A wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling, medium and equipment

Abstract

The application discloses a wind turbine blade icing prediction method based on micro-terrain classification and LWC_MVD vertical distribution modeling, a medium and equipment, relates to the wind power generation technical field, and comprises the following steps: obtaining digital elevation data of a target wind farm, extracting terrain feature parameters, classifying micro-terrain of a position where a wind turbine is located, and obtaining corresponding micro-terrain types. The application realizes high-precision prediction of wind turbine blade icing by constructing a micro-terrain classification and cloud microphysical parameter vertical distribution coupling model, significantly reduces prediction error, simultaneously establishes a three-dimensional icing calculation method, outputs spanwise and circumferential distribution characteristics, provides a basis for differentiated deicing, constructs a complete physical chain to improve model reliability, realizes a prediction and control closed loop through a standardized data interface, and fuses multi-source meteorological and terrain data to improve system robustness and adaptability.

Description

Technical Field

[0001] This invention relates to the field of wind power generation technology, specifically to a method, medium, and equipment for predicting icing on wind turbine blades based on micro-topography classification and LWC_MVD vertical distribution modeling. Background Technology

[0002] Blade icing in wind turbines is a key technical challenge hindering the development of wind power in cold and humid regions. Blade icing alters the aerodynamic characteristics of the airfoil, leading to a 10%-40% decrease in power generation efficiency. In severe cases, it can cause safety issues such as blade mass imbalance, increased vibration, and even structural damage. With the increasing size of wind turbines, rotor diameters have reached 150-300 meters, significantly expanding the swept surface area of ​​the blades. This results in a distinctly non-uniform spatial distribution of icing conditions. Under these circumstances, traditional icing prediction methods based on single-point meteorological parameters are insufficient to meet the needs of refined operation and maintenance of wind turbines. Therefore, there is an urgent need to develop a high-precision icing prediction method that can reflect spatial differences.

[0003] The shortcomings of existing technologies: Existing wind turbine blade icing prediction technologies mainly suffer from the following shortcomings: First, micro-topographic meteorological models mostly rely on empirical statistical methods, modeling only based on conventional meteorological elements such as temperature, humidity, and wind speed. They lack descriptions of the vertical structure of key cloud microphysical parameters such as liquid water content (LWC) and median volume diameter (MVD) of droplets, and fail to establish the coupling relationship between micro-topographic types and their vertical distribution, making it difficult to reflect the differences in icing at different spanwise locations of the blade. Second, existing methods typically assume that meteorological parameters are uniformly distributed in the vertical direction, using only single-point data of hub height for prediction. This fails to characterize the significant meteorological gradient changes caused by large-scale height differences on the rotor swept surface, resulting in deviations in the prediction of icing location and thickness. Furthermore, existing technologies lack a complete physical chain from micro-topographic identification to three-dimensional icing morphology prediction. Icing models are mostly simplified and do not consider aerodynamic changes during blade rotation, making it difficult to output three-dimensional icing distribution. Finally, the prediction results are mostly single icing thickness or level, failing to provide a basis for zonal control of anti-icing and de-icing systems, leading to a disconnect between prediction results and actual control strategies.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method, medium, and equipment for predicting icing on wind turbine blades based on micro-topography classification and LWC_MVD vertical distribution modeling, so as to solve the problems in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting icing on wind turbine blades based on micro-topography classification and LWC_MVD vertical distribution modeling, comprising the following steps: Obtain digital elevation data of the target wind farm, extract terrain feature parameters, classify the micro-topography of the wind turbine location, and obtain the corresponding micro-topography type. For the obtained micro-topography type, the pre-established vertical distribution statistical model of liquid water content and median droplet volume diameter is called to generate the vertical profile of liquid water content and median droplet volume diameter at the corresponding location. The vertical distribution statistical model is used to characterize the nonlinear variation characteristics of liquid water content and median droplet volume diameter along the height direction under different micro-topography conditions. A three-dimensional computational grid covering the swept area of ​​the wind turbine rotor is constructed. The generated vertical profiles of liquid water content and median volume diameter of droplets are mapped to each node of the three-dimensional computational grid to obtain the meteorological parameter distribution corresponding to each node. Based on the meteorological parameters of each node obtained after mapping, the local icing rate of each node is calculated using the icing growth model, and the local icing rate is integrated based on the time dimension to obtain the three-dimensional icing thickness distribution on the blade surface. Post-processing is performed based on the obtained three-dimensional icing thickness distribution to extract the blade spanwise icing distribution features and classify the icing severity level, outputting the corresponding icing prediction results.

[0007] Preferably, the statistical model for the vertical distribution of liquid water content adopts the following functional form: ; The statistical model for the vertical distribution of the median droplet volume diameter adopts the following functional form: ; in, Indicates the height above the ground. Indicates temperature. Indicates relative humidity. Indicates peak liquid water content. Indicates peak height. Indicates the standard deviation of the vertical distribution. This represents the micro-topography adjustment function. Indicates near-surface reference value, , Represents the gradient coefficients. This indicates a micro-topography correction term.

[0008] Preferably, the temporal evolution processing is as follows: the temperature and humidity change trends for future periods output by the mesoscale numerical weather prediction model are obtained, climatological constraints are introduced, and the vertical profiles of liquid water content and median droplet volume diameter are dynamically updated to realize the spatiotemporal evolution of meteorological parameters within the forecast lead time.

[0009] Preferably, the three-dimensional computational mesh adopts a cylindrical coordinate system with the wind turbine hub as the origin. The radial range is set to Vertical range set to The angle range is set to ,in Indicates the blade length. Indicates the wheel hub height.

