characteristic information contained in the indicator; finally, through the formula... Determine the weight of each sub-indicator Taking gearbox health assessment as an example, the weight distribution identified based on historical operating data is typically as follows: vibration health ( =0.45), bearing health ( =0.25), oil health ( =0.20) and temperature health status ( =0.10), this allocation scheme scientifically reflects the high sensitivity of vibration signals in early fault identification.

[0112] Calculate the overall health of components Health grading standards: For "health" (green). The condition is classified as "sub-healthy" (yellow). This is marked as an "early failure" (orange). The "serious fault" (red) indicates that a warning is triggered when the health of a component drops to the orange or red range.

[0113] S105. Based on the real-time health assessment results, calculate the safe operating boundary of the offshore wind turbine under the current health status.

[0114] In this embodiment, the safe operating boundary of the offshore wind turbine under the current healthy state is calculated, and the specific process includes:

[0115] S1051. Determination of Dynamic Maximum Wind Force Withstand: Based on the structural health of the blades and tower, the limit state method is used to calculate in real time the maximum aerodynamic load that the structure can withstand under normal serviceability limit states, including the blade ultimate load. ,in For the structural strength and health of the blades, To design the ultimate load; taking into account real-time wind conditions and turbulence intensity Wind shear index Reverse thrust shear wind speed For example, when the design cut-out wind speed is 25 m / s, the blade health is 0.8, and the turbulence intensity is 0.15, the dynamic cut-out wind speed is adjusted to 22.3 m / s.

[0116] S1052, Dynamic Power Generation Capacity Assessment: Based on Gearbox Health and generator health Assess the current maximum safe output power. ,in The rated power is used; when the gearbox health is lower than the preset health (0.7 in this embodiment), the power limit will not exceed 70% of the rated power regardless of the wind speed, so as to avoid overload of the transmission chain. The dynamic power curve is updated every 10 seconds to ensure a smooth transition and avoid sudden power changes.

[0117] S1053. Taking the maximum aerodynamic load and maximum safe output power that the structure can withstand under normal serviceability limits as the safe operating boundary, and based on the dynamic safe operating boundary and real-time health status, the load utilization rate of components such as the computer cabin, tower, and foundation is determined through a digital twin model. ,in For actual load, For the allowable load, The system dynamically adjusts based on health status; when U>0.85, a Level 1 load warning is triggered, limiting the pitch rate; when U>0.95, a Level 2 warning is triggered, and a load reduction shutdown is executed.

[0118] S106. The dynamic safe operating boundary and real-time health status are converted into pitch control commands for each blade, thereby realizing the pitch control of the blade.

[0119] This embodiment transforms dynamic safe operating boundaries and real-time health status into refined pitch commands, enabling health-driven unified / independent pitch switching and emergency feathering control during typhoon warnings. The specific process includes:

[0120] S1061. Basic strategy of Unified Pitch Control (CPC): Below the rated wind speed, the maximum power point tracking algorithm is used, and the optimal pitch angle is calculated every second preset time interval (10ms in this embodiment). ,in To maintain the tip speed ratio, the pitch angle adjustment rate is limited to 5° / s to avoid structural impact caused by rapid pitch changes. Above the rated wind speed, constant power control is used, and the pitch angle adjustment employs an adaptive proportional-integral-derivative (PID) algorithm, as shown in the following formula:

[0121] ;

[0122] Among them, PID gain , , Dynamically tuned based on real-time health status: When the health status is greater than 0.9, a high gain (e.g., ...) is used. =8), improve response speed; when health is less than 0.7, use low gain (e.g., =4), to ensure stability.

[0123] S1062, Independent Pitch Control (IPC) Differentiation Strategy: When an abnormality in the health of a blade is identified (e.g., When this occurs, it automatically switches to independent pitch control mode, applying independent pitch angle compensation to each blade, as shown in the following formula:

[0124] ;

[0125] in, The blade azimuth angle. This is the phase compensation angle (usually taken as 90° to compensate for aerodynamic imbalance). The gain coefficient for independent pitch control is set to 0.5-2° / health difference; simultaneously, the pitch angles of other healthy blades are adjusted accordingly. To maintain the overall aerodynamic thrust of the machine.

