Complex terrain wind resource assessment method and system based on multi-source data coupling

Abstract

The application discloses a complex terrain wind resource evaluation method and system based on multi-source data coupling. The method comprises the following steps: S1, multi-source data acquisition and fusion pretreatment; S2, initial wind field construction based on a terrain physical model; S3, fine correction of a variational data assimilation wind field; S4, multi-dimensional wind resource parameter extraction and evaluation; S5, spatial visualization and result output; and S6, evaluation result verification and uncertainty quantification. The system comprises the following modules: a multi-source data acquisition and communication module, a terrain data processing and initial wind field modeling module, a variational data assimilation wind field correction module, a wind resource evaluation and visualization module, and an evaluation verification and uncertainty analysis module. The application couples multi-source observation data and a physical model, overcomes the limitations of traditional methods in terms of inaccurate wind field simulation in complex terrain areas and single evaluation dimension, and significantly improves the accuracy, reliability and spatial resolution of wind resource evaluation. The application relates to the technical field of wind energy resource evaluation and wind power planning.

Description

Technical Field

[0001] This invention relates to the field of wind energy resource assessment and wind power planning technology, specifically to a method and system for assessing wind resources in complex terrain based on multi-source data coupling. Background Technology

[0002] As wind power development extends to complex terrain areas such as mountains, hills, and coastal zones, the accuracy and reliability of wind resource assessment face unprecedented challenges. Wind fields in complex terrain areas are affected by various aerodynamic processes such as orographic uplift, canyon effect, flow around, separation, and reattachment, exhibiting strong three-dimensional non-uniformity and spatiotemporal variability. Traditional assessment methods based on uniformity assumptions or simple interpolation are severely inadequate in such areas.

[0003] Current wind resource assessment in complex terrain mainly relies on the following types of methods: 1. Field observation method: This method relies on a limited number of wind measurement towers for single-point observation. Although it has high accuracy, it has poor spatial representativeness, high cost, and is difficult to deploy in areas with complex terrain. 2. Numerical simulation method: Computational fluid dynamics (CFD) models (such as RANS and LES) are used to simulate wind fields, which can reflect the influence of terrain. However, due to the limitations of model parameterization and boundary condition uncertainties, the simulation results have systematic biases. 3. Mesoscale-microscale coupling method: Combining mesoscale meteorological models with microscale CFD models can take into account both large-scale meteorological background and topographic details, but the computational cost is high and the observational data is not fully integrated. 4. Data interpolation methods: such as Kriging interpolation and inverse distance weighting, which are based on sparse observation points for spatial extrapolation, cannot reflect the flow field distortion caused by topography and have limited accuracy.

[0004] Existing methods generally suffer from the following problems: 1) Insufficient data utilization: The rich data resources of wind farm operation (such as SCADA and lidar) were not fully integrated, resulting in the observation information not being effectively incorporated into the evaluation system; 2) Model and observation disconnect: Although CFD models can simulate topographic effects, they lack real-time observation constraints, and there are uncorrected systematic errors between simulated wind fields and real wind fields; 3) Single evaluation index: It focuses more on wind speed and wind power density, and lacks a coordinated evaluation of key parameters that affect the safety and life of wind turbines, such as turbulence intensity, wind shear, and extreme wind speed. 4) Lack of quantification of uncertainty: The assessment results often provide a single value without providing confidence intervals or probability distributions, which is not conducive to risk assessment and decision optimization.

[0005] Therefore, there is an urgent need for a complex terrain wind resource assessment method that can deeply integrate multi-source observation data, couple terrain physical models, and has high spatial resolution and reliability to support the scientific site selection, capacity prediction and safe operation of wind farms. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for assessing wind resources in complex terrain based on multi-source data coupling. This method enables high-precision reconstruction of three-dimensional wind fields and refined spatial assessment of wind resource parameters under complex terrain, providing reliable data support for the early planning, micro-site selection and later operation and maintenance of wind power projects.