[0010] Preferably, the icing growth model adopts an improved form that includes rotational effects. The local icing rate calculation process includes correcting the collision rate based on the instantaneous angle of attack change caused by blade rotation, and correcting the centrifugal force effect based on the blade spanwise position function, thereby obtaining the icing growth characteristics corresponding to the spatial location.

[0011] Preferably, the output includes standardized interface data, which is used for calling the anti-icing and de-icing control system. The data content includes blade spanwise icing information and anti-icing and de-icing priority identifiers for each icing zone.

[0012] Preferably, the micro-topography classification process includes extracting slope factors, topographic curvature and elevation anomalies, and outputting classification results for topographic factors, such as uplift type, canyon wind corridor type, water vapor enhancement type, open and flat type and leeward slope type, through the combination of topographic factors and discrimination rules.

[0013] Preferably, the three-dimensional icing thickness distribution results are used to calculate the spanwise average icing thickness, maximum icing thickness, and icing non-uniformity index, and the spatial distribution characteristics of icing are obtained through circumferential statistics.

[0014] A wind turbine blade icing prediction device based on micro-topography classification and LWC / MVD vertical distribution modeling includes a micro-topography classification module, an LWC / MVD vertical distribution modeling module, a three-dimensional icing calculation grid module, a local icing rate calculation module, and a prediction result output module. The micro-topography classification module, based on the digital elevation model data of wind farms, extracts slope, elevation, topographic curvature, and valley and ridge line identification features. It uses a topographic factor combination discrimination algorithm to classify wind turbine locations into five micro-topography types: terrain-lifted type, canyon wind tunnel type, water vapor-enhanced type, open and flat type, and leeward slope type. The LWC / MVD vertical distribution modeling module, targeting different micro-topographic types, integrates historical radiosonde data, microwave radiometer observation data, cloud radar echo data, and mesoscale numerical weather prediction output data to construct a statistical model of the vertical profile of liquid water content and median droplet volume diameter, including the vertical distribution function of liquid water content. and the vertical distribution function of the median volume diameter of the droplet ; The 3D icing calculation mesh module is based on the hub height of the wind turbine. and blade length A three-dimensional computational mesh is established in a cylindrical coordinate system centered on the impeller hub. This mesh covers the entire swept surface of the impeller and has a height range of [missing information]. to The radial range covers the entire blade length, with a vertical resolution of 1-5 meters and an angular resolution of 1-5 degrees. The local icing rate calculation module maps the vertical profiles of liquid water content and the median volume diameter of droplets to each node of the three-dimensional computational grid. For each spatial position traversed during the blade rotation, the improved Makkonen icing model is used to calculate the local collision rate, freezing rate, and icing rate increase, thereby obtaining the icing thickness distribution at different spanwise positions and different rotation angles on the blade surface. The prediction result output module, based on the obtained icing thickness distribution, outputs the blade spanwise icing thickness curve, the maximum icing location, the icing mass distribution, and the icing evolution trend within the prediction time range, and generates an icing level warning signal, providing a standardized input interface for the anti-icing and de-icing control system.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention effectively overcomes the limitations of traditional methods that rely on single-point meteorological data at hub height by establishing a coupled model between micro-topography classification and the vertical distribution of liquid water content and median droplet volume diameter. This allows for a more detailed characterization of the icing environment at different heights on the rotor swept surface. Under complex terrain conditions, meteorological elements exhibit significant vertical differences; introducing vertical profile modeling more realistically reflects actual icing conditions. Verification using actual wind farm operation data shows that the root mean square error of icing thickness prediction is reduced by 35%-50% compared to traditional methods, significantly improving prediction accuracy and reliability, and providing more accurate data support for the safe operation of wind turbines.

[0016] This invention enables refined prediction of the three-dimensional ice thickness distribution on blade surfaces, overcoming the limitations of traditional methods that only output a single ice thickness or simple level. By constructing a three-dimensional computational grid and combining it with the blade's rotational motion characteristics, it can output multi-dimensional structured information including spanwise ice thickness curves and circumferential non-uniformity indices, thus comprehensively reflecting the spatial distribution characteristics of ice. This multi-dimensional output provides a direct basis for differentiated anti-icing and de-icing strategies, enabling maintenance personnel to take targeted measures for different areas, improving anti-icing and de-icing efficiency and reducing energy consumption.

[0017] This invention constructs a complete physical mechanism chain from terrain identification to icing prediction, covering key links such as terrain identification, vertical structure analysis of cloud microphysical parameters, three-dimensional icing calculation, and standardized output. Each link has a clear physical causal relationship. By introducing a method combining cloud microphysical parameter modeling with an aerodynamic icing model, the extrapolation failure problem of traditional empirical statistical models under complex environmental conditions is avoided, thereby enhancing the physical rationality and applicability of the model and making the prediction results more stable and reliable.

[0018] This invention employs a standardized data structure in its output format, converting prediction results into JSON format data packets that include partition priority identification information. This allows for direct interface integration with the API of anti-icing and de-icing control systems, achieving seamless connection between prediction results and control execution. This approach enables the construction of a closed-loop system of prediction-decision-execution, ensuring that icing prediction not only remains at the analysis level but also directly guides the implementation of control strategies, thereby improving the intelligence level and response efficiency of wind turbine operation and management.

[0019] This invention constructs a multi-dimensional meteorological data support system by integrating multi-source information such as digital elevation model data, radiosonde data, microwave radiometer observation data, cloud radar data, and outputs from mesoscale numerical weather prediction models. Based on this, by combining physical constraints and statistical modeling methods, meteorological parameters are uniformly described and dynamically corrected, effectively improving the model's adaptability and stability under different environmental conditions. This multi-source data fusion mechanism not only enhances the robustness of the prediction system but also improves its generalization ability under complex climatic and topographical conditions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0021] Figure 1This is a flowchart of a wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling, according to the present invention.