[0126] S1063, Protective Load Reduction Control Strategy: When the health of the gearbox or generator drops below a preset health level, regardless of whether the wind speed exceeds the rated value, the blade pitch angle is actively increased to a preset angle range (2-5° in this example). This reduces the impeller aerodynamic efficiency Cp by 10%-20%, achieving protective load reduction in the transmission chain; the load reduction rate is controlled at 0.5° / s to avoid sudden load changes; simultaneously, the power command is reduced accordingly to keep the tip speed ratio λ near the optimal range.

[0127] S1064. Emergency feathering control strategy under extreme conditions: When the predicted wind speed exceeds the dynamic cut-out wind speed or a typhoon warning is received, the emergency feathering procedure is initiated before the third preset time (30 seconds in this embodiment). The three blades feather synchronously in a coordinated manner, with a feathering rate of 8° / s and a target pitch angle of 85°-90°. The feathering process adopts an S-shaped speed curve (acceleration-uniform acceleration-deceleration) to avoid impact loads on the tower. After feathering is completed, the offshore wind turbine enters a safe shutdown state, and the impeller speed drops to below 1 rpm.

[0128] It should be noted that although the method operations of the above embodiments are described in a specific order, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the described steps may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0129] Example 2:

[0130] like Figure 4 As shown, this embodiment provides a pitch control device for offshore wind turbine blades. The device includes a data receiving module 401, a model building module 402, a data purification module 403, an evaluation system establishment module 404, a calculation module 405, and a pitch control module 406. The specific descriptions of each module are as follows:

[0131] Data receiving module 401 is used to receive pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine end. The multi-source heterogeneous monitoring data is obtained through a high-frequency multi-dimensional monitoring network that integrates mechanical status and marine environmental characteristics.

[0132] Model building module 402 is used to integrate the physical mechanism and data compensation of offshore wind turbines to build a digital twin model;

[0133] The data cleansing module 403 is used to continuously clean multi-source heterogeneous monitoring data through a potential denoising autoencoder model and calibrate the digital twin model.

[0134] The assessment system establishment module 404 is used to establish a multi-dimensional health assessment indicator system based on purification data and combined with a digital twin model, so as to realize the calculation of health assessment indicators for the health status of key components of offshore wind turbines.

[0135] The calculation module 405 is used to calculate the safe operating boundary of the offshore wind turbine under the current health state based on the real-time health assessment results.

[0136] The pitch control module 406 is used to convert the dynamic safe operating boundary and real-time health status into pitch commands for each blade, thereby realizing pitch control of the blade.

[0137] The specific implementation of each module in this embodiment can be found in Embodiment 1 above, and will not be repeated here. It should be noted that the device provided in this embodiment is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure can be divided into different functional modules to complete all or part of the functions described above.

[0138] Example 3:

[0139] This embodiment provides a computer device, such as... Figure 5 As shown, it includes a processor 502, a memory, and a network interface 503 connected via a system bus 501. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium 504 and internal memory 505. The non-volatile storage medium 504 stores an operating system, computer programs, and a database. The internal memory 505 provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. When the processor 502 executes the computer programs stored in the memory, it implements the offshore wind turbine blade pitch control method of Embodiment 1 described above, as follows:

[0140] The system receives pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine. This data is acquired through a high-frequency, multi-dimensional monitoring network that integrates mechanical conditions and marine environmental characteristics. A digital twin model is constructed by integrating the physical mechanisms and data compensation of the offshore wind turbine. The multi-source heterogeneous monitoring data is continuously purified using a potential denoising autoencoder model, and the digital twin model is calibrated. Based on the purified data and the digital twin model, a multi-dimensional health assessment index system is established to calculate health assessment indicators for the health status of key components of the offshore wind turbine. The safe operating boundary of the offshore wind turbine under the current health status is calculated based on the real-time health assessment results. The dynamic safe operating boundary and real-time health status are converted into pitch control commands for each blade, enabling pitch control of the blades.

