Wind turbine generator self-adaptive protection method based on high-precision storm surge numerical forecasting

Abstract

The invention relates to the technical field of data processing, and discloses a wind turbine generator self-adaptive protection method based on high-precision storm surge numerical forecasting. The method comprises the steps that a dynamic resolution forecasting system is established by improving a spherical coordinate system Arakawa C-type grid, high-precision storm surge forecasting is generated by fusing data of a Jason-3 satellite and a tide station and adopting an ensemble Kalman filtering algorithm, a storm surge intensity-fan action mapping table is established based on forecasting data, edge computing nodes execute millisecond-level protection control, and a high-precision storm surge model is established. The multi-parameter sensor evaluates the protection effect, and protection strategy parameters are adaptively optimized according to the evaluation result. The technical problems that in an existing wind turbine generator protection technology, storm surge forecasting precision is insufficient, a protection decision lacks dynamic adaptability, and a protection strategy cannot be optimized autonomously are solved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting. Background Technology

[0002] Existing offshore wind turbine protection technologies primarily employ passive protection methods based on fixed thresholds. These methods trigger turbine shutdown or propeller retraction actions using preset parameters such as wind speed and wave height. Protection decisions rely on static thresholds set by human experience, lacking the ability to dynamically respond to complex marine environments. Traditional storm surge forecasting systems use uniform grid resolution for numerical calculations, which cannot balance computational efficiency with forecast accuracy in complex nearshore terrain areas. Furthermore, they often rely on single observation data sources for forecasting, and their data assimilation methods are relatively simple, making it difficult to fully utilize multi-source observation information to improve forecast accuracy.

[0003] However, existing technologies have significant shortcomings: First, fixed threshold protection strategies cannot adapt to storm surge characteristics of different intensities and development stages, leading to inappropriate timing of protective measures activation, either being too conservative and affecting power generation efficiency, or having a delayed response that threatens equipment safety; second, traditional forecasting systems have insufficient spatial resolution, especially in nearshore shallow water areas where they cannot accurately capture the propagation and evolution of storm surges, resulting in limited forecast accuracy; third, existing protection systems lack real-time effect evaluation and strategy optimization mechanisms, making it impossible to adaptively adjust based on the actual implementation effect of protective measures, thus hindering continuous improvement of protection strategies.

[0004] Based on the above analysis, the core technical problem faced by existing technologies is: how to establish a high-precision storm surge numerical forecasting system and achieve intelligent linkage with wind turbine protection measures, improve forecast accuracy through multi-source data fusion, establish a dynamic threshold decision-making mechanism to achieve precise triggering of protection measures, and at the same time build a real-time evaluation system for protection effect and a strategy adaptive optimization system to solve the technical bottlenecks of traditional protection technologies such as slow response, insufficient accuracy and lack of self-learning ability. Summary of the Invention

[0005] This application provides an adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting, which addresses the technical problems of insufficient storm surge forecast accuracy, lack of dynamic adaptability in protection decisions, and inability to autonomously optimize protection strategies in existing wind turbine protection technologies.

[0006] Firstly, this application provides an adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting, the adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting comprising:

[0007] By improving the Arakawa C-type grid in spherical coordinate system, the storm surge propagation area in the sea area is dynamically divided into resolutions, resulting in a dynamic grid system with a 200-meter grid near the coast and a 1000-meter grid in the open sea.

[0008] Based on Jason-3 satellite altimeter data and coastal tide gauge station observation data, the multi-source sea surface height data in the dynamic grid system is fused and assimilated using an ensemble Kalman filter algorithm to obtain high-precision storm surge water level forecast data.

[0009] The high-precision storm surge water level forecast data is compared with the preset water level threshold. When the water level reaches 1.5 meters, a yaw control command is generated, and when the water level reaches 3.0 meters, a blade retraction control command is generated, thus obtaining a storm surge intensity-wind turbine action mapping table.

[0010] The edge computing nodes perform millisecond-level execution processing on the yaw control commands and blade pitch control commands in the storm surge intensity-wind turbine action mapping table to obtain real-time control actions for wind turbine yaw angle adjustment and blade pitch angle adjustment.

[0011] By using multi-parameter sensors to monitor and process the tower stress, blade load, and equipment vibration data of the wind turbine after the execution of the real-time control actions, a comprehensive evaluation result of the effectiveness of the protective measures is obtained.

[0012] Based on the comprehensive evaluation results, the water level threshold parameters in the storm surge intensity-wind turbine operation mapping table are adaptively adjusted to obtain optimized protection strategy parameters.

[0013] The technical solution provided in this application dynamically divides the storm surge propagation area by improving the Arakawa C-type grid in spherical coordinates. The differentiated settings of a 200-meter grid near the coast and a 1000-meter grid in the open sea ensure both computational accuracy in complex terrain areas and control of overall computational costs, significantly improving forecast efficiency and accuracy compared to traditional uniform grid methods. By fusing Jason-3 satellite altimeter data and coastal tide gauge observation data using an ensemble Kalman filter algorithm, the problem of insufficient information from a single data source is effectively solved. Multi-source data assimilation significantly improves the reliability and timeliness of storm surge level forecasts. The storm surge intensity-wind turbine action mapping table established based on high-precision forecast data achieves precise matching between protective measures and storm surge development stages. The graded protection strategy of triggering yaw control with a 1.5-meter water level threshold and triggering blade retraction with a 3.0-meter water level threshold avoids the blindness of traditional fixed threshold methods. The millisecond-level response processing capability of edge computing nodes ensures the real-time execution of protection commands, significantly shortens the response time from forecast to protection action, and improves the survivability of wind turbines in sudden storm surge environments.

