Acousto-optic fusion driven wind energy resource map inversion method and system

Abstract

The present application relates to the technical field of wind energy resource assessment, and discloses a method and system for wind energy resource atlas inversion driven by sound-light fusion. The method comprises synchronously collecting and processing original sound waves and optical remote sensing data of a target region to generate a primary wind energy characteristic field containing terrain parameters. By performing hierarchical wind energy contribution assessment on the characteristic field, wind energy components dominated by terrain and atmospheric turbulence are separated. The above components are fused, and external measured atmospheric profile data is introduced as a constraint to perform constrained three-dimensional wind field reconstruction to generate a high-precision three-dimensional wind vector field. Based on this, wind energy power density is calculated, and after surface roughness correction, a high-resolution wind energy power density distribution field is obtained, and then a full-area wind energy resource atlas is drawn, and micro-siting simulation and wake assessment are completed. The method realizes high-precision and high-reliability assessment of wind energy resources in complex terrain areas.

Description

Technical Field

[0001] This invention relates to the field of wind energy resource assessment technology, specifically to a method and system for wind energy resource map inversion driven by acoustic-optical fusion. Background Technology

[0002] Existing wind energy resource assessment technologies mainly rely on single-point observations from meteorological towers, downscaling of numerical weather prediction models, and single-type remote sensing. While meteorological tower data offers high accuracy, its spatial representativeness is limited, making it difficult to meet the needs of wind field surveys in complex terrain areas. Numerical models can generate spatially continuous fields, but their simulation accuracy in complex terrain is limited by grid resolution and physical parameterization schemes. Single optical remote sensing methods such as lidar can invert horizontal wind fields, but lack the ability to directly detect key vertical structural information affecting wind energy quality, such as atmospheric turbulence intensity and thermal stratification.

[0003] A common drawback of existing technologies is their inability to effectively separate and quantify the different contributions of topographic dynamic processes and atmospheric turbulence to near-surface wind energy distribution under complex underlying surface conditions. The mechanical processes such as flow around and acceleration caused by topography are coupled with the thermal turbulence of the atmosphere itself. Traditional methods treat these as a mixed whole, making it difficult to analyze their respective mechanisms. This results in insufficient physical consistency of the constructed three-dimensional wind field, failing to accurately characterize the spatial differentiation of wind energy resources, and consequently reducing the reliability of wind power density estimation and resource mapping.

[0004] The technical problem to be solved by this invention is: how to synergistically utilize multi-dimensional information from acoustic and optical remote sensing to effectively separate and quantitatively evaluate the topographic-dominant and turbulence-dominant wind energy components in wind field inversion, and to introduce independent atmospheric vertical profile observation data to impose physical constraints on the fusion and reconstruction process, thereby achieving high-precision and high-reliability three-dimensional wind energy resource inversion and map generation in complex terrain areas. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for wind energy resource map inversion driven by acoustic-optical fusion, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a wind energy resource map inversion method driven by acoustic-optical fusion, the method comprising: Collect raw acoustic and optical remote sensing data within the target geographic area; The original acoustic data and original optical remote sensing data are processed to generate a primary wind energy feature field that includes terrain influence parameters. A stratified wind energy contribution assessment was performed on the primary wind energy characteristic field to distinguish between the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence. By integrating the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence, and introducing external measured atmospheric profile data as constraints, a constrained three-dimensional wind field reconstruction is performed to generate a three-dimensional wind vector field. Based on the three-dimensional wind vector field, the wind power density of each grid cell is calculated, and spatial diffusion correction is performed in combination with the surface roughness parameter to generate a high-resolution wind power density distribution field. The high-resolution wind power density distribution field is divided into multi-wind-direction sectors, and the wind frequency and average wind speed in each sector are statistically analyzed to draw a wind energy resource map of the entire region. Based on the wind energy resource map of the entire region, a micro-site selection simulation of wind turbine units is performed to generate a wind turbine location layout plan and a wake impact assessment report.