[0141] Example 4:

[0142] This embodiment provides a storage medium, which is a computer-readable storage medium, storing a computer program. When the program is executed by a processor, the processor executes the computer program stored in the memory to implement the offshore wind turbine blade pitch control method of Embodiment 1 above, as follows:

[0143] The system receives pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine. This data is acquired through a high-frequency, multi-dimensional monitoring network that integrates mechanical conditions and marine environmental characteristics. A digital twin model is constructed by integrating the physical mechanisms and data compensation of the offshore wind turbine. The multi-source heterogeneous monitoring data is continuously purified using a potential denoising autoencoder model, and the digital twin model is calibrated. Based on the purified data and the digital twin model, a multi-dimensional health assessment index system is established to calculate health assessment indicators for the health status of key components of the offshore wind turbine. The safe operating boundary of the offshore wind turbine under the current health status is calculated based on the real-time health assessment results. The dynamic safe operating boundary and real-time health status are converted into pitch control commands for each blade, enabling pitch control of the blades.

[0144] It should be noted that the computer-readable storage medium in this embodiment can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor device, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0145] In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used or combined with an instruction execution device, apparatus, or device. In this embodiment, the computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use or combined with an instruction execution device, apparatus, or device. The computer program contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0146] The computer-readable storage medium described above can be used to write computer programs for executing this embodiment in one or more programming languages ​​or combinations thereof. These programming languages ​​include object-oriented programming languages—such as Java, Python, and C++—and conventional procedural programming languages—such as C or similar programming languages. The program can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0147] In summary, this invention addresses the shortcomings of traditional control systems that cannot dynamically adjust operating boundaries based on component health status. It constructs a closed-loop architecture of "multi-dimensional monitoring - digital twin - health assessment - intelligent control": deploying heterogeneous sensors for vibration, strain, oil, icing, and corrosion, and synchronizing them; transmitting data after three-stage filtering and feature compression at the edge; establishing a high-fidelity digital twin model by integrating blade element momentum theory and finite element analysis, embedding a potential denoising autoencoder to purify data in real time, and calibrating online through unscented Kalman filtering; establishing a multi-dimensional health assessment index system for blades, gearboxes, and generators to quantify overall health; dynamically calculating and cutting off wind speed and power limits based on health status, automatically reducing load when component health falls below a preset level; and converting health status into refined pitch commands to achieve unified / independent pitch switching driven by health status and emergency feathering control during typhoon warnings, thereby improving wind turbine safety and power generation economy.