[0014] In the field of adaptive protection applications for wind turbines based on high-precision storm surge numerical forecasting, the core contribution of the ensemble Kalman filter algorithm lies in its ability to effectively handle the spatiotemporal uncertainties of marine observation data. Through parameter estimation and error covariance calculation of 50 ensemble members, it systematically quantifies and reduces forecast errors, providing a more reliable data foundation for subsequent protection decisions. The multi-parameter fusion comprehensive evaluation algorithm, through a dynamic weight allocation mechanism, adaptively adjusts the importance weights of tower stress and blade load according to different stages of storm surge development, achieving an objective quantitative evaluation of protection effectiveness and overcoming the limitations of traditional methods that rely on a single indicator. The application value of genetic algorithms in protection strategy optimization lies in their global search capability and adaptive learning characteristics. Through in-depth mining of historical successful cases and parameter sensitivity analysis, it continuously optimizes water level threshold settings, enabling protection strategies to continuously improve with the accumulation of operational experience, forming an intelligent protection system with self-learning capabilities. This fundamentally improves the safety and economy of offshore wind turbines in complex marine environments. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of an embodiment of the adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting in this application. Detailed Implementation

[0017] This application provides an adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0018] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting in this application includes:

[0019] Step S101: Dynamically divide the storm surge propagation area of ​​the sea area by improving the Arakawa C-type grid in spherical coordinate system to obtain a dynamic grid system with a 200-meter grid near the shore and a 1000-meter grid in the open sea.

[0020] Step S102: Based on Jason-3 satellite altimeter data and coastal tide gauge station observation data, the multi-source sea surface height data in the dynamic grid system is fused and assimilated using an ensemble Kalman filter algorithm to obtain high-precision storm surge water level forecast data;

[0021] Step S103: Compare the high-precision storm surge water level forecast data with the preset water level threshold. When the water level reaches 1.5 meters, generate a yaw control command. When the water level reaches 3.0 meters, generate a blade retraction control command to obtain a storm surge intensity-wind turbine action mapping table.

[0022] Step S104: The edge computing node performs millisecond-level execution processing on the yaw control command and blade pitch control command in the storm surge intensity-wind turbine action mapping table to obtain the real-time control actions of wind turbine yaw angle adjustment and blade pitch angle adjustment.

[0023] Step S105: Real-time monitoring and processing of wind turbine tower stress, blade load, and equipment vibration data after the execution of the real-time control action is carried out using multi-parameter sensors to obtain a comprehensive evaluation result of the effectiveness of the protective measures.

[0024] Step S106: Based on the comprehensive evaluation results, the water level threshold parameters in the storm surge intensity-wind turbine operation mapping table are adaptively adjusted to obtain optimized protection strategy parameters.

[0025] It is understood that the implementing entity of this application can be a wind turbine adaptive protection system based on high-precision storm surge numerical forecasting, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0026] Specifically, the improved spherical coordinate system Arakawa C-type grid is a numerical grid technique that spatially discretizes the ocean area according to the water depth gradient. By performing spherical coordinate transformation on the ocean topography and water depth data, the original bathymetry data is converted into a three-dimensional water depth distribution matrix in latitude and longitude coordinates. Then, based on the water depth distribution matrix, areas with drastic changes in water depth gradient are identified as the boundary between the nearshore and offshore areas. In the complex nearshore topography areas, 200-meter high-resolution grid cells are set to capture topographic details, while 1000-meter grid cells are set in the deep water areas of the offshore to reduce the computational load. The numerical continuity between areas of different resolutions is ensured through Arakawa C-type grid interpolation connection processing.

[0027] The ensemble Kalman filter algorithm is a data assimilation method based on Bayesian theory. First, it corrects the orbital error and atmospheric delay of the raw data from the Jason-3 satellite altimeter to eliminate satellite orbital deviations and ionospheric influences, thus obtaining accurate sea surface height anomaly data. Simultaneously, it separates the astronomical tide component from the coastal tide gauge observation data through tidal harmonic analysis and extracts the pure storm surge observation signal. Then, it establishes an observation operator matrix to achieve spatial interpolation matching between satellite grid point data and tide gauge station data. The forecast error covariance is estimated through Monte Carlo sampling of 50 ensemble members. Finally, it fuses multi-source observation information to correct the sea surface height forecast value within the dynamic grid system.

[0028] High-precision storm surge water level forecast data is input into the threshold judgment module. The water level change rate is calculated through a continuous 3-hour time window. When the water level rise rate is detected to exceed 0.3 meters per hour, the early warning mechanism is activated. When the forecast water level reaches 1.5 meters, the system automatically generates a yaw control command to drive the wind turbine to turn towards the wave angle. When the forecast water level reaches 3.0 meters, a blade retraction control command is generated to adjust the blade pitch angle to 80% retraction state. A storm surge intensity-wind turbine action mapping table is established through the mapping relationship between water level thresholds and equipment actions.