[0007] The technical solution adopted in this invention is a method for assessing wind resources in complex terrain based on multi-source data coupling, which includes the following steps: S1. Multi-source data acquisition and fusion preprocessing: Simultaneously acquire SCADA operation data, laser wind radar point cloud data, and wind tower observation data of wind farms in the target complex terrain area. Perform time synchronization, coordinate unification, quality control, and standardization preprocessing on the acquired data to form a spatiotemporally consistent multi-source dataset. S2. Initial wind field construction based on terrain physical model: Import high-resolution digital elevation model, obtain high-resolution terrain data of the evaluation area, establish terrain adaptive three-dimensional computing grid, use RANS model to simulate three-dimensional steady-state wind field under multiple prevailing wind directions, and construct multi-wind direction initial wind field database. S3. Variational data assimilation wind field refinement correction: Using the multi-source data processed in step S1 as the observation value and the initial wind field obtained in step S2 as the background field, construct an objective function containing background and observation terms, and use the variational data assimilation algorithm to solve for a high-precision three-dimensional corrected wind field. S4. Multidimensional wind resource parameter extraction and evaluation: Extract multidimensional parameters such as wind speed, wind direction, turbulence intensity, and wind shear from the modified wind field, and construct a three-level evaluation index system that includes wind energy resource potential, wind energy quality, and development suitability. S5. Spatial Visualization and Output: Based on the geographic information system platform, realize spatial interpolation, classification and thematic mapping of wind resource parameters, and identify wind energy rich areas and suitable development areas; S6. Evaluation Result Verification and Uncertainty Quantification: The accuracy of the evaluation is verified by cross-validation using independent observation data, and the uncertainty of the evaluation results is quantified by ensemble perturbation method and Monte Carlo simulation.

[0008] Furthermore, in step S1, the SCADA operating data includes at least the wind turbine pitch angle, generator speed, generator power, nacelle wind speed and direction; the laser wind measurement radar point cloud data is radial wind speed point cloud data acquired by three-dimensional scanning radar, covering key terrain and flow field feature areas of the assessment area; the wind measurement tower observation data includes at least multi-layer wind speed, wind direction, temperature, air pressure and turbulence intensity observations.

[0009] Furthermore, in step S2, the high-resolution terrain data is derived from a high-resolution digital elevation model or lidar terrain scanning data, with a spatial resolution of not less than 10m; the RANS model uses a RANS solver based on a k-ε or k-ω turbulence model, combined with terrain roughness and vegetation cover parameters, to simulate steady-state wind fields under multiple prevailing wind directions and construct an initial three-dimensional wind speed vector field.

[0010] Furthermore, in step S3, the variational data assimilation algorithm adopts a three-dimensional variational (3D-Var) or four-dimensional variational (4D-Var) algorithm; the observation items include lidar radial wind speed projection, SCADA data inversion equivalent wind speed, and wind tower multi-layer observation interpolation; in the background item, the background error covariance matrix is ​​adaptively constructed using terrain-dependent flow field characteristics.

[0011] Furthermore, in step S4, the wind resource assessment indicators of the three-level assessment indicator system include: annual average wind speed, wind power density, turbulence intensity, wind shear index, and effective wind time ratio; in step S5, a geographic information system platform is used to realize the spatial distribution map of wind resources, the identification of rich areas, and the classification of levels.

[0012] Furthermore, step S6 is as follows: S61. Cross-validate using reserved independent wind measurement tower or wind turbine power data, and calculate the mean deviation (MBE), root mean square error (RMSE), and consistency index (IOA). S62. Using the ensemble perturbation method, random perturbations are applied to the observed data and model parameters to generate an assimilation result set, and the probability distribution and confidence interval of key evaluation indicators are statistically analyzed.

[0013] A complex terrain wind resource assessment system based on multi-source data coupling is provided to implement the method described above. The system includes a multi-source data acquisition and communication module, a terrain data processing and initial wind field modeling module, a variational data assimilation and wind field correction module, a wind resource assessment and visualization module, and an assessment verification and uncertainty analysis module. The multi-source data acquisition and communication module includes a SCADA data interface unit, a lidar point cloud receiving unit, and a wind tower data acquisition unit, which are used to simultaneously acquire wind farm operation data, three-dimensional radial wind speed point clouds, and multi-layer meteorological observation data, and perform time alignment and coordinate transformation. The terrain data processing and initial wind field modeling module includes a high-resolution digital elevation model import unit, a terrain adaptive mesh generation unit, and a RANS wind field simulation unit. The RANS wind field simulation unit is equipped with a k-ε or k-ω turbulence model solver, which is used to construct an initial three-dimensional wind field database based on terrain roughness and prevailing wind direction. The variational data assimilation wind field correction module includes an objective function construction unit and a variational assimilation calculation unit. The objective function construction unit is used to construct an energy functional that includes background field terms and observation terms. The variational assimilation calculation unit uses a three-dimensional variational or four-dimensional variational algorithm to obtain a high-precision three-dimensional corrected wind field by minimizing the objective function. The wind resource assessment and visualization module includes a wind resource parameter extraction unit, a three-level assessment index calculation unit, and a geographic information system mapping unit. The three-level assessment index calculation unit constructs an assessment matrix based on wind energy resource potential, wind energy quality, and development suitability. The geographic information system mapping unit is used to output a wind resource spatial distribution map and a rich area identification map. The evaluation verification and uncertainty analysis module includes an accuracy verification unit and an uncertainty quantification unit. The accuracy verification unit uses independent observation data to calculate the average deviation, root mean square error, and consistency index to evaluate the evaluation accuracy. The uncertainty quantification unit uses the ensemble perturbation method and Monte Carlo simulation to apply random perturbations to the model parameters and observation data, and statistically analyzes the probability distribution and confidence interval of key evaluation indicators.