[0022] Figure 2 This is a schematic diagram of a wind turbine blade icing prediction device based on micro-topography classification and LWC_MVD vertical distribution modeling according to the present invention. Detailed Implementation

[0023] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0024] This invention provides, for example Figure 1 The wind turbine blade icing prediction method shown includes the following steps: (The method is based on micro-topography classification and LWC_MVD vertical distribution modeling.) In the process of wind farm micro-topography identification, the first step is to acquire high-precision digital elevation model (DEM) data of the target area, with a spatial resolution of no less than 30 meters, to ensure the accuracy of topographic feature representation. After acquiring the raw DEM data, to avoid interference from topographic noise in subsequent calculations, a median filtering method is used to preprocess the DEM, effectively removing isolated outliers and local noise, and improving the stability of topographic derivative calculations. After completing the data preprocessing, key topographic factors are extracted from each DEM grid point to build the basis for subsequent micro-topography classification.

[0025] The calculation of topographic factors includes three core parameters: slope factor, topographic curvature, and elevation anomalies. The formula for calculating the slope factor is: in, It indicates the slope angle, reflecting the degree of inclination of the ground surface; and These represent DEM at direction and The first derivative of the direction is the rate of change of slope of the terrain in two horizontal orthogonal directions. The larger the value, the higher the terrain. The steeper the slope in the direction, The larger the value, the higher the terrain. The steeper the slope in a given direction, the square root of the sum of the squares of the two factors reflects the overall slope intensity. After transformation by the arctangent function, the slope angle is obtained.

[0026] Topographic curvature is used to describe the unevenness of the earth's surface, and its calculation formula is as follows: in, Indicates the curvature of the terrain; and DEM in direction and The second derivative of the direction is used to describe the curvature of the terrain in the corresponding direction; The mixed second derivative represents the terrain in... and The coupling changes in direction. The numerator reflects the combined effect of the curvature characteristics of the terrain surface, while the denominator normalizes the slope, thus ensuring scale consistency in the curvature results. Negative curvature values ​​typically indicate concave terrain (such as valleys), while positive curvature values ​​typically indicate convex terrain (such as ridges).

[0027] Elevation anomalies are used to reflect the degree of elevation deviation of a location relative to the overall topography of the area. The calculation formula is as follows: in, Indicates elevation anomalies; This represents the actual elevation of the current grid point; This represents the average elevation of the entire wind farm area. This parameter is used to identify local highlands or low-lying areas and is an important basis for distinguishing between uplifted terrain and leeward areas.

[0028] After calculating the aforementioned terrain factors, a terrain factor combination discrimination algorithm is used to classify the micro-topography of wind turbine locations. This classification method is based on the threshold combination relationship of multiple terrain factors to achieve fine identification of different terrain environments, specifically including the following five types: The first category is the land uplift type, and its identification criteria are: in, This represents the standard deviation of elevation, used to measure the degree of elevation dispersion within a region. When the elevation of a point is significantly higher than the average level and the slope is relatively gentle, it indicates that the location is in a relatively elevated plateau region.

[0029] The second category is the canyon wind tunnel type, and its identification criteria are: This type relies on curvature and valley structure density for identification, where It indicates obvious concave terrain features, while valley line density reflects the degree of development of valley structure in the region. When the density exceeds a set threshold, it indicates that the region has typical wind channel effect characteristics.

[0030] The third category is the water vapor growth type, and its discrimination criteria are: This type of data takes into account both spatial location and meteorological statistical characteristics. The distance to nearby water bodies represents the spatial distance between the aircraft location and the water source (such as lakes or reservoirs). A distance of less than 500m indicates sufficient water vapor supply. The historical average RH value represents the long-term relative humidity level. When it is greater than 85%, it indicates that the air humidity is relatively high for a long period of time, which is conducive to an increase in liquid water content.

[0031] The fourth category is the open and flat type, and its discrimination criteria are: This type represents regions with flat terrain and near-zero curvature, where This indicates that the ground is almost level. This indicates that the surface has neither obvious protrusions nor depressions, and belongs to a typical uniform underlying surface area.

[0032] The fifth type is the leeward slope type, and its identification criteria are: This type is determined by combining the relationship between elevation anomaly and wind direction, among which... This indicates that the area is significantly below the average elevation and is a low-lying area; at the same time, it is located downstream of the prevailing wind direction, which means that the area is significantly affected by descending airflow, usually accompanied by dry air and reduced moisture.

[0033] The above multi-factor combination discrimination method can achieve fine division of the micro-topography where the wind turbine site is located. Different types have significant differences in terrain structure, water vapor conditions and airflow characteristics, providing reliable terrain foundation support for subsequent meteorological parameter modeling and icing prediction.

[0034] In the vertical structural modeling of liquid water content, a parametric expression is constructed to describe the distribution of liquid water content at different altitudes, thereby reflecting the vertical variation of liquid water within the cloud layer. Its parametric expression is as follows: This expression comprehensively considers the peak distribution characteristics, spatial extent, and environmental factors of liquid water content, and can characterize the nonlinear variation trend of liquid water content at different altitudes. The meanings of each parameter are as follows: Indicates height as The liquid water content at a certain altitude is the target variable, used to describe the liquid water concentration level at a certain altitude within the cloud layer; The peak liquid water content is the maximum value in the vertical distribution. Its magnitude is obtained by inverting the liquid water path (LWP) observation value obtained by the ground-based microwave radiometer and combined with the cloud vertical structure information provided by cloud radar. Indicates the height above the ground; This indicates the altitude position corresponding to the maximum liquid water content, and this parameter is usually closely related to the cloud base height; This represents the standard deviation of the vertical distribution, used to characterize the vertical spread of liquid water. Its value is related to cloud thickness. The larger the value, the more dispersed the liquid water distribution. The smaller the value, the more concentrated the liquid water is within a narrower height range. It is a Gaussian distribution function used to describe the liquid water content around the peak height. It exhibits a symmetrical decay pattern, when near When this value is close to 1, it indicates that the liquid water content is near its peak; as... Deviation This item decreases rapidly, reflecting the decay characteristics of liquid water content with altitude.