[0029] After receiving the control commands from the mapping table, the edge computing node parses and calculates the yaw control commands to obtain the target angle parameters of the yaw system. It then sends angle control signals to the yaw motor through the fieldbus network to adjust the wind turbine orientation. For blade retraction control commands, the processor calculates the target pitch angle parameters corresponding to the 80% retraction state and completes the blade angle adjustment within 500 milliseconds through a hydraulic or electric pitch actuator. The entire control process uses a real-time operating system to ensure that the response latency is controlled at the millisecond level.

[0030] By collecting equipment operating status data through strain sensors and acceleration sensors deployed in the tower, blade roots, and nacelle, the measured values ​​of the tower stress sensors are compared in real time with the 75% safety threshold of the design strength, and the data of the blade root bending moment sensors are compared and analyzed with the 85% threshold of the rated bending moment. Based on the storm surge development stage, dynamic weighting coefficients are set for parameters such as water level, wind speed, and wave height. The equipment safety margin value is obtained by weighted averaging, and the safety margin is converted into a comprehensive evaluation result of 0-100 points using a normalized scoring algorithm.

[0031] The comprehensive evaluation results are input into the historical case database for feature matching. Similar historical events are identified through parameters such as storm surge path, intensity, and development trend. Threshold setting experience from successful protection cases is extracted. Parameter sensitivity analysis is performed on the 1.5-meter and 3.0-meter water level thresholds to calculate the influence weight of each threshold on the protection effect. The optimal threshold combination is searched through selection, crossover, and mutation operations using a genetic algorithm. The new threshold parameters obtained from the optimization are then updated to the storm surge intensity-wind turbine operation mapping table to replace the original settings.

[0032] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0033] The marine topographic and water depth data were transformed into spherical coordinates to obtain the water depth distribution matrix in latitude and longitude coordinates.

[0034] Based on the water depth distribution matrix, the boundary between the nearshore area and the offshore area of ​​the sea area is divided to obtain the regional boundary line of water depth gradient change.

[0035] Based on the aforementioned regional boundary line, a 200-meter grid density is set for the nearshore area to obtain high-resolution grid cells;

[0036] The high-resolution grid cells are connected to the 1000-meter grid in the outer sea area using Arakawa C-type grid interpolation to obtain the dynamic grid system.

[0037] Specifically, the spherical coordinate system transformation of marine topographic and water depth data involves converting the three-dimensional coordinates (x, y, z) in the Cartesian coordinate system collected by the original depth sounder into latitude and longitude representations in the spherical coordinate system. The planar coordinates are then converted into geographic coordinates represented by longitude λ and latitude φ using coordinate transformation formulas, while retaining the depth information z as the third dimension. This results in a two-dimensional matrix structure with latitude and longitude as indices and water depth as values. Each element D(λ, φ) in the matrix represents the seabed depth value at the corresponding latitude and longitude location.

[0038] The water depth distribution matrix is ​​used to delineate the boundaries between the nearshore and offshore areas. By calculating the rate of change of water depth gradient between adjacent grid points, areas of abrupt topographic changes are identified. The gradient operator is used to perform spatial differentiation on the water depth matrix to calculate the rate of change of water depth at each grid point in the longitude and latitude directions. When the gradient value exceeds a preset threshold, it is marked as a region of abrupt topographic change. Connecting these abrupt change points forms a continuous isogradability line as the boundary between the nearshore and offshore areas. The area inside the boundary line is the nearshore area with complex topography, and the area outside the boundary line is the offshore area with relatively gentle topography.

[0039] The 200-meter grid density setting for nearshore areas based on regional boundaries involves establishing uniformly distributed square grids within the nearshore area defined by the boundary line. Each grid cell has a side length of 200 meters. The required number of grids is calculated based on the geographical extent of the nearshore area. Grid node coordinates are generated along the longitude and latitude directions at 200-meter intervals. Each grid cell contains four vertices and one center point. The center point carries the average water depth information of the area, forming a high-resolution grid cell array that covers the entire complex nearshore terrain area.

[0040] Arakawa C-type grid interpolation connection between high-resolution grid cells and 1000-meter grids in the offshore area establishes a transition buffer zone at the boundary between the nearshore 200-meter grid and the offshore 1000-meter grid. The Arakawa C-type grid places scalar variables such as water level at the grid center and vector variables such as flow velocity components at the midpoints of the grid boundaries. The physical quantity values ​​at any point in the transition zone are calculated using a bilinear interpolation algorithm. The interpolation weight is determined based on the distance from the point to the center of the adjacent grid, ensuring numerical continuity and computational stability between grids of different resolutions. Ultimately, a dynamic grid system with high precision nearshore and moderate precision offshore is formed, balancing computational accuracy and efficiency requirements.

[0041] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0042] The raw data from the Jason-3 satellite altimeter were processed with orbital error correction and atmospheric delay correction to obtain the corrected satellite sea level data.

[0043] The storm surge observation data after separating the astronomical tide was obtained by tidal harmonic analysis of the observation data from the coastal tide gauge stations.

[0044] Based on the corrected satellite sea surface height data and the storm surge observation data, an observation operator matrix is ​​established to obtain spatial interpolation weight coefficients.