[0007] Preferably, the step of processing the original acoustic data and the original optical remote sensing data to generate a primary wind energy feature field including terrain influence parameters includes: Multi-scale acoustic energy spectrum decomposition processing is performed on the original acoustic data to decompose it into atmospheric turbulence acoustic energy spectrum components and background noise acoustic energy spectrum components. Temporal image difference enhancement processing is performed on the original optical remote sensing data to obtain the surface deformation feature sequence and the atmospheric motion trajectory sequence. The atmospheric turbulent acoustic energy spectrum component, the surface deformation characteristic sequence, and the atmospheric motion trajectory sequence are spatiotemporally synchronized and coupled with physical fields to construct an acoustic-optical synchronization feature body. Using the aforementioned acoustic-optical synchronization feature, adaptive noise suppression processing is performed on the background noise acoustic energy spectrum components to generate a pure acoustic energy perturbation field; The acoustic disturbance propagation path is extracted from the pure acoustic disturbance field and physically mapped to the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters.

[0008] Preferably, the step of performing multi-scale acoustic spectrum decomposition on the original acoustic data to decompose it into atmospheric turbulence acoustic spectrum components and background noise acoustic spectrum components includes: Perform a Fourier transform on the original acoustic wave data to obtain its frequency domain representation; A turbulence characteristic frequency range that matches the atmospheric turbulence physical model is preset, and frequency components falling within the turbulence characteristic frequency range and their corresponding amplitudes and phases are extracted in the frequency domain representation; The frequency components and their corresponding amplitudes and phases are subjected to inverse Fourier transform to reconstruct the atmospheric turbulent acoustic energy spectrum components; The frequency components corresponding to the atmospheric turbulent acoustic energy spectrum components are removed from the frequency domain representation of the original acoustic data to obtain the remaining frequency domain components. Wavelet threshold denoising is performed on the remaining frequency domain components to filter out high-frequency random noise and obtain smooth remaining frequency domain components. Perform an inverse Fourier transform on the smoothed residual frequency domain components to reconstruct the background noise acoustic energy spectrum components; The step of performing time-series image differential enhancement processing on the original optical remote sensing data to obtain a land surface deformation feature sequence and an atmospheric motion trajectory sequence includes: Optical remote sensing images of the target geographic area at multiple consecutive times are acquired to form the original image time series; Radiometric calibration and atmospheric correction are performed on each optical remote sensing image in the original image time series to obtain a standard surface reflectance image; In the standard surface reflectance image sequence, the reflectance difference value at the same pixel position in two consecutive images is calculated; A threshold is set for the reflectivity difference value, and pixels whose absolute difference value exceeds the threshold are marked as potentially changed pixels; Morphological filtering and region growing are performed on the potential changed pixels to extract connected regions as surface deformation patches, and the surface deformation feature sequence is composed of the surface deformation patches at all times. Meanwhile, in the standard surface reflectance image sequence, images of specific bands are selected, and the motion vectors of atmospheric clouds or aerosols between adjacent images are calculated using the optical flow method. The motion vectors at multiple consecutive moments are connected to form the atmospheric motion trajectory sequence.

[0009] Preferably, the step of performing spatiotemporal synchronization and physical field coupling processing on the atmospheric turbulent acoustic energy spectrum components, the surface deformation characteristic sequence, and the atmospheric motion trajectory sequence to construct an acoustic-optical synchronization feature body includes: Establish a unified spatial geographic coordinate system and timestamp, and transform the acoustic detection point positions of the atmospheric turbulence acoustic energy spectrum components, the pixel positions of the surface deformation feature sequences, and the trajectory point positions of the atmospheric motion trajectory sequences to the unified spatial geographic coordinate system. Based on the timestamp, time-aligned interpolation is performed on the atmospheric turbulence acoustic energy spectrum component, the surface deformation feature sequence, and the atmospheric motion trajectory sequence to ensure that the three data have corresponding spatial descriptions at the same time point; At the data points where spatiotemporal synchronization is completed, the acoustic energy intensity of the atmospheric turbulent acoustic energy spectrum component, the deformation intensity of the surface deformation characteristic sequence, and the motion velocity vector of the atmospheric motion trajectory sequence are used as coupled physical quantities. Based on the fluid dynamics equations, a constitutive model of the coupled physical quantities is established. This constitutive model is used to describe the energy transfer and momentum exchange relationships between acoustic energy fluctuations, surface deformation, and atmospheric motion. Using the constitutive relation model, the spatiotemporally synchronized data is jointly solved to generate a multidimensional data volume containing acoustic energy field, deformation field and motion vector field. This multidimensional data volume is the acoustic-optical synchronization feature volume.