[0035] This is a micro-topography adjustment function used to correct the basic Gaussian distribution, thereby reflecting temperature. relative humidity And the influence of different micro-topographic types on the distribution of liquid water content. Among them, Indicates ambient temperature. Relative humidity and cloud water generation and evaporation processes are both affected by topographical conditions. Micro-topography types are used to reflect the modulating effect of topographical conditions on airflow lifting, sinking, and water vapor transport.

[0036] Under different micro-topographical conditions, the adjustment function exhibits differentiated correction characteristics: The above parameterized expression method can be used to uniformly model the vertical distribution of liquid water content under different micro-topographic conditions, while preserving the different effects of various terrain environments on cloud micro-physical structures, providing a refined input parameter basis for subsequent icing process calculations.

[0037] In modeling the vertical distribution of median volume diameter (MVD) of droplets, a parametric expression is constructed to characterize the droplet size variation at different heights, thereby reflecting the vertical variation characteristics of the cloud microphysical structure. Its parametric expression is as follows: This expression describes the combined variation trend of droplet size with height, temperature, and micro-topographic conditions by linearly superimposing different influencing factors. The specific meanings of each parameter are as follows: Indicates height as The median volume diameter of droplets at a certain altitude is a key physical quantity reflecting the characteristics of droplet size in clouds, and is used to describe a representative scale of droplet size distribution at a certain altitude.

[0038] This represents the near-surface reference value, serving as the baseline starting point for the entire vertical distribution. Its value is obtained through observation data from a laser raindrop spectrometer or a simple optical particle counter. This parameter reflects the initial state of the near-surface water droplet size.

[0039] It represents the height above the ground and is a spatial variable that describes the vertical position.

[0040] This represents the height gradient coefficient, used to describe the trend of droplet size changing with height. Its typical value range is... to When this parameter is negative, it indicates that the droplet size gradually decreases with increasing height. This phenomenon is usually related to the decrease in air pressure, temperature changes, and the evaporation process of water droplets. The larger the absolute value, the more significant the change in droplet size with height.

[0041] This represents the temperature correction factor, used to reflect the effect of temperature changes on droplet size. Indicates the current ambient temperature. This indicates the reference temperature. When there is a difference between the ambient temperature and the reference temperature, it is indicated by... The particle size is corrected to reflect the regulatory effect of temperature on condensation, evaporation and droplet growth processes.

[0042] This represents the micro-topography type correction term, used to reflect the influence of different terrain conditions on droplet size distribution. This term is a correction function related to micro-topography type, and has different value characteristics under different terrain environments.

[0043] For example, in canyon-type wind tunnel areas, the increased airflow and water vapor condensation due to orographic lifting lead to a relatively larger droplet size, with a corresponding correction range of [missing information]. to (duetoorographicenhancement).

[0044] In leeward slope regions, due to the temperature increase caused by descending airflow and the enhanced evaporation or sublimation processes, the droplet size will decrease, and the corresponding correction range is [missing information]. to (due to evaporation / sublimation).

[0045] The terms in the expression work together to form the vertical distribution pattern of the median droplet volume diameter. Among them, Provide a baseline value. Depicting the trend of altitude change, It reflects the effect of temperature, and Furthermore, the modulating effect of topography on cloud microphysical processes is introduced. This parameterized model enables a unified description of droplet size distribution under different altitudes and topographical conditions, while preserving the differential influence of various environmental factors on droplet size variation, providing key input parameters for subsequent icing process calculations.

[0046] In the temporal evolution modeling of liquid water content (LWC) and median droplet volume diameter (MVD), numerical weather prediction results and climate statistics constraints are introduced to characterize the dynamic changes of meteorological parameters within the prediction lead time, thereby achieving a continuous description of the spatiotemporal evolution of cloud microphysical parameters. This method is based on future meteorological field information output by mesoscale numerical weather prediction models, including temperature. relative humidity Key variables such as wind speed and cloud cover are used to drive the update and change of liquid water content over time.

[0047] The time evolution expression of liquid water content is as follows: This expression comprehensively considers both the numerical forecast trend term and the climatological constraint term to correct the liquid water content at future times. The meanings of each parameter are as follows: Indicates the current time Height is The liquid water content at that point represents the initial state of time evolution.

[0048] Indicates the elapsed time step Subsequently, the updated liquid water content was obtained at the same elevation.

[0049] The time step is indicated by the forecast time resolution and is used to control the scale of time progression.

[0050] This represents the trend of liquid water content changes given by the numerical weather prediction model. This term originates from the forecast results of the mesoscale numerical model and reflects the trend of the impact of changes in meteorological conditions on liquid water content over a future time period. This term is related to the time step. Multiplication is used to estimate the value at the time of multiplication. The change in liquid water content over a period of time.

[0051] The coefficient representing the regression to climatological conditions has a range of values. to This is used to control the rate at which meteorological parameters regress to their long-term statistical state. The larger the value, the faster the regression speed, and the stronger the ability to suppress the accumulation of numerical forecast errors; When the value is small, more information about the dynamic changes in the numerical pattern is preserved.

