Operation control method of wind power generation system

Abstract

The invention discloses an operation control method for a wind power generation system, and relates to the technical field of operation control, and the method comprises the steps: obtaining the historical operation data of a wind power generation system in a target sea area, taking an offshore environment factor as an interference factor, and coupling the time-space parameters of ocean turbulence intensity, wave load and salt spray corrosion rate; establishing a fan-marine environment digital twinborn body; operation data is acquired in real time, synchronous mapping and dynamic updating of the digital twin are realized, and a state prediction model is established in combination with historical operation characteristics; taking the maximum power generation efficiency and the minimum mechanical load as a multi-objective optimization function, and introducing a model prediction controller to obtain an optimal control sequence of the wind power generation system in a future time domain of each control period; and selecting a first control instruction in the optimal control sequence, issuing the first control instruction to a physical fan torque controller and a variable pitch system in real time, realizing rolling optimization of the wind power generation system, and obtaining a self-adaptive operation control scheme of the wind power generation system. The power generation benefit of the whole life cycle is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of operation control, in particular to an operation control method of a wind power generation system. BACKGROUND

[0002] The conventional wind power generation system control method, especially when applied to complex marine environment, usually has significant defects. The control strategy is mostly based on the steady-state model of the wind turbine itself, without fully considering the dynamic coupling interference of marine turbulence, wave load and salt mist corrosion and other multi-time and space scale environmental factors, resulting in insufficient model accuracy and control lag. Feedback control based on current or historical state is generally used, which lacks forward-looking prediction of future operating state and is difficult to actively suppress mechanical load while ensuring power generation efficiency. The control parameters are fixed and cannot be adjusted online according to the performance degradation of the wind turbine and the time-varying characteristics of the marine environment, resulting in high operation risk of the system in severe sea conditions and poor economic benefits throughout the life cycle, which seriously restricts the efficiency and safety of offshore wind power generation. SUMMARY

[0003] To solve the above technical problems, an operation control method of a wind power generation system is provided, which solves the above problems.

[0004] To achieve the above purpose, the technical scheme adopted by the present application is as follows: An operation control method of a wind power generation system, comprising: Obtaining historical operation data of the wind power generation system in the target sea area, taking marine environmental factors as external interference factors, coupling the time and space parameters of marine turbulence intensity, wave load and salt mist corrosion rate, and establishing a wind turbine-marine environment digital twin; Based on the sensor, real-time acquisition of the wind power generation system operation data is realized, the wind turbine-marine environment digital twin is synchronously mapped and dynamically updated, the wind power generation system state prediction model is established in combination with the historical operation characteristics of the wind power generation system; Based on the wind power generation system state prediction model, taking maximizing power generation efficiency and minimizing mechanical load as a multi-objective optimization function, a model predictive controller is cited to obtain the optimal control sequence of the wind power generation system in the future time domain of each control period; Selecting the first control instruction in the optimal control sequence of the wind power generation system in the future time domain of each control period, and real-time issuing it to the physical wind turbine torque controller and the variable pitch system to realize rolling optimization of the wind power generation system and obtain the control scheme of adaptive operation of the wind power generation system.

[0005] Preferably, obtaining historical operation data of the wind power generation system in the target sea area, taking marine environmental factors as external interference factors, coupling the time and space parameters of marine turbulence intensity, wave load and salt mist corrosion rate, and establishing a wind turbine-marine environment digital twin specifically comprises: Based on the historical SCADA system of the wind farm in the target sea area, acquire historical operating data of the wind power generation system; The operational data includes: wind speed, wind turbine speed, power generation, blade pitch angle, and vibration acceleration; Acquire marine meteorological buoy data for the target sea area, including marine meteorological data and marine hydrological data, to obtain marine environmental data; The marine environmental data includes: wind speed / direction, wave height / period / direction, seawater salinity, air temperature and humidity, and turbulence intensity at 10 meters above the sea surface; Based on historical operational data of wind power generation systems, marine meteorological data, and marine hydrological data, the data is unified to the same timestamp, preprocessed, and denoised using a low-pass filter to obtain a historical dataset of wind power generation systems. Using wind speed / direction and turbulence intensity at 10 meters above sea level as input, the aerodynamic characteristics of wind turbine blades are simulated using blade element momentum theory. The average wind speed at hub height at 10 meters above sea level is calculated using the logarithmic wind profile formula. For a monopile tower of an offshore wind turbine, the coordinates of each blade element relative to the tower are calculated in the rotor rotation plane. The potential flow theory is used to correct the model and obtain a time-varying wind speed component related to the azimuth angle. Using turbulence intensity as a random input, a random wind speed field that conforms to the characteristics of high ocean turbulence is generated using the Kaimal spectrum.