[0010] Preferably, the step of using the acoustic-optical synchronization feature to perform adaptive noise suppression processing on the background noise acoustic energy spectrum components to generate a pure acoustic energy perturbation field includes: Extract the local deformation intensity and local motion velocity that correspond in time and space to the background noise acoustic energy spectrum components from the acoustic-optical synchronization feature body; Analyze the statistical correlation between the local deformation intensity, local motion velocity, and the energy values ​​of the background noise acoustic energy spectrum components; Based on the statistical correlation, an adaptive filter is constructed, and the transfer function of the adaptive filter is dynamically adjusted by the instantaneous values ​​of the local deformation intensity and the local motion velocity. The background noise acoustic energy spectrum component is input into the adaptive filter, and the adaptive filter dynamically attenuates random noise energy that is unrelated to surface or atmospheric motion based on the local deformation intensity and local motion velocity at the current spatiotemporal point. Energy compensation is performed on the filtered signal to retain the acoustic energy components related to the actual atmospheric disturbances, and the output signal after noise suppression is the pure acoustic energy disturbance field.

[0011] Preferably, the step of extracting the acoustic disturbance propagation path from the pure acoustic disturbance field and physically mapping it with the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters includes: In the pure acoustic energy disturbance field, the propagation trajectory of the peak acoustic energy intensity in space and time is traced, and the propagation trajectory is defined as the acoustic energy disturbance propagation path; Calculate the propagation direction and speed at each point along the propagation path of the acoustic disturbance; The acoustic disturbance propagation path is superimposed onto the digital elevation model corresponding to the surface deformation feature sequence, and the terrain slope, aspect and roughness length at each point on the path are calculated. Based on the physical model of sound wave propagation in non-uniform atmosphere and complex terrain, and combined with the propagation direction, propagation speed, terrain slope, aspect and roughness length, the equivalent wind speed profile along the sound energy disturbance propagation path is inverted. The equivalent wind speed profile, along with the corresponding terrain slope, aspect, and roughness length, are encoded together into a data structure with spatial location information. This data structure is the primary wind energy feature field.

[0012] Preferably, the step of performing a stratified wind energy contribution assessment on the primary wind energy characteristic field, distinguishing between the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence, includes: In the primary wind energy characteristic field, the set of parameters characterizing the influence of topography and the parameter characterizing the equivalent wind speed are separated. A preset terrain influence weighting function is used, which takes terrain slope, aspect and roughness length as inputs to calculate the contribution coefficient of terrain to local wind speed. Multiply the equivalent wind speed by the contribution coefficient to obtain the terrain-modulated wind speed, and define the wind energy corresponding to the terrain-modulated wind speed as the terrain-dominated wind energy component. From the total wind energy of the primary wind energy characteristic field, the wind energy component dominated by topography is subtracted, and the remaining wind energy is defined as the wind energy component dominated by atmospheric turbulence. The spatial distribution characteristics of the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence were statistically analyzed separately.