[0052] This represents the climatological liquid water content obtained through statistical analysis of long-term historical observation data. This parameter reflects the climatological liquid water content at a given altitude. The average climate level at a given location describes the long-term stable state of liquid water content. This is achieved by introducing... It can correct numerical forecast results that deviate from the normal climate over time.

[0053] The entire expression consists of three parts: the first part Indicates the current state; Part Two Indicating the changing trends driven by numerical weather prediction; Part Three This represents the adjustment term for regression to climatological conditions. When numerical forecast results show deviations, the third term can constrain them, preventing errors from accumulating and diverging over time.

[0054] By using this time-evolution parameterized expression, climate statistics constraints can be introduced while preserving the dynamic information of numerical weather prediction models. This makes the changes in liquid water content over time both physically reasonable and long-term stable, thus providing reliable time-varying input parameter support for subsequent icing process calculations.

[0055] In the process of predicting icing on wind turbine blades, a refined three-dimensional computational grid needs to be constructed to accurately characterize the spatial variation features of meteorological parameters and icing distribution within the turbine's swept area. This grid is based on a cylindrical coordinate system, with the center of the wind turbine hub as the spatial reference origin, thus naturally describing the spatial positional changes of the blades as they rotate around the hub.

[0056] Cylindrical coordinate system adopts Form, in which Indicates radial distance. Indicates azimuth. This indicates the vertical height. The radial range is set as follows: in, This refers to the entire length of the blade, covering the space from the center of the hub to the blade tip. The vertical range is set as follows: in, This indicates the hub height, covering the lowest and highest positions the blades may reach during rotation, thus ensuring complete vertical coverage of the entire impeller sweeping surface. The angle range is defined as: Used to represent the complete cycle of one rotation of the blade, it can reflect the differences in aerodynamic and icing conditions at different azimuth angles.

[0057] In terms of spatial discretization, the three-dimensional space is meshed, with radial resolution... A resolution of 1-2m is used to describe the detailed variations in the blade spanwise direction; vertical resolution A range of 1-5m is used to characterize the meteorological differences between different altitude layers; angular resolution A resolution of 1-5° is used to describe the changes in different azimuth angles during blade rotation. This resolution setting allows for a balance between computational accuracy and computational cost.

[0058] During blade rotation, meteorological parameters at any spatial location need to be calculated. This includes the distance from the blade to the hub. At a certain point, its angle ,time The spatial position below can be represented as: in, and These represent the coordinates of the point in the horizontal plane. This indicates the vertical height of the point; Indicates the current rotation angle of the blade; This indicates the radial distance of the point relative to the wheel hub; Indicates the wheel hub height; This indicates the blade cone angle, used to describe the degree of inclination of the blade relative to the plane of rotation. When... When changes occur, the position of that point in space changes periodically, thus reflecting the geometric characteristics of the blade's rotational motion.

[0059] After obtaining the spatial coordinates, the height of the corresponding location needs to be calculated. This is mapped onto the vertical profile of liquid water content and median droplet volume diameter. Using linear interpolation, the corresponding local liquid water content and median droplet volume diameter values ​​can be calculated based on the position of this height within the vertical profile. This process realizes the transformation from a continuous vertical distribution function to discrete grid node meteorological parameters, ensuring that each three-dimensional grid node has corresponding meteorological input parameters.

[0060] By using the above-mentioned three-dimensional mesh construction and spatial mapping methods, it is possible to achieve a detailed characterization of meteorological conditions at any location within the impeller sweep area, providing a reliable spatial basis for subsequent calculation of icing growth rate and three-dimensional icing morphology analysis.

[0061] In modeling the blade icing process, calculating the local icing rate is a crucial step connecting meteorological parameters with the final icing morphology. The improved Makkonen icing model allows for a quantitative description of the local icing process, taking into account both aerodynamic characteristics and thermodynamic processes. First, the collision rate describes the probability of water droplets in the air contacting the blade surface, expressed as: in, The Stokes number represents the collision rate; it measures the ability of a water droplet to deviate from its streamline and collide with a solid surface due to airflow, and its formula is: In the formula, For Stokes numbers; Indicates the density of water droplets; The diameter of the water droplet can be calculated from the median volume diameter of the droplet. This represents the local relative wind speed of the blade, which is the result of the combination of the incoming wind speed and the blade rotational linear velocity. It represents aerodynamic viscosity, used to reflect the resistance characteristics of gas to the movement of water droplets; This represents the chord length of the blade at the corresponding spanwise position. A higher Stokes number indicates stronger inertia in the water droplets, making them more likely to detach from the airflow path and impact the blade surface, thus increasing the collision rate.

[0062] After obtaining the collision rate, the freezing rate needs to be further calculated. This rate describes the proportion of water droplets impacting the blade surface that can actually freeze into ice, and its expression is: in, Indicates the freeze rate; This represents the surface temperature of the blade, which is calculated using the heat balance equation. Indicates ambient air temperature; It represents the convective heat transfer coefficient, used to describe the intensity of heat exchange between air and the blade surface; The latent heat of condensation of water is the heat released when a unit mass of water changes from a liquid state to a solid state. Indicates the liquid water content; This indicates the local relative wind speed on the blades; This represents the collision rate. This expression reflects the combined influence of thermodynamic and fluid conditions on the freezing process; when the surface temperature is lower than the ambient temperature and the heat transfer intensity is high, the freezing efficiency increases.