[0006] Preferably, acquiring historical operational data of the wind power generation system in the target sea area, using marine environmental factors as external disturbance factors, and coupling the spatiotemporal parameters of ocean turbulence intensity, wave load, and salt spray corrosion rate to establish a wind turbine-marine environment digital twin also includes: The random wind speed field that conforms to the characteristics of high ocean turbulence is discretized into N frequency components. The phase angle corresponding to each frequency component is randomly generated. The time series of transverse and vertical pulsating wind speeds on the wind turbine sweep surface are synthesized to establish a dynamic, three-dimensional and marine environment-corrected equivalent wind speed field. Based on the dynamic, three-dimensional and marine environment-corrected equivalent wind speed field, combined with the wind turbine blade geometric parameters, airfoil aerodynamic data and wind turbine parameters, the wind turbine is discretized into M blade elements along the radial direction. Using the blade element momentum theory, the axial and circumferential induction factors of each blade element are iteratively solved, and the lift and drag of each blade element are calculated. Based on the lift and drag of each blade element, calculate the aerodynamic torque and thrust of the entire wind turbine. Based on the aerodynamic torque and thrust of the entire wind turbine, the yaw moment and overturning moment of the entire wind turbine are obtained by integration using the azimuth distribution as a compensation factor. Using marine salt spray corrosion as an aerodynamic performance degradation factor, we corrected the airfoil aerodynamic data, updated the aerodynamic loads on each blade element, and established an aerodynamic sub-model. Using the aerodynamic load on each blade element as the excitation function, the vibration response of the wind turbine tower is calculated using the wind turbine structural dynamics theory, and the structural safety of the wind turbine in the marine environment is evaluated. For monopile foundations of offshore wind turbines, the Morrison equations are used to calculate the drag force and inertial force caused by waves. The wave force / moment is calculated as an external excitation function and applied to the corresponding position of the tower in the aerodynamic sub-model. The wave load is calculated, the wind turbine structural dynamic model is established, and the natural frequency and mode shape of the wind turbine are obtained. Using seawater salinity and air temperature and humidity as inputs, and employing the principles of materials science and electrochemistry, a corrosion kinetic model is established with the change in component corrosion depth over time as the target. Based on the change of component corrosion depth over time, the finite element model of the wind turbine tower is substituted into it, and the corresponding component material properties, stiffness matrix and strength parameters are dynamically updated to establish a corrosion degradation sub-model. By integrating aerodynamic sub-models, wind turbine structural dynamics models, and corrosion degradation sub-models, a digital twin of the wind turbine and marine environment is established.

[0007] Preferably, based on sensors, real-time acquisition of wind power system operation data is achieved to realize synchronous mapping and dynamic updating of the wind turbine-marine environment digital twin. Combined with the historical operating characteristics of the wind power system, a wind power system state prediction model is established, specifically including: Based on sensors, real-time operational data of the wind power generation system is acquired and preprocessed. Using real-time operating data of the wind power generation system as input, the parameters of the wind turbine-marine environment digital twin are updated synchronously. The parameters include: wind turbine speed, blade pitch angle, and generator torque; Based on the meteorological forecasting system, short-term forecast data of the target sea area is obtained, including the sea area wind speed, wind direction and wave parameters for a period of time in the future, and a short-term meteorological forecast sequence for the sea area is generated. Taking the real-time operation of the wind power generation system as the initial state, and combining the planned execution of the future control sequence of the wind turbine controller with the short-term forecast sequence of marine meteorology, the wind turbine-marine environment digital twin is substituted into the forward dynamics simulation to deduce the evolution trajectory of the wind power generation system's operating state and control strategy over a period of time, and to establish a physical prediction model of the wind power generation system. The future control sequence includes: generator torque setpoint and pitch angle setpoint.

[0008] Preferably, based on sensors, real-time acquisition of wind power system operation data is used to achieve synchronous mapping and dynamic updating of the wind turbine-marine environment digital twin. Combined with historical operating characteristics of the wind power system, the establishment of a wind power system state prediction model also includes: Based on the historical dataset of wind power generation system, historical operating data at h time steps are selected as training samples. The goal is to minimize the residual between the actual historical operating data and the predicted value of the physical prediction model of wind power generation system, and to train the data-driven prediction model of wind power generation system. By integrating the physical prediction model and the data-driven prediction model of the wind power system, and using the residual correction method, the error between the prediction results of the integrated wind power system operation data and the actual operation data of the wind power system is calculated and corrected, and a state prediction model of the wind power system is established. If the predictive performance of the data-driven prediction model for the wind power system is continuously declining, the model will be automatically retrained and dynamically updated using the latest wind power system operating data. If a systematic deviation is detected, the physical prediction model of the wind power generation system will be dynamically adjusted.

[0009] Preferably, based on the wind power generation system state prediction model, using maximizing power generation efficiency and minimizing mechanical load as multi-objective optimization functions, and employing a model predictive controller, the optimal control sequence of the wind power generation system in the future time domain for each control cycle specifically includes: Based on the wind turbine design parameters, the wind speed operating range is divided. If it is below the rated wind speed range, the primary goal is to capture the maximum wind energy and the secondary goal is to smooth the mechanical load. If it is above the rated wind speed range, the primary goal is to maintain the rated power control and minimize the mechanical load. Based on real-time operating data of the wind power generation system, a digital twin of the wind turbine and marine environment is simulated to analyze the coupling relationship between power generation and mechanical load under the current operating conditions and dynamically adjust the weight coefficients of power generation and mechanical load. Based on the weighting coefficients of power generation and mechanical load, the system state prediction model of wind power generation is substituted into the model for forward verification. If the verification result meets the preset lower limit of power generation efficiency and upper limit of mechanical load, the current weighting coefficient is determined to be an effective weight. If the verification result does not meet the preset lower limit of power generation efficiency and upper limit of mechanical load, the power generation efficiency and mechanical load coefficients are readjusted and substituted into the verification again until an effective weight that meets the requirements is obtained.