[0148] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A method for controlling the pitch of offshore wind turbine blades, characterized in that, The method includes: The system receives pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine. This data is acquired through a high-frequency multi-dimensional monitoring network that integrates mechanical condition and marine environmental characteristics. The high-frequency multi-dimensional monitoring network includes a triaxial vibration sensor, an online oil monitoring unit, a fiber optic strain sensor, a triaxial accelerometer, a biaxial tilt sensor, an optical icing detector, an electrochemical corrosion rate sensor, a lidar, an acoustic Doppler wave velocity profiler, a preload sensor, an acoustic sensor array, and a SCADA (Supervisory Control and Data Acquisition) system. The triaxial vibration sensor is deployed at the meshing points of gears in each stage of the gearbox, at each bearing housing, on the front and rear bearings of the generator, and on the main shaft bearing. The online oil monitoring unit is connected in series in the gearbox lubrication circuit. The fiber optic strain sensor is attached along the principal stress direction at the blade root section, blade mid-section, and the bottom and middle of the tower. The triaxial accelerometer... The degree sensor and dual-axis tilt sensor are installed at the top of the tower and the center of mass of the nacelle; the optical icing detector and electrochemical corrosion rate sensor are embedded on the leading edge surface of the blades; the lidar is installed at the center of the top of the nacelle; the acoustic Doppler wave velocity profiler is deployed near the monopile foundation or jacket foundation platform; the preload sensor is installed on the bolts connecting the pitch bearing to the hub and the T-bolts at the blade root; the acoustic sensor array is arranged inside the nacelle, inside the hub, and in the electrical cabinet at the bottom of the tower; the preprocessing process of the multi-source heterogeneous monitoring data includes: in the edge computing unit near the sensor, the vibration data and acoustic data are processed in three levels, namely signal filtering, outlier removal, and feature extraction; a precise time protocol is used to achieve sensor time synchronization, a timestamp is embedded in each data point, and spline interpolation is used to resample asynchronous data; Integrating the physical mechanisms and data compensation of offshore wind turbines, a digital twin model is constructed, including: establishing an aerodynamic load model for the blades based on blade element momentum theory, dividing the blades along the spanwise direction into multiple blade element segments, calculating the local angle of attack, lift coefficient, and drag coefficient for each blade element segment, and integrating tip loss, tower shadow effect, wake effect, wind shear, and blade elastic deformation based on Euler-Bernoulli beam theory; establishing a dynamic model of the tower and foundation structure based on finite element analysis, using beam elements to simulate the tower and spring elements to simulate soil-structure interaction; and employing a dual-mass transmission chain model, equating the low-speed shaft to a damped torsion beam. The high-speed shaft is equivalent to a rigid body by rotating springs; an electromagnetic transient model of the generator and a converter control model are established to reproduce the torque control and pitch control logic of the offshore wind turbine main control system; a long short-term memory network model is used to identify aerodynamic efficiency and transmission efficiency, and historical SCADA data of offshore wind turbine operation are collected to train the long short-term memory network model to predict the deviation between actual power and theoretical power as a correction term for the physical model; the blade aerodynamic load model, tower and foundation structure dynamic model, dual-mass transmission chain model, generator electromagnetic transient model and converter control model are integrated to form a digital twin model; The potential denoising autoencoder model is used to continuously purify multi-source heterogeneous monitoring data and calibrate the digital twin model. Based on purification data and combined with a digital twin model, a multi-dimensional health assessment index system is established to calculate the health assessment indexes for key components of offshore wind turbines. This includes: establishing wind turbine structural health indicators: aerodynamic efficiency health based on aerodynamic efficiency deviation calculated using the digital twin model; structural strength health based on fatigue damage accumulation from blade strain monitoring; and surface condition health based on blade icing and corrosion monitoring data. Gearbox health indicators include: vibration health based on the deviation rate of gear meshing frequency in the nacelle; bearing health based on the characteristic frequency energy of the outer / inner ring of the power generation system; oil health based on the concentration of wear particles in the gearbox; and temperature health based on the deviation of the gearbox lubrication system oil temperature from the theoretical value. Generator health indicators are also established. The health indicators include: winding insulation health based on generator partial discharge amplitude, bearing health based on generator high-frequency vibration kurtosis, and cooling system health based on generator coolant temperature gradient. The digital twin model collects multiple sets of steady-state and dynamic data at each preset wind speed range from the cut-in wind speed to the rated wind speed, records the theoretical parameters of each component, forms a health baseline, and stores these data according to season and environment to form a benchmark health database. During offshore wind turbine operation, at first preset time intervals, the theoretical state values ​​of the components output by the digital twin model and the measured values ​​purified by the latent denoising autoencoder model are acquired, and the deviation rate is calculated. A membership function is used to map the deviation rate to health sub-indicators. The entropy weight method is used to determine the weights of each health sub-indicator, and the overall component health is calculated. Based on the real-time health assessment results, calculate the safe operating boundary of the offshore wind turbine under the current health status; The dynamic safety operating boundary and real-time health status are converted into pitch control commands for each blade, thereby enabling pitch control of the blades.