[0014] The beneficial effects of this invention are as follows: (1) Significantly improved assessment accuracy: By assimilating and fusing multi-source data, the observation information of SCADA, lidar, and wind tower is organically combined with the CFD physical model. Compared with the traditional single method, the wind speed assessment error is reduced by 40% to 60%, and the turbulence intensity assessment accuracy is improved by more than 30%. (2) Breakthrough improvement in spatial resolution: By adopting terrain-adaptive grid densification technology and multi-source data fusion, the spatial resolution of the evaluation results reaches the level of 10 to 50 meters, which can accurately capture the local wind field acceleration, separation and flow around effects caused by complex terrain; (3) Comprehensive and systematic evaluation dimensions: A three-level evaluation index system including wind energy resource potential, wind energy quality and development suitability has been constructed. It not only evaluates the amount of wind energy resources, but also the stability of resources, development risks and economic benefits. (4) Strong engineering practicality: A complete technical process from data acquisition, model calculation, assimilation correction to visualization output has been established. It can be integrated into the existing wind resource assessment platform, supports batch automated processing, and shortens the assessment cycle by more than 50%. (5) Uncertainty can be quantified: Through set perturbation and Monte Carlo methods, the uncertainty of the evaluation results is quantified, providing confidence intervals and probability distributions, and providing a basis for risk quantification for investment decisions; (6) High adaptability: The method can be applied to various complex terrain types such as mountains, hills, coastal zones, and canyons, and is particularly suitable for the expansion assessment and micro-site optimization of existing wind farms. Attached Figure Description

[0015] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a simplified structural diagram of the system of the present invention; Figure 3 This is a schematic diagram illustrating the construction of initial wind fields in multiple wind directions based on terrain data and RANS models. Figure 4 A spatial distribution visualization of downwind resource assessment results in complex terrain; Figure 5 This is a comparison chart to verify the evaluation results with the measured data from an independent wind measurement tower. Detailed Implementation

[0016] like Figures 1-5 As shown, a method for assessing wind resources in complex terrain based on multi-source data coupling is proposed. This method includes the following steps: S1. Multi-source data acquisition and fusion preprocessing: Simultaneously acquire SCADA operation data, laser wind radar point cloud data, and wind tower observation data of wind farms in the target complex terrain area. Perform time synchronization, coordinate unification, quality control, and standardization preprocessing on the acquired data to form a spatiotemporally consistent multi-source dataset. S2. Initial wind field construction based on terrain physical model: Import high-resolution digital elevation model, obtain high-resolution terrain data of the evaluation area, establish terrain adaptive three-dimensional computing grid, use RANS model to simulate three-dimensional steady-state wind field under multiple prevailing wind directions, and construct multi-wind direction initial wind field database. S3. Variational data assimilation wind field refinement correction: Using the multi-source data processed in step S1 as the observation value and the initial wind field obtained in step S2 as the background field, construct an objective function containing background and observation terms, and use the variational data assimilation algorithm to solve for a high-precision three-dimensional corrected wind field. S4. Multidimensional wind resource parameter extraction and evaluation: Extract multidimensional parameters such as wind speed, wind direction, turbulence intensity, and wind shear from the modified wind field, and construct a three-level evaluation index system that includes wind energy resource potential, wind energy quality, and development suitability. S5. Spatial Visualization and Output: Based on the geographic information system platform, realize spatial interpolation, classification and thematic mapping of wind resource parameters, and identify wind energy rich areas and suitable development areas; S6. Evaluation Result Verification and Uncertainty Quantification: The accuracy of the evaluation is verified by cross-validation using independent observation data, and the uncertainty of the evaluation results is quantified by ensemble perturbation method and Monte Carlo simulation.

[0017] Specifically, in step S1, the SCADA operating data includes at least the wind turbine pitch angle, generator speed, generator power, nacelle wind speed and wind direction; the laser wind measurement radar point cloud data is radial wind speed point cloud data acquired by three-dimensional scanning radar, covering key terrain and flow field feature areas of the assessment area; the wind measurement tower observation data includes at least multi-layer wind speed, wind direction, temperature, air pressure and turbulence intensity observations.