[0010] Preferably, based on the wind power generation system state prediction model, using maximizing power generation efficiency and minimizing mechanical load as multi-objective optimization functions, and employing a model predictive controller, the optimal control sequence of the wind power generation system in the future time domain for each control cycle further includes: Using effective weights as input, a weighted multi-objective optimization function is constructed to maximize power generation efficiency and minimize mechanical load; The control input constraints are the range of pitch angle, the limit of pitch angle change rate, the range of generator torque and the limit of generator torque change rate; the state constraints are the wind turbine rotor speed and the engine output power. Design a model predictive controller based on a multi-objective optimization function, combining control input constraints and state constraints, and discretize the future time domain for each control cycle; For each discrete control time step, the wind power system state prediction model is used to predict the system operating state over a period of time in the future, and the wind turbine speed, power generation efficiency and mechanical load parameters are obtained. Based on the prediction of the system's operating status over a future period, a weighted multi-objective optimization function that maximizes power generation efficiency and minimizes mechanical load is calculated, resulting in the generator torque setpoint sequence and the optimal paddle moment angle setpoint sequence within this time domain. By integrating the time-domain generator torque setpoint sequence with the optimal propeller torque angle setpoint sequence, the optimal control sequence for the wind power generation system in the future time domain is generated for each control cycle.

[0011] Preferably, the first control command in the optimal control sequence of the wind power generation system in the future time domain of each control cycle is selected and sent to the physical wind turbine torque controller and pitch system in real time to achieve rolling optimization of the wind power generation system. The resulting control scheme for adaptive operation of the wind power generation system specifically includes: The first control instruction in the optimal control sequence of the wind power generation system in the future time domain of each control cycle is selected as the execution instruction of the current control cycle, and the generator torque setpoint and the propeller pitch angle setpoint are extracted at the corresponding time. The generator torque setpoint and pitch angle setpoint are sent to the wind turbine execution system in real time via fieldbus. For the torque controller, the generator torque setpoint is received in real time and the generator electromagnetic torque is adjusted to ensure that the wind turbine power output meets expectations. For the pitch servo controller, the pitch angle setpoint is received in real time and the wind turbine blade rotation angle is controlled to ensure that the blade is in the predetermined position.

[0012] Preferably, the first control command in the optimal control sequence of the wind power generation system in the future time domain of each control cycle is selected and sent to the physical wind turbine torque controller and pitch system in real time to achieve rolling optimization of the wind power generation system. The control scheme for adaptive operation of the wind power generation system also includes: Based on the data sent to the wind turbine execution system, the actual operating data of the wind power generation system is collected synchronously and compared with the wind turbine-marine environment digital twin to calculate the deviation between the two in real time. Set a preset threshold; The preset threshold includes: performance deviation threshold. 3%, load safety deviation threshold 8%, Actuator deviation threshold: Pitch angle 0.5°, torque deviation 2%; If the deviation between the two is less than the preset threshold, the current control command continues to be executed. If the deviation is greater than or equal to the preset threshold, the control strategy replanning mechanism is triggered, the wind power system state prediction model is called again, and the optimal control sequence of the wind power system in the future time domain is regenerated by combining the actual operation data of the wind power system and the marine environment data. The first control command is selected and sent to the wind turbine execution system to realize the rolling optimization of the wind power system and obtain the control scheme for the adaptive operation of the wind power system.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes an operation control scheme for a wind power generation system. This scheme constructs a digital twin that integrates multiple environmental disturbance factors such as ocean turbulence, wave loads, and salt spray corrosion, and dynamically updates it based on real-time data. This achieves high-precision mapping and forward-looking prediction of the wind turbine's operating status in complex marine environments. A model predictive control method is employed, with rolling optimization aimed at improving power generation efficiency and reducing mechanical loads, ensuring the system always executes the optimal control commands in the future time domain. This enhances the wind power generation system's adaptability, control accuracy, and operational reliability in harsh marine environments. While effectively ensuring unit safety, it maximizes power generation benefits throughout the entire lifecycle. This method increases the average annual power generation of the wind turbine by 3.7%, while reducing fatigue loads on key components by 12.4%, verifying its effectiveness and reliability in complex marine environments. Attached Figure Description

[0014] Figure 1 This is a flowchart of an operation control method for a wind power generation system. Detailed Implementation

[0015] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0016] Reference Figure 1 As shown, a method for operating and controlling a wind power generation system includes: S1. Obtain historical operating data of wind power generation systems in the target sea area, use marine environmental factors as external interference factors, couple the spatiotemporal parameters of ocean turbulence intensity, wave load and salt spray corrosion rate, and establish a digital twin of wind turbine and marine environment. Step S1 specifically includes: Based on the historical SCADA system of the wind farm in the target sea area, acquire historical operating data of the wind power generation system; The operational data includes: wind speed, wind turbine speed, power generation, blade pitch angle, and vibration acceleration; Acquire marine meteorological buoy data for the target sea area, including marine meteorological data and marine hydrological data, to obtain marine environmental data; The marine environmental data includes: wind speed / direction, wave height / period / direction, seawater salinity, air temperature and humidity, and turbulence intensity at 10 meters above the sea surface; Based on historical operational data of wind power generation systems, marine meteorological data, and marine hydrological data, the data is unified to the same timestamp, preprocessed, and denoised using a low-pass filter to obtain a historical dataset of wind power generation systems. Using wind speed / direction and turbulence intensity at 10 meters above sea level as input, the aerodynamic characteristics of wind turbine blades are simulated using blade element momentum theory. The average wind speed at hub height at 10 meters above sea level is calculated using the logarithmic wind profile formula. For a monopile tower of an offshore wind turbine, the coordinates of each blade element relative to the tower are calculated in the rotor rotation plane. The potential flow theory is used to correct the model and obtain a time-varying wind speed component related to the azimuth angle. Using turbulence intensity as a random input, a random wind speed field that conforms to the characteristics of high ocean turbulence is generated using the Kaimal spectrum.