[0045] The high-precision storm surge water level forecast data is obtained by performing parameter estimation on 50 ensemble members of the observation operator matrix using the ensemble Kalman filter algorithm.

[0046] Specifically, the Jason-3 satellite altimeter raw data undergoes orbit error correction and atmospheric delay correction. First, the impact of satellite orbit determination error on ranging accuracy needs to be eliminated. The deviation between the actual satellite position and the theoretical orbit position is corrected using GPS precise orbit determination data, and the true distance from the satellite to the sea surface is calculated. Then, atmospheric delay correction is performed, which includes two parts: ionospheric delay and tropospheric delay. Ionospheric delay is eliminated by dual-frequency ranging technology to eliminate the influence of electron density on radar signal propagation speed. Tropospheric delay is corrected by meteorological models to correct the delay effect of water vapor and dry air on signal propagation. After these two correction steps, accurate satellite sea surface height data is obtained, and the data accuracy is reduced from the original centimeter-level error to the millimeter level.

[0047] The tidal harmonic analysis of coastal tide gauge observation data uses the least squares method to decompose the observed total water level signal into multiple harmonic components of different frequencies, mainly including astronomical tidal components such as the semi-diurnal tide M2 ​​and S2 components and the diurnal tide K1 and O1 components. The amplitude and phase parameters of each harmonic component are identified by Fourier transform technology, and an astronomical tide forecasting model is established to calculate the theoretical astronomical tide level. The storm surge observation data is obtained by subtracting the astronomical tide level from the measured total water level. This processing eliminates the periodic tidal changes caused by the gravitational influence of the moon and the sun, and retains the non-periodic water level change signal caused by meteorological factors.

[0048] Based on corrected satellite sea surface height data and storm surge observation data, an observation operator matrix is ​​established. Spatial interpolation technology is used to map observation data from different locations onto a unified model grid. Each row of the observation operator matrix H corresponds to an observation point, and each column corresponds to a model grid point. The matrix element Hij represents the influence weight of the j-th grid point on the i-th observation point. The weight calculation adopts the inverse distance weighting method, where the weight of the grid points around the observation point is inversely proportional to the square of its distance. The weight coefficients are normalized to ensure that the sum of the weights of all grid points corresponding to the same observation point is equal to 1, forming a linear mapping relationship between the observation space and the model space.

[0049] The ensemble Kalman filter algorithm first generates 50 different initial state vectors as ensemble members, each representing a possible sea surface height distribution state. Random perturbations are added around the initial states using Monte Carlo sampling to generate the ensemble. In the forecasting step, each ensemble member independently runs a numerical model to obtain the forecast state. In the analysis step, the ensemble mean is calculated as the background field, and the covariance matrix among the ensemble members represents the forecast error. The model state is projected onto the observation space using the observation operator matrix and compared with the actual observation data. Each ensemble member is then corrected based on the Kalman gain matrix, ultimately yielding high-precision storm surge level forecast data after fusing observation information. The forecast accuracy is significantly improved compared to a single data source.

[0050] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0051] The high-precision storm surge water level forecast data is processed by calculating the rate of change over a continuous 3-hour time window to obtain the water level rise rate parameter.

[0052] The storm surge development trend is predicted by comparing the water level rise rate parameter with the 0.3 m / h rate threshold.

[0053] Based on the predicted results, the yaw control command is obtained by adjusting the wave angle of the yaw system triggered by the 1.5-meter water level threshold.

[0054] By associating the 3.0-meter water level threshold with the 80% blade retraction angle, the blade retraction control command and the storm surge intensity-wind turbine action mapping table are obtained.

[0055] Specifically, the high-precision storm surge water level forecast data is processed by calculating the rate of change over a continuous 3-hour time window. The sliding window technique is used to perform differential calculations on the time series data. The water level value H(t) at the current time t is compared with the water level value H(t-3) 3 hours ago at time t-3. The formula for calculating the rate of water level rise v is v = [H(t) - H(t-3)] / 3, where H(t) represents the forecast water level at the current time, H(t-3) represents the forecast water level 3 hours ago, and the denominator 3 represents the time interval of 3 hours. The unit of the calculated result v is meters per hour. A positive value indicates a rise in water level, and a negative value indicates a fall in water level. The water level change rate parameter sequence for each time moment is obtained through continuous calculation.

[0056] The storm surge development status is determined by comparing the water level rise rate parameter with a rate threshold of 0.3 meters per hour. When the calculated water level rise rate v is greater than 0.3 meters per hour, the storm surge is judged to be in a rapid development stage, and the system outputs a development trend prediction result of "intensification". When the water level rise rate v is less than 0.3 meters per hour but greater than 0, it is judged to be in a "slow rise" state. When v is less than 0, it is judged to be in a "receding" state. The prediction result serves as the trigger condition for the activation of subsequent protective measures. The 0.3 meters per hour threshold is set based on the statistical analysis of historical storm surge events and represents the critical rate at which the storm surge changes from ordinary sea state to dangerous sea state.