[0018] In step S2, the high-resolution terrain data comes from a high-resolution digital elevation model or lidar terrain scanning data, with a spatial resolution of not less than 10m; the RANS model uses a RANS solver based on a k-ε or k-ω turbulence model, combined with terrain roughness and vegetation cover parameters, to simulate steady-state wind fields under multiple prevailing wind directions and construct an initial three-dimensional wind speed vector field.

[0019] In step S3, the variational data assimilation algorithm uses either a three-dimensional variational (3D-Var) or a four-dimensional variational (4D-Var) algorithm; the observation items include lidar radial wind speed projection, SCADA data inversion equivalent wind speed, and wind tower multi-layer observation interpolation; in the background item, the background error covariance matrix is ​​adaptively constructed using terrain-dependent flow field features.

[0020] In step S4, the wind resource assessment indicators of the three-level assessment index system include: annual average wind speed, wind power density, turbulence intensity, wind shear index, and effective wind time ratio; in step S5, a geographic information system platform is used to realize the spatial distribution map of wind resources, the identification of rich areas, and the classification of levels.

[0021] The S6 steps are as follows: S61. Cross-validate using reserved independent wind measurement tower or wind turbine power data, and calculate the mean deviation (MBE), root mean square error (RMSE), and consistency index (IOA). S62. Using the ensemble perturbation method, random perturbations are applied to the observed data and model parameters to generate an assimilation result set, and the probability distribution and confidence interval of key evaluation indicators are statistically analyzed.

[0022] A complex terrain wind resource assessment system based on multi-source data coupling, used to implement the method described above, comprises: 1. Multi-source data acquisition and communication module; 2. Terrain data processing and initial wind field modeling module; 3. Variational data assimilation and wind field correction module; 4. Wind resource assessment and visualization module; and 5. Assessment verification and uncertainty analysis module. The multi-source data acquisition and communication module 1 includes a SCADA data interface unit 11, a lidar point cloud receiving unit 12, and a wind tower data acquisition unit 13, which are used to simultaneously acquire wind farm operation data, three-dimensional radial wind speed point cloud, and multi-layer meteorological observation data, and perform time alignment and coordinate transformation. The terrain data processing and initial wind field modeling module 2 includes a high-resolution digital elevation model import unit 21, a terrain adaptive mesh generation unit 22, and a RANS wind field simulation unit 23. The RANS wind field simulation unit 23 is equipped with a k-ε or k-ω turbulence model solver, which is used to construct an initial three-dimensional wind field database based on terrain roughness and prevailing wind direction. The variational data assimilation wind field correction module 3 includes an objective function construction unit 31 and a variational assimilation calculation unit 32. The objective function construction unit 31 is used to construct an energy functional containing background field terms and observation terms. The variational assimilation calculation unit 32 uses a three-dimensional variational or four-dimensional variational algorithm to obtain a high-precision three-dimensional corrected wind field by minimizing the objective function. The wind resource assessment and visualization module 4 includes a wind resource parameter extraction unit 41, a three-level assessment index calculation unit 42, and a geographic information system mapping unit 43. The three-level assessment index calculation unit 42 constructs an assessment matrix based on wind energy resource potential, wind energy quality, and development suitability. The geographic information system mapping unit 43 is used to output a wind resource spatial distribution map and a rich area identification map. The evaluation verification and uncertainty analysis module 5 includes an accuracy verification unit 51 and an uncertainty quantification unit 52. The accuracy verification unit 51 uses independent observation data to calculate the average deviation, root mean square error and consistency index to evaluate the evaluation accuracy. The uncertainty quantification unit 52 uses the ensemble perturbation method and Monte Carlo simulation to apply random perturbations to the model parameters and observation data, and statistically analyzes the probability distribution and confidence interval of key evaluation indicators.

[0023] The method of the present invention will now be described in a further manner.

[0024] The specific technical solution of this invention includes the following steps: Step S1: Multi-source data acquisition and fusion preprocessing 1. Simultaneously collect three types of data from wind farms within the target area's complex terrain: 1) SCADA data: High-frequency operating data (e.g., 1Hz) of each wind turbine are obtained from the wind farm monitoring system, including but not limited to: turbine ID, timestamp, active power, generator speed, pitch angle, nacelle wind speed, nacelle wind direction, yaw angle, etc. 2) Laser wind radar data: Three-dimensional scanning laser radars deployed at key terrain locations (such as ridges, canyon entrances, and wake areas between wind turbines) acquire radial wind speed point cloud data of spatially discrete points, including point coordinates (x, y, z) and radial wind speed; 3) Wind tower data: Existing wind tower observation data in the area, including wind speed, wind direction, temperature, air pressure, turbulence intensity, etc. at multiple heights such as 10m, 50m, and 100m, with a time resolution of not less than 10 minutes.