[0017] Step S1 also includes: The random wind speed field that conforms to the characteristics of high ocean turbulence is discretized into N frequency components. The phase angle corresponding to each frequency component is randomly generated. The time series of transverse and vertical pulsating wind speeds on the wind turbine sweep surface are synthesized to establish a dynamic, three-dimensional and marine environment-corrected equivalent wind speed field. Based on the dynamic, three-dimensional and marine environment-corrected equivalent wind speed field, combined with the wind turbine blade geometric parameters, airfoil aerodynamic data and wind turbine parameters, the wind turbine is discretized into M blade elements along the radial direction. Using the blade element momentum theory, the axial and circumferential induction factors of each blade element are iteratively solved, and the lift and drag of each blade element are calculated. Based on the lift and drag of each blade element, calculate the aerodynamic torque and thrust of the entire wind turbine. Based on the aerodynamic torque and thrust of the entire wind turbine, the yaw moment and overturning moment of the entire wind turbine are obtained by integration using the azimuth distribution as a compensation factor. Using marine salt spray corrosion as an aerodynamic performance degradation factor, we corrected the airfoil aerodynamic data, updated the aerodynamic loads on each blade element, and established an aerodynamic sub-model. Using the aerodynamic load on each blade element as the excitation function, the vibration response of the wind turbine tower is calculated using the wind turbine structural dynamics theory, and the structural safety of the wind turbine in the marine environment is evaluated. For monopile foundations of offshore wind turbines, the Morrison equations are used to calculate the drag force and inertial force caused by waves. The wave force / moment is calculated as an external excitation function and applied to the corresponding position of the tower in the aerodynamic sub-model. The wave load is calculated, the wind turbine structural dynamic model is established, and the natural frequency and mode shape of the wind turbine are obtained. Using seawater salinity and air temperature and humidity as inputs, and employing the principles of materials science and electrochemistry, a corrosion kinetic model is established with the change in component corrosion depth over time as the target. Based on the change of component corrosion depth over time, the finite element model of the wind turbine tower is substituted into it, and the corresponding component material properties, stiffness matrix and strength parameters are dynamically updated to establish a corrosion degradation sub-model. By integrating aerodynamic sub-models, wind turbine structural dynamics models, and corrosion degradation sub-models, a digital twin of the wind turbine and marine environment is established.

[0018] When using it, please refer to the steps outlined above: In existing technologies, modeling the operating status of offshore wind turbines typically neglects the multi-physics coupling effect of the marine environment and fails to comprehensively consider the spatiotemporal correlation of multiple factors such as high-turbulence wind fields, wave dynamic loads, and salt spray corrosion degradation, resulting in insufficient model accuracy and life prediction deviations. This step constructs a digital twin of the wind turbine and marine environment by coupling the spatiotemporal parameters of ocean turbulence, wave loads, and salt spray corrosion, realizing multi-disciplinary dynamic coupling modeling of aerodynamics, structure, and corrosion, thereby improving the simulation accuracy of wind turbine loads, the reliability of structural safety assessment, and the accuracy of life prediction in the marine environment.

[0019] S2. Based on sensors, real-time operation data of the wind power generation system is acquired to realize the synchronous mapping and dynamic updating of the wind turbine-ocean environment digital twin. Combined with the historical operation characteristics of the wind power generation system, a wind power generation system status prediction model is established. Step S2 specifically includes: Based on sensors, real-time operational data of the wind power generation system is acquired and preprocessed. Using real-time operating data of the wind power generation system as input, the parameters of the wind turbine-marine environment digital twin are updated synchronously. The parameters include: wind turbine speed, blade pitch angle, and generator torque; Based on the meteorological forecasting system, short-term forecast data of the target sea area is obtained, including the sea area wind speed, wind direction and wave parameters for a period of time in the future, and a short-term meteorological forecast sequence for the sea area is generated. Taking the real-time operation of the wind power generation system as the initial state, and combining the planned execution of the future control sequence of the wind turbine controller with the short-term forecast sequence of marine meteorology, the wind turbine-marine environment digital twin is substituted into the forward dynamics simulation to deduce the evolution trajectory of the wind power generation system's operating state and control strategy over a period of time, and to establish a physical prediction model of the wind power generation system. The future control sequence includes: generator torque setpoint and pitch angle setpoint.

[0020] Step S2 also includes: Based on the historical dataset of wind power generation system, historical operating data at h time steps are selected as training samples. The goal is to minimize the residual between the actual historical operating data and the predicted value of the physical prediction model of wind power generation system, and to train the data-driven prediction model of wind power generation system. By integrating the physical prediction model and the data-driven prediction model of the wind power system, and using the residual correction method, the error between the prediction results of the integrated wind power system operation data and the actual operation data of the wind power system is calculated and corrected, and a state prediction model of the wind power system is established. If the predictive performance of the data-driven prediction model for the wind power system is continuously declining, the model will be automatically retrained and dynamically updated using the latest wind power system operating data. If a systematic deviation is detected, the physical prediction model of the wind power generation system will be dynamically adjusted.

[0021] When using it, please refer to the steps outlined above: In existing technologies, wind power system state prediction typically relies on a single physical model or data-driven method. Physical models struggle to accurately simulate the dynamic changes in complex marine environments, while purely data-driven models lack extrapolation capabilities due to neglecting physical mechanisms, resulting in insufficient prediction accuracy and an inability to adapt to sudden environmental changes and equipment performance degradation. This step, by integrating physical and data-driven models and combining real-time data with weather forecasts, constructs a dynamically updated digital twin, improving the accuracy of short-term state predictions. Furthermore, it can adapt to system changes through residual correction and model self-updating mechanisms, enhancing the real-time perception and forward-looking control capabilities of wind turbine operating status in complex marine environments.