[0057] Based on the prediction results, the yaw system's wave-facing angle adjustment is triggered by the 1.5-meter water level threshold. The system makes a dual judgment based on the current absolute value and trend of the water level. When the predicted water level reaches 1.5 meters and the prediction result shows "strengthening" or "slowly rising", the system automatically generates a yaw control command. The command includes the target yaw angle and the execution time window. The yaw angle is calculated and determined based on the current wind direction and the main wave direction, so that the wind turbine head faces the incoming wave direction to form a wave-facing attitude, reducing the impact load of the lateral wave force on the tower and blades. The yaw control command is output in the form of a digital signal, including parameters such as angle value, rotation direction and execution priority.

[0058] A correlation is established between the 3.0-meter water level threshold and the 80% pitch angle to create a relationship between water level and blade pitch angle. When the predicted water level reaches 3.0 meters, the system determines it to be a high-risk state and requires blade pitch protection measures. The 80% pitch angle refers to adjusting the blade from its optimal pitch angle during normal operation to a position close to feathering. The specific angle value is determined according to the wind turbine model, usually between 70 and 85 degrees. Pitch angle reduction significantly reduces the blade's windward area and aerodynamic load. The blade pitch reduction control command includes parameters such as the target pitch angle, adjustment rate, and completion time limit. By systematically combining different water level thresholds with corresponding protective actions, a storm surge intensity-wind turbine action mapping table is formed, which includes yaw control commands and blade pitch reduction control commands. The mapping table is stored in the form of a lookup table to provide data support for subsequent rapid decision-making and execution.

[0059] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0060] The yaw control command is input into the ARM Cortex-A78 processor for instruction parsing and processing to obtain the target angle parameters of the yaw system.

[0061] Based on the target angle parameters, the yaw motor is controlled via a fieldbus network to obtain the yaw angle adjustment action.

[0062] The blade retraction control command is processed by pitch angle calculation to obtain the target pitch angle parameter for the 80% retraction state.

[0063] Based on the target pitch angle parameter, the blade pitch actuator is subjected to response control processing within 500 milliseconds to obtain the real-time control action for adjusting the blade pitch angle.

[0064] Specifically, the yaw control command is input to the ARM Cortex-A78 processor for instruction parsing and processing. The received digital instruction packets are decoded and analyzed using a multi-threaded parallel computing architecture. The processor first verifies the integrity and validity of the instruction by checking the packet header identifier and checksum. Then, it extracts key parameters from the instruction, including the target yaw azimuth angle, the current cabin orientation angle, the yaw speed limit, and the execution priority. The floating-point unit calculates the difference between the target angle and the current angle to determine the required yaw amplitude. Considering the mechanical limitations and safety constraints of the yaw system, the target angle is checked for boundaries. When the target angle exceeds the ±720 degree range, angle normalization is performed. Finally, standardized yaw system target angle parameters are output for use by subsequent execution modules.

[0065] Based on the target angle parameters, the yaw motor is controlled via a fieldbus network. The control commands are transmitted to the yaw driver using the CAN bus protocol. The fieldbus data frame contains fields such as device address, function code, target angle value, and execution time. After receiving the command, the yaw motor driver starts the servo control algorithm. The encoder feeds back the current actual angle position of the nacelle and compares it with the target angle in real time. The PID controller calculates the motor torque output based on the angle deviation, which drives the gear reduction mechanism to rotate the nacelle around the vertical axis of the tower. During yaw, the system continuously monitors the rate of change of angle and torque load to ensure smooth rotation. When the actual angle is close to the target angle, the motor output is gradually reduced to achieve precise positioning, complete the yaw angle adjustment action, and report the execution status to the upper control system.

[0066] The blade retraction control command calculates the pitch angle based on the current blade pitch angle and the target retraction state. Under normal operating conditions, the blade pitch angle typically varies within the range of 0-30 degrees to optimize power generation efficiency. In the 80% retraction state, the pitch angle needs to be adjusted to 75-85 degrees, close to the feathering position. The processor reads the real-time pitch angle sensor values ​​of the three blades and calculates the angle increment required for each blade to reach the target pitch angle. Considering the response characteristics of the hydraulic actuator and angular velocity limitations, the pitch adjustment trajectory is determined. A smooth angle change curve is generated through an interpolation algorithm to avoid abrupt shocks. The output is a control data packet containing the target pitch angle parameters of the three blades, providing precise angle commands to the pitch execution system.

[0067] The target pitch angle parameter enables the blade pitch actuator to respond and control within 500 milliseconds. The hydraulic servo system drives the pitch bearing to achieve rapid blade angle adjustment. After receiving the target angle command, the hydraulic pump station immediately starts to establish system pressure. The proportional valve adjusts the hydraulic oil flow according to the angle deviation to control the piston rod extension and retraction speed. The angle sensor provides real-time feedback on the current blade pitch angle to form a closed-loop control circuit. The controller adopts a composite control strategy of feedforward and feedback to compensate for system delay and disturbance effects. The hydraulic cylinder pushes the pitch bearing to rotate the blade around the longitudinal axis to change the angle of attack. The entire pitch adjustment process is completed within a 500-millisecond time constraint to ensure rapid response to storm surge threats. The three blades simultaneously perform pitch angle adjustment to reach 80% of the blade retraction state, significantly reducing the aerodynamic load and bending moment stress on the blade. The real-time control action of blade pitch angle adjustment effectively protects the structural safety of the wind turbine.