[0063] Based on this, the icing growth rate is used to describe the increase in ice mass on the blade surface per unit time, and its expression is: in, This indicates the rate of change of icing mass over time; This indicates ice density, which is usually selected based on the type of icing. Values ​​within the range; Indicates the collision rate; Indicates the freeze rate; Indicates the liquid water content; Indicates local relative wind speed; This represents the swept area of ​​the blade at the corresponding position. This formula comprehensively reflects the impact of water droplet supply, collision efficiency, and freezing efficiency on icing growth.

[0064] When considering the blade rotation effect, the angle of attack and relative wind speed change with the azimuth angle due to the rotation process, requiring corrections to the local aerodynamic parameters. The instantaneous angle of attack is expressed as: in, Indicates the instantaneous angle of attack as the azimuth changes; Indicates the design angle of attack; This represents the change in angle of attack caused by rotation. Changes in angle of attack directly affect the airflow separation state, thus altering the collision rate. The value of therefore needs to be based on right Corrections are made. Meanwhile, the centrifugal force generated by the rotation affects the ice layer distribution, causing the ice layer on the outer side of the blade to become relatively thinner.

[0065] After obtaining the icing growth rate, the icing thickness distribution can be obtained through time integration, and its expression is: in, Indicates radial position and azimuth Increase in ice thickness at that location; Indicates the forecast time range; Indicates the rate of ice accumulation; Indicates the density of ice; Indicates the radial position of the blade chord length at the point; This represents the spanwise infinitesimal element length. By integrating the icing growth rate over a time interval, the icing thickness distribution at different spatial locations on the blade surface can be obtained.

[0066] By calculating the collision rate, freezing rate, and icing growth rate layer by layer, and combining the rotation correction and time integration process, a fine characterization of the three-dimensional icing morphology on the blade surface can be achieved, thus providing a reliable physical basis for icing distribution analysis.

[0067] After calculating the three-dimensional icing thickness field, the results need to be standardized and structured to enable quantitative analysis of the spatial distribution characteristics of icing and subsequent application outputs. First, statistical processing is performed on the three-dimensional icing thickness field in the circumferential direction to obtain the icing distribution characteristics in the spanwise direction of the blade. The expression for calculating the spanwise average icing thickness is: in, Indicates radial position The average ice thickness at that location; Indicates radial position and azimuth Ice thickness at the location; The blade rotation azimuth angle has a range of values. to This integration process averages the ice thickness over a complete rotation cycle, thereby eliminating the periodic effects of azimuth changes and obtaining a stable spanwise distribution.

[0068] Based on the average value calculation, extreme value information also needs to be extracted to identify the areas with the most severe icing, which is expressed as follows: in, Indicates radial position The maximum icing thickness along the circumference is given. This parameter characterizes the degree of icing under the most unfavorable operating conditions. Simultaneously, to reflect the non-uniformity of the icing distribution in the circumference, an icing non-uniformity index is introduced, expressed as: in, Indicates the icing unevenness index; Indicates a fixed radial position Circumferential ice thickness Standard deviation; This indicates the average ice thickness at the corresponding location. This indicator measures the dispersion of ice distribution; a higher value indicates more uneven ice coverage.

[0069] After obtaining the thickness distribution, the icing mass distribution can be further calculated, and its expression is: in, Indicates radial position The quality of icing at the location; Indicates the density of ice; This indicates the blade chord length corresponding to that radial position; Indicates the average icing thickness; This represents the discrete length in the radial direction. The total icing mass can be obtained by summing the icing mass at each radial position: in, This indicates the total mass of ice covering the entire blade.

[0070] Based on the maximum icing thickness and total icing mass, the degree of icing can be classified into levels for risk warning and operational decision-making. The specific classification rules are as follows: This grading system uses thickness as the core indicator to intuitively reflect the degree of icing risk.

[0071] After completing the above calculations and classifications, the icing prediction results are converted into a standardized data structure for data interaction and control calls between systems. The output data is organized in JSON format, containing key fields such as wind turbine identifier, prediction time range, icing level, spanwise distribution information, and total icing mass. The spanwise distribution information is given as an array, with each element containing the radial ratio r_ratio, average icing thickness h_mean, maximum icing thickness h_max, and a priority flag to indicate the severity of icing in different areas. The total icing mass is given in the form total_mass_kg, and a recommended_action field is provided to indicate the type of anti-icing and de-icing strategy. This standardized data structure enables direct transmission of icing prediction results to the anti-icing and de-icing control system, thereby supporting the rapid generation and execution of control commands.

[0072] The workflow of this method is as follows: 1. Acquire high-precision digital elevation model data of wind farms, preprocess the original terrain data to eliminate the influence of noise, and extract key terrain factors such as slope, elevation anomaly and terrain curvature. On this basis, complete micro-topography classification to determine the micro-topography type corresponding to the location of wind turbines, and provide terrain conditions for subsequent meteorological parameter modeling. Second, for the identified micro-topography type, the pre-constructed vertical distribution statistical model of liquid water content and median droplet volume diameter is invoked. Combined with historical meteorological observation data and numerical weather forecast information, the vertical profile of cloud microphysical parameters at the current moment and within the future forecast period is generated to achieve continuous characterization of the changes in liquid water content and droplet size with height. Third, based on the parameters of the wind turbine hub height and blade length, a three-dimensional computational grid covering the swept area of ​​the impeller is constructed. The obtained vertical profiles of liquid water content and droplet median volume diameter are mapped to each grid node, so that each spatial location has corresponding local meteorological parameters, thereby forming a spatially continuous meteorological input field. Fourth, an improved Makkonen icing model is introduced at each three-dimensional mesh node. Combined with the changes in relative wind speed and angle of attack generated during blade rotation, the collision rate, freezing rate and icing growth rate at the corresponding locations are calculated to obtain the local icing dynamic characteristics under different spatial locations and azimuth angles. 5. Within the prediction time range, the icing growth rate at each grid node is integrated over time to transform the instantaneous growth process into a cumulative icing effect, thereby obtaining the three-dimensional icing thickness distribution on the blade surface and realizing an overall description of the spatial morphology of icing. VI. Post-processing analysis of the three-dimensional icing thickness field is performed to extract the blade spanwise icing thickness distribution curve and the location of maximum icing, calculate the icing mass distribution and total icing mass, and generate early warning signals based on the icing thickness for classification. At the same time, the results are organized into a standardized data interface format for output to support the invocation and execution of anti-icing and de-icing control strategies.