[0022] S3. Based on the wind power generation system state prediction model, with maximizing power generation efficiency and minimizing mechanical load as the multi-objective optimization function, the model predictive controller is used to obtain the optimal control sequence of the wind power generation system in the future time domain for each control cycle. Step S3 specifically includes: Based on the wind turbine design parameters, the wind speed operating range is divided. If it is below the rated wind speed range, the primary goal is to capture the maximum wind energy and the secondary goal is to smooth the mechanical load. If it is above the rated wind speed range, the primary goal is to maintain the rated power control and minimize the mechanical load. Based on real-time operating data of the wind power generation system, a digital twin of the wind turbine and marine environment is simulated to analyze the coupling relationship between power generation and mechanical load under the current operating conditions and dynamically adjust the weight coefficients of power generation and mechanical load. Based on the weighting coefficients of power generation and mechanical load, the system state prediction model of wind power generation is substituted into the model for forward verification. If the verification result meets the preset lower limit of power generation efficiency and upper limit of mechanical load, the current weighting coefficient is determined to be an effective weight. If the verification result does not meet the preset lower limit of power generation efficiency and upper limit of mechanical load, the power generation efficiency and mechanical load coefficients are readjusted and substituted into the verification again until an effective weight that meets the requirements is obtained.

[0023] Step S3 also includes: Using effective weights as input, a weighted multi-objective optimization function is constructed to maximize power generation efficiency and minimize mechanical load; The control input constraints are the range of pitch angle, the limit of pitch angle change rate, the range of generator torque and the limit of generator torque change rate; the state constraints are the wind turbine rotor speed and the engine output power. Design a model predictive controller based on a multi-objective optimization function, combining control input constraints and state constraints, and discretize the future time domain for each control cycle; For each discrete control time step, the wind power system state prediction model is used to predict the system operating state over a period of time in the future, and the wind turbine speed, power generation efficiency and mechanical load parameters are obtained. Based on the prediction of the system's operating status over a future period, a weighted multi-objective optimization function that maximizes power generation efficiency and minimizes mechanical load is calculated, resulting in the generator torque setpoint sequence and the optimal paddle moment angle setpoint sequence within this time domain. By integrating the time-domain generator torque setpoint sequence with the optimal propeller torque angle setpoint sequence, the optimal control sequence for the wind power generation system in the future time domain is generated for each control cycle.

[0024] When using it, please refer to the steps outlined above: In existing technologies, wind power generation system control strategies typically employ fixed-weight single-objective optimization or static multi-objective optimization methods, which struggle to dynamically adapt to complex and ever-changing offshore wind conditions. This results in insufficient balancing ability between power generation efficiency and mechanical load, and a lack of forward verification of the predictive model. Consequently, the control strategy is prone to failing to meet power generation targets or exceeding mechanical load limits under extreme operating conditions. This approach introduces a digital twin-driven dynamic weight adjustment mechanism and a model predictive control framework, enabling real-time optimization of the weight allocation between power generation efficiency and mechanical load. Based on forward verification, it ensures that the control sequence simultaneously meets the lower limit of power generation efficiency and the upper limit of mechanical load, thereby improving wind energy capture efficiency while reducing structural fatigue and achieving adaptive and precise control under multi-objective collaborative optimization.

[0025] S4. Select the first control instruction in the optimal control sequence of the wind power generation system in the future time domain of each control cycle, and send it to the physical wind turbine torque controller and pitch system in real time to realize the rolling optimization of the wind power generation system and obtain the control scheme for the adaptive operation of the wind power generation system. Step S4 specifically includes: The first control instruction in the optimal control sequence of the wind power generation system in the future time domain of each control cycle is selected as the execution instruction of the current control cycle, and the generator torque setpoint and the propeller pitch angle setpoint are extracted at the corresponding time. The generator torque setpoint and pitch angle setpoint are sent to the wind turbine execution system in real time via fieldbus. For the torque controller, the generator torque setpoint is received in real time and the generator electromagnetic torque is adjusted to ensure that the wind turbine power output meets expectations. For the pitch servo controller, the pitch angle setpoint is received in real time and the wind turbine blade rotation angle is controlled to ensure that the blade is in the predetermined position.

[0026] Step S4 also includes: Based on the data sent to the wind turbine execution system, the actual operating data of the wind power generation system is collected synchronously and compared with the wind turbine-marine environment digital twin to calculate the deviation between the two in real time. Set a preset threshold; The preset threshold includes: performance deviation threshold. 3%, load safety deviation threshold 8%, Actuator deviation threshold: Pitch angle 0.5°, torque deviation 2%; If the deviation between the two is less than the preset threshold, the current control command continues to be executed. If the deviation is greater than or equal to the preset threshold, the control strategy replanning mechanism is triggered, the wind power system state prediction model is called again, and the optimal control sequence of the wind power system in the future time domain is regenerated by combining the actual operation data of the wind power system and the marine environment data. The first control command is selected and sent to the wind turbine execution system to realize the rolling optimization of the wind power system and obtain the control scheme for the adaptive operation of the wind power system.