[0013] Preferably, the fusion of the topographically dominated wind energy component and the atmospheric turbulence-dominated wind energy component, and the introduction of external measured atmospheric profile data as constraints, performs constrained three-dimensional wind field reconstruction to generate a three-dimensional wind vector field, including: The wind energy components dominated by topography and wind energy components dominated by atmospheric turbulence in space are weighted and superimposed according to their energy values ​​to obtain a fused two-dimensional wind energy field. Acquire measured atmospheric profile data provided by meteorological towers or radiosondes within the target area, wherein the measured atmospheric profile data includes wind speed and wind direction at different altitude levels; The fused two-dimensional wind energy field is used as the prior distribution of the horizontal wind field intensity, and the measured atmospheric profile data is used as the hard constraint condition in the vertical direction. A three-dimensional wind field reconstruction variational model is established. The objective function of the variational model includes: making the reconstructed three-dimensional wind field as close as possible to the prior distribution in the horizontal direction, and passing through the constraint points of the measured atmospheric profile data in the vertical section, while the entire wind field satisfies the fluid continuity and boundary layer physical laws. Solving the three-dimensional wind field reconstruction variational model yields the wind speed vector at each three-dimensional grid point in space, thus forming the three-dimensional wind vector field.

[0014] Preferably, the step of performing a micro-site selection simulation of wind turbines based on the full-area wind energy resource map to generate a wind turbine site layout plan and a wake impact assessment report includes: On the entire region's wind energy resource map, potential turbine locations with wind power density exceeding the engineering development threshold were identified; Within the potential machine location area, the hub height and rotor diameter of the fan are determined according to the preset single-unit capacity and fan model; With the goal of maximizing total power generation, an iterative simulation of automatic wind turbine location layout is performed within the potential turbine location area. In each iteration, the wake effect between any two locations based on the prevailing wind direction is calculated, and the power generation loss caused by the wake is deducted. When the increase in total power generation is less than a preset threshold, the iteration stops, and the final set of spatial coordinates of the wind turbine locations is obtained, forming the wind turbine location layout scheme. Based on the wind turbine location layout scheme and the three-dimensional wind vector field, a computational fluid dynamics wake model is run to simulate the wake superposition effect of all wind turbines under typical wind conditions. The increase in turbulence intensity and the annual power generation reduction coefficient of each wind turbine are output to form the wake impact assessment report.

[0015] Preferably, the present invention also includes an acoustic-optical fusion-driven wind energy resource map inversion system, the system including a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the acoustic-optical fusion-driven wind energy resource map inversion method as described above.

[0016] Compared with the prior art, the beneficial effects of the present invention are: By simultaneously acquiring and fusing raw acoustic and optical remote sensing data, a primary wind energy characteristic field containing topographic influence parameters is generated. Acoustic data includes atmospheric turbulence and thermal information, while optical data provides wind vector information; the two are fused in the early processing stage. This technique constructs an initial field that simultaneously contains kinematic and dynamic characteristics and incorporates topographic effects. This overcomes the limitations of single data sources, resulting in a more complete characterization of the interaction between the near-surface wind field and the complex underlying surface, providing a more comprehensive data foundation and more clearly defined physical properties for subsequent mechanism separation.

[0017] A stratified wind energy contribution assessment was performed, distinguishing between topographically dominated and atmospheric turbulence-dominated wind energy components, and externally measured atmospheric profiles were introduced for constrained 3D wind field reconstruction. First, the total wind energy contribution was physically analyzed into topographic dynamic components and atmospheric turbulence components. When reconstructing and fusing each component in 3D, the introduced independent measured atmospheric vertical profile data provided objective physical constraints. This ensured that the vertical structure and thermodynamic properties of the reconstructed 3D wind vector field remained consistent with the actual atmospheric conditions, improving the accuracy and physical plausibility of the wind field inversion results under complex flow field conditions.