[0025] 2. Preprocessing includes: 1) Time synchronization: Based on UTC time, timestamp alignment and interpolation synchronization are performed on all data streams; 2) Quality control: Remove outliers that exceed the physical range and filter low-quality lidar points based on the signal-to-noise ratio; 3) Missing value handling: For missing values ​​caused by short-term communication interruptions, linear interpolation or nearest neighbor values ​​are used to fill the gaps; 4) Coordinate unification: Convert all data to the same coordinate system (such as UTM or the local Cartesian coordinate system); 5) Standardization: Perform zero-mean, unit-variance standardization on continuous variables such as wind speed and power.

[0026] Step S2: Initial wind field construction based on terrain data and RANS model 1) Terrain data processing: Import high-resolution digital elevation model (DEM, resolution ≤10m), combine with land use / vegetation cover data to generate terrain elevation field and surface roughness field, and obtain high-resolution terrain data of the assessment area; 2) Computational mesh generation: Construct a three-dimensional structured or unstructured mesh covering the evaluation area, and refine the near-surface layer mesh to capture terrain gradients; 3) RANS model settings: Use a steady-state RANS solver based on the k-ε or k-ω turbulence model, set the inlet boundary conditions to multiple prevailing wind directions (e.g., 16 directions), the outlet to be a pressure outlet, the ground to be a no-slip wall, and the top to be a symmetric or free-slip boundary. 4) Solving the initial wind field for multiple wind directions: For each dominant wind direction, solve the RANS equation to obtain the three-dimensional wind speed, turbulent kinetic energy, dissipation rate and other field variables under that wind direction; integrate the simulation results of all wind directions to construct a multi-wind-direction initial wind field library as the background field basis for data assimilation.

[0027] Step S3: Wind field correction based on variational data assimilation 1. Preparation of observation data: 1) Radial wind speed is observed directly using lidar; 2) SCADA data is used as an indirect observation by inverting the equivalent hub height wind speed through aerodynamic models; 3) The wind speed and direction of the wind measuring tower are observed as a vertical profile.

[0028] 2. Background field preparation: Based on the current actual wind direction, select the closest background wind field from the multi-wind-direction initial wind field library, or generate a real-time background field through wind direction interpolation.

[0029] 3. Variational Assimilation Execution: In this embodiment, a three-dimensional variational (3D-Var) method is used to construct the objective function: J(x) = ½(xx) b ) T B -1 (xx b )+½[yH(x)] T R -1 [yH(x)] in: x is the state vector (three-dimensional wind speed field); x b For background scene; xx b Let x be the residual vector of the background field, representing the state vector x (i.e., the optimal three-dimensional wind speed field to be found) and the background field vector x. b The difference between them; (xx b ) T The transpose of the background field residual vector represents the transpose of the column vector (xx). b Convert the matrix into row vectors so that matrix inner product operations can be performed. B is the background error covariance matrix, which is adaptively constructed using a terrain flow field correlation structure. B -1 It is the inverse of the background error covariance matrix; y is the observation vector; H(x) is the observation operator that maps the state vector x to the observation space; [yH(x)] is the observation residual vector, representing the difference between the observation vector y (the actual observation value of SCADA, radar, and wind tower) and the simulated observation value after the observation operator H(x) is applied to the state vector; [yH(x)] T To observe the transpose of residual direction error; R is the observation error covariance matrix; R -1 It is the inverse of the observation error covariance matrix.

[0030] 4. Variational assimilation solution process: The objective function minimization problem is solved using a quasi-Newton optimization algorithm (L-BFGS). The specific iterative steps are as follows: 1) Initialization: Set the iteration number k=0, and the initial state vector. Calculate the initial objective function value ; 2) Gradient calculation: Calculate the gradient of the objective function. ; 3) Search direction determination: Calculate the search direction using the L-BFGS method. ; 4) Line search: along The step size is determined by performing an Armijo line search in the direction. ; 5) Status Update: ; 6) Convergence judgment: If (usually taken) If the maximum number of iterations is reached, the iteration stops and the analysis field is output. Otherwise, let k = k + 1 and return to step 2).

[0031] 5. Assimilation quality control: 1) Observational data quality control: The Buddy Check method is used to check the spatial consistency of the observational data and remove abnormal observation points; 2) Background field error covariance localization: The Gaspari-Cohn function is used to spatially localize the background error covariance matrix B, reducing long-distance spurious correlations; 3) Incremental analysis: The incremental 4D-Var method is used to assimilate the observation data within the assimilation window in batches, thereby improving computational efficiency.