[0068] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0069] The tower stress sensor data is compared and processed with the 75% threshold of the design strength to obtain the tower safety status assessment parameters.

[0070] By comparing and analyzing the data from the bending moment sensor at the blade root with the 85% threshold of the rated bending moment, the blade load safety assessment parameters are obtained.

[0071] Based on the safety status assessment parameters and the load safety assessment parameters, dynamic weight allocation calculation is performed to obtain the equipment safety margin value.

[0072] The safety margin value of the equipment is processed using a 0-100 score algorithm to obtain a comprehensive evaluation result of the effectiveness of the protective measures.

[0073] Specifically, the tower stress sensor data is compared with the 75% design strength threshold. Strain gauge sensors installed at different heights on the tower collect structural stress information in real time. The sensors convert mechanical strain into voltage signals, which are amplified and filtered by the signal conditioning circuit before being input into the data acquisition system. The system continuously records the stress distribution of the tower under bending, torsion, and axial loads at a sampling frequency of 10 Hz. The measured stress value is compared with the 75% safety threshold of the tower material's yield strength. When the measured stress is less than 75% of the threshold, a safety status code of 1 is output, indicating normal operation. When the measured stress is between 75% of the threshold and 100% of the design strength, a code of 2 is output, indicating a warning is needed. When the measured stress exceeds 100% of the design strength, a code of 3 is output, indicating a dangerous state. By statistically analyzing the stress data over the past 15 minutes, statistical parameters such as the average value, peak value, and standard deviation are calculated. The safety status assessment parameters reflecting the structural integrity of the tower are obtained by comprehensively considering the stress level and fluctuation degree.

[0074] The data from the blade root bending moment sensor is compared and analyzed with the 85% threshold of the rated bending moment. A torque sensor installed at the hub connection at the blade root monitors the flapping and oscillation bending moments borne by the blade. Based on the strain measurement principle, the sensor converts the bending deformation at the blade root into an electrical signal proportional to the bending moment. The data acquisition system synchronously records the bending moment time series data of the three blades and performs coordinate transformation to eliminate the influence of blade rotation on the measurement. The measured bending moment of each blade is compared and analyzed with the 85% safety threshold of the rated bending moment of this type of wind turbine. The bending moment safety margin is calculated as 85% of the rated bending moment minus the measured bending moment and then divided by 85% of the rated bending moment to obtain the relative safety margin. When the safety margin is greater than 30%, it is rated as a safe state; when the margin is between 10% and 30%, it is a state of caution; and when the margin is less than 10%, it is a state of danger. The safety assessment parameters reflecting the blade load level are calculated by combining the bending moment data of the three blades and the safety margin.

[0075] Dynamic weight allocation calculations are performed based on safety status assessment parameters and load safety assessment parameters. The importance weights of tower stress and blade load are adjusted according to different stages of storm surge development. In the early stage of storm surge, when the wave impact on the tower is relatively small, the tower weight is set to 0.3 and the blade weight to 0.7. As the storm surge intensity increases, the wave load has a significant impact on the tower, so the tower weight is adjusted to 0.6 and the blade weight to 0.4. The weight allocation algorithm calculates the dynamic weight coefficient based on the current water level and wave height. The tower weight increases with the rise in water level, gradually increasing from the initial value of 0.3 to the maximum value of 0.6. The tower safety assessment value and the blade safety assessment value are synthesized by a weighted average method according to the dynamic weights. The tower assessment value multiplied by the tower weight and the blade assessment value multiplied by the blade weight are used to obtain the comprehensive safety index. Finally, the equipment safety margin value reflecting the overall structural safety status is output.

[0076] The equipment safety margin is scored from 0 to 100 using a piecewise linear mapping function. This converts the safety margin into a straightforward percentage score. The algorithm divides the safety margin into four intervals, each corresponding to a different scoring range. A safety margin greater than 0.8 (scoring 90-100 points) indicates completely safe equipment operation. A safety margin between 0.6 and 0.8 (scoring 70-90 points) indicates good equipment condition with no concerns. A safety margin between 0.3 and 0.6 (scoring 40-70 points) indicates close monitoring of the equipment. A safety margin less than 0.3 (scoring 0-40 points) indicates significant safety risk. The scoring uses linear interpolation to proportionally allocate scores within each interval, ensuring a continuous correlation between the score and the safety margin. A time decay factor is introduced to incorporate historical scores with weights into the current score, preventing drastic fluctuations. The final output is a comprehensive evaluation result of 0-100 points, directly reflecting the overall safety level of the wind turbine and the effectiveness of the protective measures.

[0077] In one specific embodiment, the process of executing step S106 may specifically include the following steps:

[0078] The comprehensive evaluation results are input into the historical case database for similarity matching to obtain successful protection experience data for similar storm surge events;

[0079] Based on the successful protection experience data, parameter sensitivity analysis was performed on the 1.5-meter and 3.0-meter water level thresholds to obtain the threshold influence weighting coefficients;

[0080] The optimal water level threshold combination parameters are obtained by iteratively optimizing the threshold influence weight coefficients using a genetic algorithm.

[0081] The optimal water level threshold combination parameters are updated to the storm surge intensity-wind turbine operation mapping table for strategy replacement processing to obtain the optimized protection strategy parameters.