[0018] Based on the aforementioned highly physically consistent three-dimensional wind vector field, wind power density is calculated and corrected for surface roughness to obtain a high-resolution wind power density distribution field. This distribution field more accurately reflects the microscopic spatial differences in wind energy resources. Based on this, wind direction sector statistics and microscopic site selection simulations result in wind resource maps, wind turbine placement schemes, and wake assessments, all of which demonstrate improved accuracy and engineering applicability. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the working principle of the acoustic-optical fusion-driven wind energy resource map inversion method described in this invention. Figure 2 A flowchart for generating the primary wind energy characteristic field; Figure 3 A flowchart for generating a clean acoustic energy perturbation field for adaptive noise suppression processing; Figure 4 A diagram showing the wind turbine location layout scheme driven by wind resource maps; Figure 5 The image shows the heat map of the superposition effect coefficient of the wake of each wind turbine under different typical wind conditions. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1This invention provides a method for wind energy resource map inversion driven by acoustic-optical fusion. The method includes: First, simultaneously acquiring raw acoustic data and raw optical remote sensing data within a target geographical area. Then, performing a series of processes on the raw acoustic and optical remote sensing data to generate a primary wind energy feature field containing parameters such as topographic slope, aspect, and roughness length. Next, performing a layered wind energy contribution assessment on the primary wind energy feature field to distinguish between wind energy components dominated by topography and those dominated by atmospheric turbulence. Then, fusing the two wind energy components and introducing external measured atmospheric profile data as a vertical constraint, performing a constrained three-dimensional wind field reconstruction to generate a three-dimensional wind vector field covering the target area. Based on this three-dimensional wind vector field, calculating the wind power density of each grid cell and performing spatial diffusion correction using surface roughness parameters to generate a high-resolution wind power density distribution field. Finally, dividing this distribution field into multi-wind-direction sectors, statistically analyzing the wind frequency and average wind speed within each sector, and drawing a wind energy resource map of the entire region. Finally, based on the wind energy resource map of the entire region, a micro-site selection simulation of wind turbine units was performed to generate a wind turbine location layout plan and a wake impact assessment report.

[0022] Example 1: See Figure 2 Raw acoustic wave data and raw optical remote sensing data were collected within the target geographic area. Multi-scale acoustic spectrum decomposition was performed on the raw acoustic wave data, decomposing it into atmospheric turbulence acoustic spectrum components and background noise acoustic spectrum components. This process specifically includes: performing a Fourier transform on the raw acoustic wave data to obtain its frequency domain representation; pre-setting a turbulence characteristic frequency range matching the atmospheric turbulence physical model; extracting frequency components falling within this turbulence characteristic frequency range and their corresponding amplitudes and phases from the frequency domain representation; then performing an inverse Fourier transform on these frequency components and their corresponding amplitudes and phases to reconstruct the atmospheric turbulence acoustic spectrum components; removing the frequency components corresponding to the atmospheric turbulence acoustic spectrum components from the frequency domain representation of the raw acoustic wave data to obtain the remaining frequency domain components; performing wavelet threshold denoising on the remaining frequency domain components to filter out high-frequency random noise to obtain smoothed remaining frequency domain components; and finally performing an inverse Fourier transform on the smoothed remaining frequency domain components to reconstruct the background noise acoustic spectrum components.

[0023] The process involves performing temporal image differential enhancement on the raw optical remote sensing data to obtain a land deformation feature sequence and an atmospheric motion trajectory sequence. Specifically, this includes: acquiring optical remote sensing images of the target geographic area at multiple consecutive time points to form a raw image time series; performing radiometric calibration and atmospheric correction on each optical remote sensing image in the raw image time series to obtain a standard land reflectance image; calculating the reflectance difference value at the same pixel location between two consecutive images in the standard land reflectance image sequence; setting a threshold for the reflectance difference value and marking pixels with absolute difference values ​​exceeding the threshold as potentially changed pixels; performing morphological filtering and region growing on the potentially changed pixels to extract connected regions as land deformation patches; and constructing a land deformation feature sequence from the land deformation patches at all time points. Simultaneously, selecting images in specific bands from the standard land reflectance image sequence and using optical flow to calculate the motion vectors of atmospheric clouds or aerosols between adjacent images; and connecting the motion vectors at multiple consecutive time points to form an atmospheric motion trajectory sequence. The obtained atmospheric turbulent acoustic energy spectrum components, surface deformation feature sequences, and atmospheric motion trajectory sequences are spatiotemporally synchronized and coupled with physical fields to construct an acoustic-optical synchronization feature body. This feature body is then used to perform adaptive noise suppression processing on the background noise acoustic energy spectrum components, generating a clean acoustic energy disturbance field. The acoustic energy disturbance propagation path is extracted from the clean acoustic energy disturbance field and physically mapped to the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters.