[0032] Step S4: Wind resource parameter extraction and evaluation index system construction 1. Key parameter extraction: Extracted grid-by-grid from the modified wind field: 1) Wind speed parameters: including horizontal wind speed (U, V components), vertical wind speed (W component), average wind speed, standard deviation, maximum value, and percentile value; 2) Wind direction parameters: prevailing wind direction, wind direction frequency distribution, and wind direction stability coefficient; 3) Turbulence characteristics: turbulence intensity, turbulence kinetic energy, and turbulence dissipation rate; 4) Wind shear parameters: vertical wind shear index, horizontal wind shear gradient; 5) Extreme wind conditions: return period of extreme wind speeds, gust factor, and wind speed frequency distribution.

[0033] 2. Construction of the evaluation indicator system: Construct a comprehensive evaluation system that includes three levels of indicators: Primary indicator: Wind energy resource potential (weight 40%), including annual average wind speed, wind power density, and effective wind time percentage; Secondary indicators: Wind energy quality indicators (weight 35%), including turbulence intensity level, wind shear characteristics, and wind direction stability; Level 3 indicators: Development suitability indicators (weight 25%), including terrain complexity, extreme wind risk, and distance from the existing power grid.

[0034] 3. Spatial visualization and classification: Using a GIS platform, draw a spatial distribution map of wind resources, identify wind energy rich areas (such as high wind speed and low turbulence areas), and classify the resources into categories (such as wind zones I-IV).

[0035] Step S5: Evaluation Result Verification and Uncertainty Analysis 1. Cross-validation: Using data from reserved independent wind measurement towers or actual wind turbine power generation, compare and evaluate the predicted wind speed with the measured value, and calculate the mean deviation (MBE), root mean square error (RMSE), and index of consistency (IOA). 2. Uncertainty Quantification: The ensemble perturbation method is used to apply random perturbations to the observed data and model parameters, generating multiple assimilation result sets, and statistically analyzing the probability distribution and confidence intervals of key evaluation indicators.

[0036] The technical solution of the present invention will be described in more detail below with reference to the accompanying drawings and specific embodiments.

[0037] Specific Implementation Example: Taking a wind farm expansion project in a southern mountainous area as an example, the terrain in this region is complex, with an elevation difference of approximately 800 meters, including multiple ridges, valleys, and slopes. The existing wind farm's first phase is equipped with 30 2MW wind turbines, and a second phase expansion is planned in the surrounding area. A detailed wind resource assessment of the expansion area is required.

[0038] Step S1: Multi-source data acquisition and preprocessing 1) SCADA Data: One year of 1Hz operation data was obtained from 30 wind turbines in Phase I. Key fields include: turbine ID, timestamp, and active power. P ), generator speed ( ω ), pitch angle ( β ), cabin wind speed ( v nac ), cabin wind direction ( θ nac ).

[0039] 2) LiDAR data: One 3D scanning LiDAR is deployed on the eastern ridge and the western valley entrance of the region. The scanning frequency is 4Hz. Each LiDAR covers about 5,000 spatial points. The data includes point coordinates and radial wind speed.

[0040] 3) Wind tower data: There are currently 3 100m wind towers in the area, providing wind speed, wind direction, temperature and air pressure data at 10m, 50m and 100m levels, with a time resolution of 10 minutes.

[0041] After preprocessing, a multi-source dataset with synchronized time, unified coordinates, and controllable quality is formed.

[0042] Step S2: Initial wind field construction based on terrain data and RANS model 1. Import 30m resolution DEM data and estimate the surface roughness length by combining it with satellite remote sensing vegetation index. z 0 ).

[0043] 2. Construct a three-dimensional structured grid with a region size of 5km×5km×1km, a horizontal resolution of 50m, and 10 vertical layers within 10m of the ground.

[0044] 3. The simpleFoam solver (k-ε turbulence model) in OpenFOAM was used, and 16 wind direction inlets were set (0°, 22.5°, ..., 337.5°). The vertical wind speed profile followed a power-law distribution (α=0.2).

[0045] 4. Solve the steady-state flow field under each wind direction, save the three-dimensional wind speed vector (U, V, W) of each grid, and build the initial wind field library.

[0046] Step S3: Wind field correction based on variational data assimilation 1. Preparation of observations: 1) Radial wind speed from lidar can be used directly; 2) The SCADA data is used to invert the equivalent wind speed at the hub height based on the CP-λ-β curve, which serves as a virtual observation point for the wind turbine location; 3) Vertically interpolate the wind tower data to the grid height.

[0047] 2. Background field selection: Select the background field closest to the prevailing wind direction from the initial wind field database based on the prevailing wind direction for the current hour.