[0073] This invention effectively overcomes the limitations of traditional methods that rely on single-point meteorological data at hub height by establishing a coupled model between micro-topography classification and the vertical distribution of liquid water content and median droplet volume diameter. This allows for a more detailed characterization of the icing environment at different heights on the rotor swept surface. Under complex terrain conditions, meteorological elements exhibit significant vertical differences; introducing vertical profile modeling more realistically reflects actual icing conditions. Verification using actual wind farm operation data shows that the root mean square error of icing thickness prediction is reduced by 35%-50% compared to traditional methods, significantly improving prediction accuracy and reliability, and providing more accurate data support for the safe operation of wind turbines.

[0074] This invention enables refined prediction of the three-dimensional ice thickness distribution on blade surfaces, overcoming the limitations of traditional methods that only output a single ice thickness or simple level. By constructing a three-dimensional computational grid and combining it with the blade's rotational motion characteristics, it can output multi-dimensional structured information including spanwise ice thickness curves and circumferential non-uniformity indices, thus comprehensively reflecting the spatial distribution characteristics of ice. This multi-dimensional output provides a direct basis for differentiated anti-icing and de-icing strategies, enabling maintenance personnel to take targeted measures for different areas, improving anti-icing and de-icing efficiency and reducing energy consumption.

[0075] This invention constructs a complete physical mechanism chain from terrain identification to icing prediction, covering key links such as terrain identification, vertical structure analysis of cloud microphysical parameters, three-dimensional icing calculation, and standardized output. Each link has a clear physical causal relationship. By introducing a method combining cloud microphysical parameter modeling with an aerodynamic icing model, the extrapolation failure problem of traditional empirical statistical models under complex environmental conditions is avoided, thereby enhancing the physical rationality and applicability of the model and making the prediction results more stable and reliable.

[0076] This invention employs a standardized data structure in its output format, converting prediction results into JSON format data packets that include partition priority identification information. This allows for direct interface integration with the API of anti-icing and de-icing control systems, achieving seamless connection between prediction results and control execution. This approach enables the construction of a closed-loop system of prediction-decision-execution, ensuring that icing prediction not only remains at the analysis level but also directly guides the implementation of control strategies, thereby improving the intelligence level and response efficiency of wind turbine operation and management.

[0077] This invention constructs a multi-dimensional meteorological data support system by integrating multi-source information such as digital elevation model data, radiosonde data, microwave radiometer observation data, cloud radar data, and outputs from mesoscale numerical weather prediction models. Based on this, by combining physical constraints and statistical modeling methods, meteorological parameters are uniformly described and dynamically corrected, effectively improving the model's adaptability and stability under different environmental conditions. This multi-source data fusion mechanism not only enhances the robustness of the prediction system but also improves its generalization ability under complex climatic and topographical conditions.

[0078] This invention provides, for example Figure 2 The wind turbine blade icing prediction device shown includes a micro-topography classification module, an LWC / MVD vertical distribution modeling module, a three-dimensional icing calculation grid module, a local icing rate calculation module, and a prediction result output module. The micro-topography classification module, based on the digital elevation model data of wind farms, extracts slope, elevation, topographic curvature, and valley and ridge line identification features. It uses a topographic factor combination discrimination algorithm to classify wind turbine locations into five micro-topography types: terrain-lifted type, canyon wind tunnel type, water vapor-enhanced type, open and flat type, and leeward slope type. The LWC / MVD vertical distribution modeling module, targeting different micro-topographic types, integrates historical radiosonde data, microwave radiometer observation data, cloud radar echo data, and mesoscale numerical weather prediction output data to construct a statistical model of the vertical profile of liquid water content and median droplet volume diameter, including the vertical distribution function of liquid water content. and the vertical distribution function of the median volume diameter of the droplet ; The 3D icing calculation mesh module is based on the hub height of the wind turbine. and blade length A three-dimensional computational mesh is established in a cylindrical coordinate system centered on the impeller hub. This mesh covers the entire swept surface of the impeller and has a height range of [missing information]. to The radial range covers the entire blade length, with a vertical resolution of 1-5 meters and an angular resolution of 1-5 degrees. The local icing rate calculation module maps the vertical profiles of liquid water content and the median volume diameter of droplets to each node of the three-dimensional computational grid. For each spatial position traversed during the blade rotation, the improved Makkonen icing model is used to calculate the local collision rate, freezing rate, and icing rate increase, thereby obtaining the icing thickness distribution at different spanwise positions and different rotation angles on the blade surface. The prediction result output module, based on the obtained icing thickness distribution, outputs the blade spanwise icing thickness curve, the maximum icing location, the icing mass distribution, and the icing evolution trend within the prediction time range, and generates an icing level warning signal, providing a standardized input interface for the anti-icing and de-icing control system.

[0079] The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling provided in this invention is implemented through the aforementioned wind turbine blade icing prediction system based on micro-topography classification and LWC_MVD vertical distribution modeling. For details of the specific methods and procedures of the wind turbine blade icing prediction system based on micro-topography classification and LWC_MVD vertical distribution modeling, please refer to the embodiments of the aforementioned wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling, which will not be repeated here.