[0024] In practical implementation, the acoustic-optical fusion-driven wind energy resource map inversion method involves collecting raw acoustic data and raw optical remote sensing data within the target geographical area. The raw acoustic data is collected by a distributed acoustic sensor array deployed in the target area, while the raw optical remote sensing data is acquired by optical sensors mounted on satellites or high-altitude flight platforms; both are recorded synchronously in time. Multi-scale acoustic spectrum decomposition processing is performed on the raw acoustic data, decomposing it into atmospheric turbulence acoustic spectrum components and background noise acoustic spectrum components. In the example scenario, monitoring a hilly area with a complex underlying surface, the specific process of multi-scale acoustic spectrum decomposition is as follows: a fast Fourier transform is performed on the raw acoustic data to obtain its frequency domain representation. A turbulence characteristic frequency range matching the atmospheric turbulence physical model is set, for example, 0.1 Hz to 1 Hz. All frequency components falling within this range, along with the amplitude and phase of each frequency component, are extracted from the frequency domain representation. The extracted frequency components, amplitudes, and phases are reconstructed into a time-domain signal using inverse Fourier transform. This reconstructed time-domain signal is the atmospheric turbulence acoustic energy spectrum component. From the complete frequency domain representation of the original acoustic data, the frequency components in the range of 0.1 Hz to 1 Hz are set to zero to obtain the remaining frequency domain components. Wavelet thresholding denoising is performed on the remaining frequency domain components. A 5-level decomposition is performed using the Daubechies wavelet basis, and a soft thresholding function is applied to the detail coefficients of each level to filter out high-frequency random noise, resulting in smoothed remaining frequency domain components. Inverse Fourier transform is performed on the smoothed remaining frequency domain components to reconstruct the background noise acoustic energy spectrum component. In some embodiments, the extraction of frequency components can be achieved using a selective transfer function, the mathematical expression of which is: , in: This represents the frequency domain representation of the original acoustic wave data. It is a window function that is 1 within a preset turbulence characteristic frequency range and 0 outside of that range. The extracted frequency components are then inversely transformed to obtain the atmospheric turbulent acoustic energy spectrum components.

[0025] In practice, temporal image differential enhancement processing is performed on the raw optical remote sensing data to obtain a land deformation feature sequence and an atmospheric motion trajectory sequence. Specifically, optical remote sensing images of the target hilly area are acquired at 24 consecutive time points with a 1-hour time interval, forming the raw image time series. Radiometric calibration and atmospheric correction processing are performed on each optical remote sensing image in the raw image time series, converting the digital quantization values ​​of the images into land reflectance to eliminate the effects of atmospheric scattering and absorption, resulting in a standard land reflectance image. In the standard land reflectance image sequence, the reflectance difference between the image at time t+1 and the image at time t is calculated at the same pixel location. A threshold is set for the reflectance difference, for example, 0.05, and pixels with an absolute difference value exceeding 0.05 are marked as potentially changed pixels. Morphological filtering was performed on all marked potentially variable pixels to remove isolated noise points. Then, region growing was performed to merge spatially adjacent potentially variable pixels, extracting connected regions as surface deformation patches. A surface deformation feature sequence was constructed from all surface deformation patches at 24 time points, with each element recording the spatial distribution of surface deformation at the corresponding time. Simultaneously, specific band images sensitive to water vapor or aerosols were selected from the standard surface reflectance image sequence. The Horn-Schunck optical flow method was used to calculate the motion vector of atmospheric clouds or aerosols at each pixel location between adjacent images. Connecting the motion vectors of feature points of the same air mass at multiple consecutive time points formed multiple sequences of atmospheric motion trajectories reflecting atmospheric transport paths.