[0048] 3. A 3D-Var assimilation system is adopted, and the background error covariance B is set to use a Gaussian distance correlation function with a horizontal correlation scale of 500m and a vertical correlation scale of 100m; the observation error is set according to the instrument accuracy.

[0049] 4. The assimilation period is 1 hour, and the output is a high-precision three-dimensional corrected wind field per hour.

[0050] Step S4: Wind Resource Enrichment Assessment 1. Based on the corrected wind field of 8760 hours throughout the year, calculate the following for each grid: 1) Average annual wind speed: 7.8 m / s; 2) Wind power density: 420 W / m²; 3) Turbulence intensity (100m height): 0.12; 4) Wind shear index (10-100m): 0.28; 5) Effective wind time percentage: 78%.

[0051] 2. Draw a wind resource distribution map in ArcGIS (see...) Figure 4 The study identified the northeastern ridge area as a Class I enrichment area (wind speed > 8.5 m / s, TI < 0.10) and the western valley as a Class III area (wind speed 6.0-7.0 m / s, TI > 0.15).

[0052] Step S5: Evaluation Result Verification and Uncertainty Analysis 1. Verification was conducted using a reserved wind measurement tower (not involved in assimilation), comparing the annual wind speed data (see...). Figure 5 ): 1) Mean deviation MBE = 0.15 m / s; 2) Root mean square error RMSE = 0.68 m / s; 3) The consistency index IOA = 0.94.

[0053] 2. By adding a 5% random perturbation to the observation data, 20 assimilation sets were generated, and the 95% confidence interval of the annual average wind speed was calculated to be [7.62, 7.98] m / s.

[0054] 3. Comparative analysis with traditional methods: To quantify the advantages of this invention, the evaluation results are compared with three traditional methods: (1) Traditional wind tower interpolation method: only data from 3 wind towers are used, and ordinary Kriging interpolation is adopted; (2) Pure CFD simulation method: only RANS model is used for simulation, without data assimilation correction; (3) Mesoscale-microscale coupling method: WRF+CFD coupling simulation is used.

[0055] The comparison results are shown in the table below:

[0056] 4. Uncertainty propagation analysis: The Monte Carlo method was used to analyze the impact of uncertainties at each stage on the final evaluation results: (1) Uncertainty in observation data: LiDAR wind measurement error ±0.3m / s, SCADA inversion error ±0.5m / s, wind tower observation error ±0.2m / s; (2) Model parameter uncertainty: Uncertainty in terrain roughness leads to wind speed error ±8% and atmospheric stability parameterization error ±12%; (3) Overall uncertainty: Through 500 Monte Carlo simulations, the 95% confidence interval for the annual average wind speed is 7.62 to 7.98 m / s, and the confidence interval for wind power density is 395 to 445 W / m².

[0057] The specific embodiments described above demonstrate that the method of this invention can achieve high-precision and high-resolution assessment of wind resources in complex mountainous terrain, with verification errors significantly lower than those of traditional interpolation methods (RMSE typically > 1.2 m / s) and simple CFD simulations (MBE can reach ± ​​1 m / s). The assessment results not only provide reliable wind speed and wind energy data, but also offer multi-dimensional data support for wind turbine micro-location, turbulence load assessment, and power generation prediction, possessing significant engineering application value.

[0058] Finally, it should be emphasized that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for assessing wind resources in complex terrain based on multi-source data coupling, characterized in that, The method includes the following steps: S1. Multi-source data acquisition and fusion preprocessing: Simultaneously acquire SCADA operation data, laser wind radar point cloud data, and wind tower observation data of wind farms in the target complex terrain area. Perform time synchronization, coordinate unification, quality control, and standardization preprocessing on the acquired data to form a spatiotemporally consistent multi-source dataset. S2. Initial wind field construction based on terrain physical model: Import high-resolution digital elevation model, obtain high-resolution terrain data of the evaluation area, establish terrain adaptive three-dimensional computing grid, use RANS model to simulate three-dimensional steady-state wind field under multiple prevailing wind directions, and construct multi-wind direction initial wind field database. S3. Variational data assimilation wind field refinement correction: Using the multi-source data processed in step S1 as the observation value and the initial wind field obtained in step S2 as the background field, construct an objective function containing background and observation terms, and use the variational data assimilation algorithm to solve for a high-precision three-dimensional corrected wind field. S4. Multidimensional wind resource parameter extraction and evaluation: Extract multidimensional parameters such as wind speed, wind direction, turbulence intensity, and wind shear from the modified wind field, and construct a three-level evaluation index system that includes wind energy resource potential, wind energy quality, and development suitability. S5. Spatial Visualization and Output: Based on the geographic information system platform, realize spatial interpolation, classification and thematic mapping of wind resource parameters, and identify wind energy rich areas and suitable development areas; S6. Evaluation Result Verification and Uncertainty Quantification: The accuracy of the evaluation is verified by cross-validation using independent observation data, and the uncertainty of the evaluation results is quantified by ensemble perturbation method and Monte Carlo simulation.