[0082] Specifically, the comprehensive evaluation results are input into the historical case database for similarity matching. Feature vector extraction technology converts the key parameters of the current storm surge event into numerical feature vectors. These feature vectors contain information on dimensions such as storm surge path direction angle, maximum water level height, water level rise rate, duration, wave cycle, and wind speed. The historical case database stores all storm surge events that occurred in the past three years and the corresponding implementation results of protective measures. The Euclidean distance algorithm is used to calculate the similarity distance between the current event's feature vector and the historical case feature vectors. The smaller the distance value, the more similar the two events are. The system automatically filters historical cases with distance values ​​less than a preset threshold as similar events. Further, based on the protection effectiveness score, successful cases with scores higher than 80 are selected. Key information such as water level threshold settings, the timing of protective measure activation, and equipment response parameters used in these successful cases are extracted to form a dataset of successful protection experiences for similar storm surge events, providing a reference for subsequent optimization analysis.

[0083] Based on successful protection experience data, parameter sensitivity analysis was performed on the 1.5-meter and 3.0-meter water level thresholds. Statistical analysis was used to quantify the impact of different threshold settings on the protection effect. Historical successful cases were grouped and statistically analyzed according to the water level thresholds used. The number of cases and average protection effect scores for triggering yaw action with a 1.5-meter threshold and triggering propeller retraction action with a 3.0-meter threshold were calculated. Analysis of variance was used to test the significance of different threshold combinations on the protection effect. The sensitivity coefficients of threshold changes on equipment safety margin changes were calculated. The sensitivity coefficient of the 1.5-meter threshold reflects the impact of the yaw action initiation timing on tower stress control, while the sensitivity coefficient of the 3.0-meter threshold reflects the impact of propeller retraction action on blade load control. Regression analysis was used to establish a mathematical relationship model between threshold parameters and protection effect. The contribution weight of each threshold parameter to the overall protection effect was calculated, and the quantified threshold influence weight coefficients were obtained to provide objective function parameters for subsequent optimization algorithms.

[0084] An optimization problem aimed at maximizing protective effect is established by iteratively optimizing the threshold influence weight coefficients using a genetic algorithm. The 1.5-meter and 3.0-meter water level thresholds are used as optimization variables within a reasonable range. The genetic algorithm first randomly generates an initial population of 50 individuals, each representing a set of threshold parameter combinations. The expected protective effect of each individual is evaluated using a fitness function, calculated based on historical experience data and weight coefficients. A selection operation uses a roulette wheel method to choose individuals with higher fitness as parents. A crossover operation exchanges gene fragments between two parent individuals to generate new offspring. A mutation operation slightly perturbs the genes of offspring individuals to increase population diversity. Through multiple generations of evolutionary iterations, the fitness gradually increases. The algorithm terminates when the change in optimal fitness over 20 consecutive generations is less than a set convergence threshold, and the threshold combination corresponding to the individual with the highest fitness is output as the optimal water level threshold combination parameter.

[0085] The optimal water level threshold combination parameters are updated to the storm surge intensity-wind turbine action mapping table for strategy replacement. Through a parameter overwrite mechanism, the optimized threshold settings are applied to the actual protection system. First, the original threshold parameters in the current mapping table are backed up for rollback operations if necessary. Then, the new threshold parameters optimized by the genetic algorithm are written into the corresponding fields of the mapping table. The updated mapping table includes optimized 1.5-meter yaw trigger thresholds and 3.0-meter propeller retraction trigger thresholds, along with corresponding action parameter settings. The system automatically verifies the effectiveness and safety of the new parameters, ensuring that the parameter values ​​are within the range allowed by the equipment technical specifications. Simulation tests verify the protective effect of the new threshold settings in typical storm surge scenarios. When the simulation results meet the expected protection objectives, the new mapping table configuration is officially activated. Simultaneously, the system logs are updated to record the parameter change time and optimization basis. After completing the strategy replacement process, protection strategy parameters that integrate historical experience and intelligent optimization are obtained, achieving continuous improvement and adaptive adjustment capabilities of the protection strategy.

[0086] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting, characterized in that, The method includes: By improving the Arakawa C-type grid in spherical coordinate system, the storm surge propagation area in the sea area is dynamically divided into resolutions, resulting in a dynamic grid system with a 200-meter grid near the coast and a 1000-meter grid in the open sea. Based on Jason-3 satellite altimeter data and coastal tide gauge station observation data, the multi-source sea surface height data in the dynamic grid system is fused and assimilated using an ensemble Kalman filter algorithm to obtain high-precision storm surge water level forecast data. The high-precision storm surge water level forecast data is compared with the preset water level threshold. When the water level reaches 1.5 meters, a yaw control command is generated, and when the water level reaches 3.0 meters, a blade retraction control command is generated, thus obtaining a storm surge intensity-wind turbine action mapping table. The edge computing nodes perform millisecond-level execution processing on the yaw control commands and blade pitch control commands in the storm surge intensity-wind turbine action mapping table to obtain real-time control actions for wind turbine yaw angle adjustment and blade pitch angle adjustment. By using multi-parameter sensors to monitor and process the tower stress, blade load, and equipment vibration data of the wind turbine after the execution of the real-time control actions, a comprehensive evaluation result of the effectiveness of the protective measures is obtained. Based on the comprehensive evaluation results, the water level threshold parameters in the storm surge intensity-wind turbine operation mapping table are adaptively adjusted to obtain optimized protection strategy parameters.