[0080] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for predicting icing on wind turbine blades based on micro-topography classification and LWC_MVD vertical distribution modeling, characterized in that, Includes the following steps: Obtain digital elevation data of the target wind farm, extract terrain feature parameters, classify the micro-topography of the wind turbine location, and obtain the corresponding micro-topography type. For the obtained micro-topography type, the pre-established vertical distribution statistical model of liquid water content and median droplet volume diameter is called to generate the vertical profile of liquid water content and median droplet volume diameter at the corresponding location. The vertical distribution statistical model is used to characterize the nonlinear variation characteristics of liquid water content and median droplet volume diameter along the height direction under different micro-topography conditions. A three-dimensional computational grid covering the swept area of ​​the wind turbine rotor is constructed. The generated vertical profiles of liquid water content and median volume diameter of droplets are mapped to each node of the three-dimensional computational grid to obtain the meteorological parameter distribution corresponding to each node. Based on the meteorological parameters of each node obtained after mapping, the local icing rate of each node is calculated using the icing growth model, and the local icing rate is integrated based on the time dimension to obtain the three-dimensional icing thickness distribution on the blade surface. Post-processing is performed based on the obtained three-dimensional icing thickness distribution to extract the blade spanwise icing distribution features and classify the icing severity level, outputting the corresponding icing prediction results.

2. The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The statistical model for the vertical distribution of liquid water content adopts the following functional form: ； The statistical model for the vertical distribution of the median droplet volume diameter adopts the following functional form: ； in, Indicates the height above the ground. Indicates temperature. Indicates relative humidity. Indicates peak liquid water content. Indicates peak height. Indicates the standard deviation of the vertical distribution. This represents the micro-topography adjustment function. Indicates near-surface reference value, , Represents the gradient coefficients. This indicates a micro-topography correction term.

3. The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The temporal evolution processing is as follows: the future temperature and humidity change trends output by the mesoscale numerical weather prediction model are obtained, climatological constraints are introduced, and the vertical profiles of liquid water content and median droplet volume diameter are dynamically updated to realize the spatiotemporal evolution of meteorological parameters within the forecast lead time.

4. The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The three-dimensional computational mesh uses a cylindrical coordinate system with the wind turbine hub as the origin. The radial range is set to Vertical range set to The angle range is set to ,in Indicates the blade length. Indicates the wheel hub height.

5. The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The icing growth model adopts an improved form that includes rotation effects. The local icing rate calculation process includes correcting the collision rate based on the instantaneous angle of attack change caused by blade rotation, and correcting the centrifugal force effect based on the blade spanwise position function, thereby obtaining the icing growth characteristics corresponding to the spatial location.

6. The method for predicting icing on wind turbine blades based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The output includes standardized interface data, which is used to call the anti-icing and de-icing control system. The data content includes blade spanwise icing information and anti-icing and de-icing priority identifiers for each icing zone.

7. The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The micro-topography classification process includes extracting slope factors, topographic curvature, and elevation anomalies, and outputting classification results for types such as uplift, canyon wind tunnel, water vapor enhancement, open and flat, and leeward slope through topographic factor combination discrimination rules.

8. The wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in claim 1, characterized in that, The three-dimensional ice thickness distribution results are used to calculate the spanwise average ice thickness, maximum ice thickness, and ice non-uniformity index, and the spatial distribution characteristics of ice are obtained through circumferential statistics.

9. A wind turbine blade icing prediction device based on micro-topography classification and LWC_MVD vertical distribution modeling, used to implement the wind turbine blade icing prediction method based on micro-topography classification and LWC_MVD vertical distribution modeling as described in any one of claims 1-8, characterized in that, It includes a micro-topography classification module, an LWC / MVD vertical distribution modeling module, a 3D icing calculation grid module, a local icing rate calculation module, and a prediction result output module: The micro-topography classification module, based on the digital elevation model data of wind farms, extracts slope, elevation, topographic curvature, and valley and ridge line identification features. It uses a topographic factor combination discrimination algorithm to classify wind turbine locations into five micro-topography types: terrain-lifted type, canyon wind tunnel type, water vapor-enhanced type, open and flat type, and leeward slope type. The LWC / MVD vertical distribution modeling module, targeting different micro-topographic types, integrates historical radiosonde data, microwave radiometer observation data, cloud radar echo data, and mesoscale numerical weather prediction output data to construct a statistical model of the vertical profile of liquid water content and median droplet volume diameter, including the vertical distribution function of liquid water content. and the vertical distribution function of the median volume diameter of the droplet ; The 3D icing calculation mesh module is based on the hub height of the wind turbine. and blade length A three-dimensional computational mesh is established in a cylindrical coordinate system centered on the impeller hub. This mesh covers the entire swept surface of the impeller and has a height range of [missing information]. to The radial range covers the entire blade length, with a vertical resolution of 1-5 meters and an angular resolution of 1-5 degrees. The local icing rate calculation module maps the vertical profiles of liquid water content and the median volume diameter of droplets to each node of the three-dimensional computational grid. For each spatial position traversed during the blade rotation, the improved Makkonen icing model is used to calculate the local collision rate, freezing rate, and icing rate increase, thereby obtaining the icing thickness distribution at different spanwise positions and different rotation angles on the blade surface. The prediction result output module, based on the obtained icing thickness distribution, outputs the blade spanwise icing thickness curve, the maximum icing location, the icing mass distribution, and the icing evolution trend within the prediction time range, and generates an icing level warning signal, providing a standardized input interface for the anti-icing and de-icing control system.

10. A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method according to any one of claims 1-8.

Images