2. The method according to claim 1, characterized in that, In step S1, the SCADA operating data includes at least the wind turbine pitch angle, generator speed, generator power, nacelle wind speed and wind direction; the laser wind measurement radar point cloud data is radial wind speed point cloud data acquired by three-dimensional scanning radar, covering key terrain and flow field feature areas of the assessment area; the wind measurement tower observation data includes at least multi-layer wind speed, wind direction, temperature, air pressure and turbulence intensity observations.

3. The method according to claim 1, characterized in that, In step S2, the high-resolution terrain data comes from a high-resolution digital elevation model or lidar terrain scanning data, with a spatial resolution of not less than 10m; the RANS model uses a RANS solver based on a k-ε or k-ω turbulence model, combined with terrain roughness and vegetation cover parameters, to simulate steady-state wind fields under multiple prevailing wind directions and construct an initial three-dimensional wind speed vector field.

4. The method according to claim 1, characterized in that, In step S3, the variational data assimilation algorithm uses either a three-dimensional variational (3D-Var) or a four-dimensional variational (4D-Var) algorithm; the observation items include lidar radial wind speed projection, SCADA data inversion equivalent wind speed, and wind tower multi-layer observation interpolation; in the background item, the background error covariance matrix is ​​adaptively constructed using terrain-dependent flow field features.

5. The method according to claim 1, characterized in that, In step S4, the wind resource assessment indicators of the three-level assessment index system include: annual average wind speed, wind power density, turbulence intensity, wind shear index, and effective wind time ratio; in step S5, a geographic information system platform is used to realize the spatial distribution map of wind resources, the identification of rich areas, and the classification of levels.

6. The method according to claim 1, characterized in that, The S6 steps are as follows: S61. Cross-validate using reserved independent wind measurement tower or wind turbine power data, and calculate the mean deviation (MBE), root mean square error (RMSE), and consistency index (IOA). S62. Using the ensemble perturbation method, random perturbations are applied to the observed data and model parameters to generate an assimilation result set, and the probability distribution and confidence interval of key evaluation indicators are statistically analyzed.

7. A complex terrain wind resource assessment system based on multi-source data coupling for implementing the method as described in any one of claims 1 to 6, characterized in that, The system includes a multi-source data acquisition and communication module (1), a terrain data processing and initial wind field modeling module (2), a variational data assimilation and wind field correction module (3), a wind resource assessment and visualization module (4), and an assessment, verification and uncertainty analysis module (5). The multi-source data acquisition and communication module (1) includes a SCADA data interface unit (11), a lidar point cloud receiving unit (12), and a wind tower data acquisition unit (13), which are used to synchronously acquire wind farm operation data, three-dimensional radial wind speed point cloud and multi-layer meteorological observation data, and perform time alignment and coordinate transformation. The terrain data processing and initial wind field modeling module (2) includes a high-resolution digital elevation model import unit (21), a terrain adaptive mesh generation unit (22), and a RANS wind field simulation unit (23). The RANS wind field simulation unit (23) is equipped with a k-ε or k-ω turbulence model solver, which is used to construct an initial three-dimensional wind field database based on terrain roughness and prevailing wind direction. The variational data assimilation wind field correction module (3) includes an objective function construction unit (31) and a variational assimilation calculation unit (32). The objective function construction unit (31) is used to construct an energy functional containing background field terms and observation terms. The variational assimilation calculation unit (32) uses a three-dimensional variational or four-dimensional variational algorithm to obtain a high-precision three-dimensional corrected wind field by minimizing the objective function. The wind resource assessment and visualization module (4) includes a wind resource parameter extraction unit (41), a three-level assessment index calculation unit (42), and a geographic information system mapping unit (43). The three-level assessment index calculation unit (42) constructs an assessment matrix based on wind energy resource potential, wind energy quality, and development suitability. The geographic information system mapping unit (43) is used to output a wind resource spatial distribution map and a rich area identification map. The evaluation verification and uncertainty analysis module (5) includes an accuracy verification unit (51) and an uncertainty quantification unit (52). The accuracy verification unit (51) uses independent observation data to calculate the average deviation, root mean square error and consistency index to evaluate the evaluation accuracy. The uncertainty quantification unit (52) uses the ensemble perturbation method and Monte Carlo simulation to apply random perturbations to the model parameters and observation data, and statistically analyzes the probability distribution and confidence interval of key evaluation indicators.

Images