[0027] When using it, please refer to the steps outlined above: Traditional wind power generation control often employs controllers designed based on fixed parameters or single operating conditions, lacking foresight and adaptability to complex and time-varying marine environments. When encountering rapid changes in wind, waves, and currents, such control strategies exhibit lag in response, making it difficult to dynamically optimize the overall system performance. The absence of a real-time feedback verification mechanism easily leads to a disconnect between control commands and actual operating conditions, resulting in decreased unit operational stability and power generation efficiency losses. This step, by implementing a rolling optimization mechanism based on model predictive control, ensures that control commands are always based on the optimal sequence in the future time domain, significantly improving the system's forward-looking control and dynamic response capabilities. The introduction of real-time comparison and feedback correction between digital twins and actual operating data forms a closed-loop adaptive optimization of prediction-execution-verification-adjustment, effectively overcoming model mismatch and environmental disturbances, and ensuring the accuracy, adaptability, and operational reliability of the unit's control in complex marine environments.

[0028] Based on the above, the specific implementation method is as follows: Taking a 50MW offshore wind farm in my country as an example, this operation and control method was deployed. First, the historical SCADA data of the wind farm over the past three years and the monitoring data of marine buoys in the same area were retrieved. Through data alignment and filtering and noise reduction, a high-quality training dataset of more than TB was formed. Using blade element momentum theory, the 158-meter diameter wind turbine was discretized into 200 blade elements. Combined with NREL S series airfoil aerodynamic data, a high-precision aerodynamic model was constructed. Targeting the unique high turbulence characteristics of this sea area, with a turbulence intensity of 18%, a three-dimensional unsteady wind speed field containing 50 frequency components was generated using the Kaimal spectrum. In terms of structural modeling, a tower-foundation finite element model containing 12,000 elements was established based on the ANSYS platform. The JONSWAP spectrum wave load was calculated using the Morrison equation, with a maximum wave height of 8.2 meters. Based on the measured salinity data of the sea area, which is 3.2-3.8%, a corrosion rate prediction model for the bottom of the tower was established, predicting a maximum corrosion depth of 2.8 mm over a 20-year service life. The integrated digital twin was verified by measured data, showing a power generation prediction error of less than 2.1% and a tower bending moment prediction error of less than 4.7% under rated operating conditions. In actual operation, the system collects operational data in real time through 128 sensors deployed in various parts of the wind turbine, including vibration, strain, and corrosion probes. It receives a detailed wind and wave forecast for the next 6 hours from the regional meteorological forecast center every 5 minutes, with a time resolution of 10 minutes. The digital twin synchronizes the state based on the current real-time pitch angle of 8.7 and the current generator torque of 2850 kN·m, and simulates the system state for the next 30 minutes. A data-driven prediction model is constructed using an LSTM neural network to predict the generator speed and output power for the next 10 minutes using historical data from the past 60 minutes and 15 time steps. During a strong wind event in October 2023, in which the wind speed suddenly increased from 8 m / s to 14 m / s, the physical model prediction showed a systematic bias, and the power prediction error continued to exceed 5%. The system automatically activated the model update mechanism and retrained the LSTM network online based on the latest 200 sets of running data, so that the average prediction error in the following 30 minutes was restored to within 2.3%. At the control level, based on the real-time wind speed (current value 12.5 m / s, higher than the rated wind speed), the system determines the operating range as the constant power zone. Through digital twin simulation analysis, the weights of the multi-objective function are dynamically adjusted: the power generation efficiency weight is set to 0.6, and the mechanical load weight is set to 0.4. The model predictive controller uses a 15-second control time domain, discretized into 30 control steps. Within 0.5 seconds of each control cycle, it solves the optimization problem based on the state prediction results, considering constraints such as pitch angle change rate limit ≤8° / s and torque change rate limit ≤15% / s. During a certain transitional operating condition, the controller successfully generated the optimal control sequence for the next 15 seconds. The first command was a pitch angle of 10.2° and a generator torque of 3120 kN·m. Forward verification confirmed that the sequence could ensure that the power generation was maintained within the range of 5.0 ± 0.15 MW, while controlling the tower base bending moment to below 82% of the design limit. The system sends control commands to the ABB torque controller and Moog pitch system in real time via the PROFIBUS bus. During execution, it continuously monitors the deviation between the actual power of 5.02MW and the digital twin prediction of 5.08MW. When extreme waves were encountered in November 2023, with an effective wave height of 4.8 meters, the measured tower vibration acceleration of 2.3m / s² exceeded the prediction value by 18%, triggering the control replanning mechanism. The system immediately re-optimized based on the updated environmental parameters and adjusted the pitch angle to 9.8° within two control cycles, successfully reducing the vibration level to 1.9m / s².

[0029] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. An operation control method of a wind power generation system for an offshore wind power generation system, characterized by, To solve the coupling interference problem of ocean turbulence, wave load and salt mist corrosion, the method comprises: Obtain the historical operation data of the wind power system in the target sea area, take the offshore environmental factors as external interference factors, couple the spatial and temporal parameters of ocean turbulence intensity, wave load and salt mist corrosion rate, and establish a wind turbine-ocean environment digital twin; Based on the sensor, the operation data of the wind power system is obtained in real time, the synchronization mapping and dynamic updating of the wind turbine-ocean environment digital twin are realized, the historical operation characteristics of the wind power system are combined to establish a wind power system state prediction model; Based on the wind power system state prediction model, taking the maximum power generation efficiency and the minimum mechanical load as the multi-objective optimization function, referring to the model predictive controller, the optimal control sequence of the wind power system in each control period in the future time domain is obtained; Select the first control instruction in the optimal control sequence of the wind power system in each control period in the future time domain, and real-time issue to the physical wind turbine torque controller and the variable pitch system to realize the rolling optimization of the wind power system and obtain the adaptive operation control scheme of the wind power system.