2. The adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting as described in claim 1, characterized in that, The process involves dynamically resolving the storm surge propagation area in the marine region using an improved Arakawa C-type spherical coordinate system, resulting in a dynamic grid system with a 200-meter grid for nearshore areas and a 1000-meter grid for offshore areas. This system includes: The marine topographic and water depth data were transformed into spherical coordinates to obtain the water depth distribution matrix in latitude and longitude coordinates. Based on the water depth distribution matrix, the boundary between the nearshore area and the offshore area of ​​the sea area is divided to obtain the regional boundary line of water depth gradient change. Based on the aforementioned regional boundary line, a 200-meter grid density is set for the nearshore area to obtain high-resolution grid cells; The high-resolution grid cells are connected to the 1000-meter grid in the outer sea area using Arakawa C-type grid interpolation to obtain the dynamic grid system.

3. The adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting as described in claim 1, characterized in that, Based on Jason-3 satellite altimeter data and coastal tide gauge observation data, the multi-source sea surface height data within the dynamic grid system is fused and assimilated using an ensemble Kalman filter algorithm to obtain high-precision storm surge level forecast data, including: The raw data from the Jason-3 satellite altimeter were processed with orbital error correction and atmospheric delay correction to obtain the corrected satellite sea level data. The storm surge observation data after separating the astronomical tide was obtained by tidal harmonic analysis of the observation data from the coastal tide gauge stations. Based on the corrected satellite sea surface height data and the storm surge observation data, an observation operator matrix is ​​established to obtain spatial interpolation weight coefficients. The high-precision storm surge water level forecast data is obtained by performing parameter estimation on 50 ensemble members of the observation operator matrix using the ensemble Kalman filter algorithm.

4. The adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting as described in claim 1, characterized in that, The process involves comparing the high-precision storm surge water level forecast data with preset water level thresholds. When the water level reaches 1.5 meters, a yaw control command is generated; when the water level reaches 3.0 meters, a blade retraction control command is generated. This yields a storm surge intensity-wind turbine operation mapping table, including: The high-precision storm surge water level forecast data is processed by calculating the rate of change over a continuous 3-hour time window to obtain the water level rise rate parameter. The storm surge development trend is predicted by comparing the water level rise rate parameter with the 0.3 m / h rate threshold. Based on the predicted results, the yaw control command is obtained by adjusting the wave angle of the yaw system triggered by the 1.5-meter water level threshold. By associating the 3.0-meter water level threshold with the 80% blade retraction angle, the blade retraction control command and the storm surge intensity-wind turbine action mapping table are obtained.

5. The adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting according to claim 1, characterized in that, The edge computing nodes perform millisecond-level execution processing on the yaw control commands and blade pitch control commands in the storm surge intensity-wind turbine action mapping table to obtain real-time control actions for wind turbine yaw angle adjustment and blade pitch angle adjustment, including: The yaw control command is input into the ARM Cortex-A78 processor for instruction parsing and processing to obtain the target angle parameters of the yaw system. Based on the target angle parameters, the yaw motor is controlled via a fieldbus network to obtain the yaw angle adjustment action. The blade retraction control command is processed by pitch angle calculation to obtain the target pitch angle parameter for the 80% retraction state. Based on the target pitch angle parameter, the blade pitch actuator is subjected to response control processing within 500 milliseconds to obtain the real-time control action for adjusting the blade pitch angle.

6. The adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting as described in claim 1, characterized in that, The process involves real-time monitoring and processing of wind turbine tower stress, blade load, and equipment vibration data after the execution of the real-time control actions using multi-parameter sensors to obtain a comprehensive evaluation result of the effectiveness of the protective measures, including: The tower stress sensor data is compared and processed with the 75% threshold of the design strength to obtain the tower safety status assessment parameters. By comparing and analyzing the data from the bending moment sensor at the blade root with the 85% threshold of the rated bending moment, the blade load safety assessment parameters are obtained. Based on the safety status assessment parameters and the load safety assessment parameters, dynamic weight allocation calculation is performed to obtain the equipment safety margin value. The safety margin value of the equipment is processed using a 0-100 score algorithm to obtain a comprehensive evaluation result of the effectiveness of the protective measures.

7. The adaptive protection method for wind turbines based on high-precision storm surge numerical forecasting as described in claim 1, characterized in that, The optimized protection strategy parameters are obtained by adaptively adjusting the water level threshold parameters in the storm surge intensity-wind turbine operation mapping table based on the comprehensive evaluation results, including: The comprehensive evaluation results are input into the historical case database for similarity matching to obtain successful protection experience data for similar storm surge events; Based on the successful protection experience data, parameter sensitivity analysis was performed on the 1.5-meter and 3.0-meter water level thresholds to obtain the threshold influence weighting coefficients; The optimal water level threshold combination parameters are obtained by iteratively optimizing the threshold influence weight coefficients using a genetic algorithm. The optimal water level threshold combination parameters are updated to the storm surge intensity-wind turbine operation mapping table for strategy replacement processing to obtain the optimized protection strategy parameters.

Images