2. The method for controlling the pitch of offshore wind turbine blades according to claim 1, characterized in that, The process of continuously purifying multi-source heterogeneous monitoring data through a potential denoising autoencoder model and calibrating the digital twin model includes: A latent denoising autoencoder model is embedded in the digital twin model, and the latent denoising autoencoder model is trained offline using historical data of offshore wind turbines. After multi-source heterogeneous monitoring data enters the digital twin model, it is compressed into low-dimensional latent variables through a latent denoising autoencoder model, and then reconstructed into clean data by a decoder. The unscented Kalman filter algorithm is used to recursively estimate and correct the deformation of offshore wind turbine blades, transmission chain torque, and tower displacement by using the purified data as observations. The process noise covariance and observation noise covariance of the unscented Kalman filter are dynamically adjusted according to the uncertainty of the output of the potential denoising autoencoder model to achieve adaptive filtering.

3. The method for controlling the pitch of offshore wind turbine blades according to claim 1, characterized in that, The calculation of the safe operating boundary of the offshore wind turbine under the current health state based on real-time health assessment results includes: Based on the health status of the blade and tower structures, the limit state method is used to calculate in real time the maximum aerodynamic load that the blade and tower structures can withstand under normal service limit states. Assess the current maximum safe output power based on the gearbox health and generator health. The maximum aerodynamic load and maximum safe output power that the blade and tower structure can withstand under normal service limit conditions are taken as the safe operating boundary.

4. The method for controlling the pitch of offshore wind turbine blades according to claim 1, characterized in that, The process of converting dynamic safety operation boundaries and real-time health status into pitch control commands for each blade, thereby achieving pitch control of the blades, includes: Below the rated wind speed, the maximum power point tracking algorithm is used to calculate the optimal pitch angle every second preset time interval; Above the rated wind speed, constant power control is adopted, and the pitch angle adjustment adopts an adaptive proportional-integral-derivative algorithm. When an abnormality in the health of a blade is detected, the system automatically switches to independent pitch control mode and applies independent pitch angle compensation to each blade. When the health of the gearbox or generator drops below the preset health level, increase the pitch angle to the preset angle range, control the load reduction rate, and reduce the power accordingly. When the predicted wind speed exceeds the dynamic cut-out wind speed or a typhoon warning is received, the emergency feathering procedure is initiated before the third preset time.

5. A pitch control device for offshore wind turbine blades, used to implement the pitch control method for offshore wind turbine blades as described in any one of claims 1-4, characterized in that, The device includes: The data receiving module is used to receive pre-processed multi-source heterogeneous monitoring data transmitted from the offshore wind turbine. The multi-source heterogeneous monitoring data is acquired through a high-frequency multi-dimensional monitoring network that integrates mechanical status and marine environmental characteristics. The model building module is used to integrate the physical mechanisms and data compensation of offshore wind turbines to build a digital twin model; The data cleansing module is used to continuously clean multi-source heterogeneous monitoring data through a potential denoising autoencoder model and calibrate the digital twin model. The assessment system establishment module is used to establish a multi-dimensional health assessment indicator system based on purification data and combined with a digital twin model, so as to realize the calculation of health assessment indicators for the health status of key components of offshore wind turbines. The calculation module is used to calculate the safe operating boundary of the offshore wind turbine under the current health status based on the real-time health assessment results; The pitch control module is used to convert dynamic safe operating boundaries and real-time health status into pitch commands for each blade, thereby enabling pitch control of the blades.

6. A computer device comprising a processor and a memory for storing a processor-executable program, characterized in that, When the processor executes the program stored in the memory, it implements the offshore wind turbine blade pitch control method according to any one of claims 1-4.

7. A storage medium storing a program, characterized in that, When the program is executed by the processor, it implements the offshore wind turbine blade pitch control method according to any one of claims 1-4.

Images