2. The method of claim 1, wherein Obtain the historical operation data of the wind power system in the target sea area, take the offshore environmental factors as external interference factors, couple the spatial and temporal parameters of ocean turbulence intensity, wave load and salt mist corrosion rate, and establish a wind turbine-ocean environment digital twin, which specifically comprises: Based on the historical SCADA system of the target sea area wind farm, the historical operation data of the wind power system is obtained; The operation data includes: wind speed, wind turbine speed, power generation power, pitch angle, vibration acceleration; Obtain the ocean meteorological buoy data of the target sea area, including ocean meteorological data and ocean hydrological data, to obtain ocean environmental data; The ocean environmental data includes: wind speed / wind direction at 10 meters above sea level, wave height / period / direction, seawater salinity, air temperature and humidity, turbulence intensity; Based on the historical operation data of the wind power system, the ocean meteorological data and the ocean hydrological data, unify to the same time stamp, perform data preprocessing, use a low-pass filter to denoise the data, and obtain a wind power system historical data set; Taking the wind speed / wind direction at 10 meters above sea level and the turbulence intensity as input, using the blade element momentum theory, simulating the aerodynamic characteristics of the wind turbine blade, using the logarithmic wind profile formula, calculating the wind speed at the hub height of 10 meters above sea level, obtaining the average wind speed at the hub height; For offshore wind turbine single pile tower, calculate the coordinates of each blade element position relative to the tower in the wind wheel rotation plane, and use the potential flow theory correction model to obtain a time-varying wind speed component related to the azimuth angle; Taking the turbulence intensity as a random input item, using Kaimal spectrum, generating a random wind speed field conforming to the characteristics of high ocean turbulence.

3. A method of operating control of a wind power generation system according to claim 2, characterized by, Obtain the historical operation data of the wind power system in the target sea area, take the offshore environmental factors as external interference factors, couple the spatial and temporal parameters of ocean turbulence intensity, wave load and salt mist corrosion rate, and establish a wind turbine-ocean environment digital twin, which further comprises: Discretize the random wind speed field with high turbulence characteristics of the sea into N frequency components, randomly generate the corresponding phase angle of each frequency component, synthesize the time series of transverse and vertical fluctuating wind speed on the wind wheel sweep surface, and establish a dynamic, three-dimensional and equivalent wind speed field modified by the sea environment; Based on the dynamic, three-dimensional and equivalent wind speed field modified by the sea environment, combined with the geometric parameters of the wind turbine blade, the airfoil aerodynamic data and the wind wheel parameters, the wind wheel is discretized into M blades along the radial direction. Using the blade element momentum theory, the axial and circumferential induction factors of each blade element are iteratively solved, and the lift and drag of each blade element are calculated. Based on the lift and drag of each blade element, the aerodynamic torque and thrust of the entire wind wheel are calculated. Based on the aerodynamic torque and thrust of the entire wind wheel, the azimuth angle distribution is taken as the compensation factor, and the integral is obtained to obtain the yaw moment and overturning moment of the entire wind wheel. Taking the sea salt spray corrosion as the aerodynamic performance degradation factor, the airfoil aerodynamic data is modified, and the aerodynamic load on each blade element is updated to establish an aerodynamics sub-model. Taking the aerodynamic load on each blade element as the excitation function, the wind turbine tower vibration response is calculated using the wind turbine structural dynamics theory, and the structural safety of the wind turbine in the marine environment is evaluated. For offshore wind turbine single pile foundation, the drag force and inertia force caused by waves are calculated using the Morison equation. The wave force / torque is calculated as the external excitation function, which is applied to the corresponding position of the tower in the aerodynamics sub-model to calculate the wave load and establish the wind turbine structural dynamics model to obtain the natural frequency and mode shape of the wind turbine. Taking seawater salinity and air temperature and humidity as inputs, using material science and electrochemistry principles, taking the change of component corrosion depth over time as the target, a corrosion kinetics model is established. Based on the change of component corrosion depth over time, the corresponding component material properties, stiffness matrix and strength parameters are dynamically updated by substituting into the wind turbine tower finite element model to establish a corrosion degradation sub-model. Integrate the aerodynamics sub-model, wind turbine structural dynamics model and corrosion degradation sub-model to establish a wind turbine-marine environment digital twin.

4. The method of claim 1, wherein Based on the sensor, the real-time operation data of the wind power system is obtained to realize the synchronous mapping and dynamic updating of the wind turbine-marine environment digital twin, and the wind power system state prediction model is established based on the historical operation characteristics of the wind power system, which includes: Based on the sensor, the real-time operation data of the wind power system is obtained to realize the synchronous mapping and dynamic updating of the wind turbine-marine environment digital twin, and the wind power system state prediction model is established based on the historical operation characteristics of the wind power system, which includes: The parameters include: wind turbine speed, pitch angle and generator torque; Based on the weather forecast system, the short-term prediction data of the target sea area is obtained, including the wind speed, wind direction and wave parameters of the sea area in the future, and the sea area weather short-term prediction sequence is generated; Taking the real-time operation of the wind power system as the initial state, combining the wind turbine group controller planned execution of the future control sequence and the sea area weather short-term prediction sequence, substituting into the wind turbine-marine environment digital twin for forward dynamics simulation, deducing the wind power system running state and control strategy evolution trajectory in the future, and establishing a wind power system physical prediction model. ​ The future control sequence includes a generator torque set value and a pitch angle set value.

5. A method of operating control of a wind power generation system according to claim 4, characterized by, Based on the sensors, real-time wind power system operation data is acquired to realize synchronous mapping and dynamic updating of the wind turbine-ocean environment digital twin, and a wind power system state prediction model is established by combining historical operation characteristics of the wind power system, which further includes: Based on the historical data set of the wind power system, historical operation data of h time steps are selected as training samples, and a wind power system data-driven prediction model is trained to minimize the residual error between the actual historical operation data and the predicted value of the wind power system physical prediction model. The wind power system physical prediction model and the wind power system data-driven prediction model are fused, and the residual error correction method is used to calculate the error between the fusion wind power system operation data prediction result and the actual wind power system operation data to correct the error, and a wind power system state prediction model is established. If the prediction performance of the wind power system data-driven prediction model is continuously decreasing, the model retraining is automatically triggered, and the latest wind power system operation data is used to dynamically update the model. If a systematic deviation occurs, the wind power system physical prediction model is dynamically adjusted.

6. A method of operating control of a wind power generation system according to claim 5, characterized by, Based on the wind power system state prediction model, the maximum power generation efficiency and the minimum mechanical load are taken as the multi-objective optimization function, and the model predictive controller is referenced to obtain the optimal control sequence of the wind power system in the future time domain of each control period, which further includes: Based on the wind turbine design parameters, the wind speed operating interval is divided, and if it is below the rated wind speed region, the maximum wind energy capture is taken as the main target, and the smooth mechanical load is taken as the secondary target, and if it is above the rated wind speed region, the rated power control and the minimum mechanical load are taken as the main target. Based on the real-time operation data of the wind power system, the wind turbine-ocean environment digital twin is driven to simulate, and the coupling relationship between power generation and mechanical load under the current working condition is analyzed, and the power generation and mechanical load weight coefficients are dynamically adjusted. Based on the power generation and mechanical load weight coefficients, the wind power system state prediction model is verified, and if the verification result meets the preset lower limit of power generation efficiency and upper limit of mechanical load, the current weight coefficient is determined as the effective weight, and if the verification result does not meet the preset lower limit of power generation efficiency and upper limit of mechanical load, the power generation efficiency and mechanical load coefficients are adjusted again to verify until the effective weight that meets the requirements is obtained.

7. A method of operating control of a wind power system according to claim 6, characterized in that, Based on the wind power system state prediction model, the maximum power generation efficiency and the minimum mechanical load are taken as the multi-objective optimization function, and the model predictive controller is referenced to obtain the optimal control sequence of the wind power system in the future time domain of each control period, which further includes: The effective weight is taken as the input to construct the weighted multi-objective optimization function of maximizing the power generation efficiency and minimizing the mechanical load. The pitch angle range, pitch angle change rate limit, generator torque range and generator torque change rate limit are taken as the control input constraint conditions, and the wind turbine rotor speed and engine output power are taken as the state constraint conditions. The model predictive controller is designed, the multi-objective optimization function is combined with the control input constraint conditions and the state constraint conditions, and the future time domain is discretized for each control period. For each discrete control time step, the wind power system state prediction model is used to predict the system operating state in the future time period, obtaining the wind turbine speed, power generation efficiency and mechanical load parameters; Based on the prediction of the system operating state in the future time period, a weighted multi-objective optimization function is calculated to maximize power generation efficiency and minimize mechanical load, obtaining the generator torque set value sequence and the optimal pitch angle set value sequence in the time domain; Integrate the generator torque set value sequence and the optimal pitch angle set value sequence in the time domain to generate the optimal control sequence of the wind power system in the future time domain of each control period.

8. The method of claim 7, wherein, Select the first control instruction in the optimal control sequence of the wind power system in the future time domain of each control period, and real-time issue it to the physical wind turbine torque controller and the variable pitch system to realize the rolling optimization of the wind power system, and obtain the adaptive operation control scheme of the wind power system, which specifically includes: Select the first control instruction in the optimal control sequence of the wind power system in the future time domain of each control period as the current control period execution instruction, and extract the generator torque set value and the pitch angle set value at the corresponding time; Through the field bus, the generator torque set value and the pitch angle set value are real-time issued to the wind turbine execution system, the generator torque set value is real-time received for the torque controller to adjust the electromagnetic torque of the generator, ensuring that the wind turbine power output meets the expectation, and the pitch angle set value is real-time received for the variable pitch servo controller to control the wind turbine blade rotation angle, ensuring that the blade is in the predetermined position.

9. A method of operating control of a wind power system according to claim 8, characterized in that, Select the first control instruction in the optimal control sequence of the wind power system in the future time domain of each control period, and real-time issue it to the physical wind turbine torque controller and the variable pitch system to realize the rolling optimization of the wind power system, and obtain the adaptive operation control scheme of the wind power system, which further includes: Based on the issue to the wind turbine execution system, the actual operation data of the wind power system is synchronously collected and compared with the wind turbine-ocean environment digital twin, and the deviation between the two is real-time calculated; Set a preset threshold; The preset threshold comprises: a performance deviation threshold 3%, a load safety deviation threshold 8%, an actuator deviation threshold: pitch angle 0.5°, torque deviation 2%; If the deviation between the two is less than the preset threshold, the current control instruction is continued to be executed, and if the deviation between the two is greater than or equal to the preset threshold, the control strategy re-planning mechanism is triggered, the wind power system state prediction model is re-called, the actual operation data of the wind power system and the ocean environment data are combined to re-generate the optimal control sequence of the wind power system in the future time domain, and the first control instruction is issued to the wind turbine execution system to realize the rolling optimization of the wind power system and obtain the adaptive operation control scheme of the wind power system.

Images