[0026] In the specific implementation, the obtained atmospheric turbulence acoustic energy spectrum components, surface deformation feature sequences, and atmospheric motion trajectory sequences are subjected to spatiotemporal synchronization and physical field coupling processing. A unified spatial geographic coordinate system and timestamp are established. The spatial geographic coordinate system adopts the WGS84 geographic coordinate system, and the timestamp adopts Coordinated Universal Time (UTC). The acoustic detection point positions of the atmospheric turbulence acoustic energy spectrum components, the pixel positions of the surface deformation feature sequences, and the trajectory point positions of the atmospheric motion trajectory sequences are all transformed to the WGS84 geographic coordinate system through coordinate transformation. Using UTC as the reference, time-aligned interpolation is performed on the atmospheric turbulence acoustic energy spectrum components, the surface deformation feature sequences, and the atmospheric motion trajectory sequences. Cubic spline interpolation is used to ensure that the three types of data have corresponding spatial descriptions at every hour. At the spatiotemporally synchronized data points, the acoustic energy intensity of the atmospheric turbulence acoustic energy spectrum components, the deformation intensity of the surface deformation feature sequences, and the motion velocity vector of the atmospheric motion trajectory sequences are used as coupling physical quantities. Based on a simplified set of fluid dynamics governing equations, a constitutive model is established to describe the energy transfer and momentum exchange relationships between acoustic energy fluctuations, surface deformation, and atmospheric motion. The model's construction begins by defining the coupled physical quantities at spatiotemporally synchronized data points: the acoustic intensity of the atmospheric turbulent acoustic energy spectrum components, the deformation intensity of the surface deformation characteristic sequence, and the velocity vector of the atmospheric motion trajectory sequence. These quantities are aligned using a unified spatial geographic coordinate system and timestamps before being incorporated into the model framework. The simplified fluid dynamics governing equations are applied to describe the fundamental laws of atmospheric motion, treating acoustic energy fluctuations as an energy source and surface deformation as a boundary condition influence term, thus establishing the energy transfer and momentum exchange relationships between acoustic energy fluctuations, surface deformation, and atmospheric motion. During model construction, higher-order nonlinear terms or specific thermodynamic effects in the fluid dynamics equations are ignored for simplification, allowing the model to focus on key physical mechanisms such as acoustic energy propagation loss in turbulent media, the modulation of airflow dynamic drag by surface deformation, and the conversion between atmospheric momentum and acoustic energy. By using a constitutive relation model to jointly solve the spatiotemporally synchronized data, and using the least squares method to optimize the solution, a multidimensional data volume containing acoustic energy field, deformation field and motion vector field is generated. This multidimensional data volume is the acoustic-optical synchronization feature volume.

[0027] In practical implementation, adaptive noise suppression processing is performed on the background noise acoustic energy spectrum component using an acoustic-optical synchronization feature. From the acoustic-optical synchronization feature, based on timestamps and geographic coordinates, the local deformation intensity and local motion velocity, which are spatiotemporally completely corresponding to each data point of the background noise acoustic energy spectrum component, are extracted. The statistical correlation between the local deformation intensity, local motion velocity, and the energy values ​​of the background noise acoustic energy spectrum component is analyzed, and the Pearson correlation coefficient matrix is ​​calculated. Based on the statistical correlation analysis results, an adaptive filter is constructed. The parameters of the adaptive filter's transfer function are dynamically adjusted by the instantaneous values ​​of the local deformation intensity and local motion velocity. When the local deformation intensity or local motion velocity increases, the attenuation degree of the background noise acoustic energy spectrum component by the filter decreases. The background noise acoustic energy spectrum component is input into the adaptive filter, which dynamically attenuates random noise energy unrelated to surface or atmospheric motion based on the local deformation intensity and local motion velocity at the current spatiotemporal point. Energy compensation is performed on the signal processed by the adaptive filter. The compensation factor is calculated based on the average energy loss of the suppressed frequency band to retain the acoustic energy components related to real atmospheric disturbances. The output signal, after noise suppression and energy compensation, is the pure acoustic energy disturbance field. Compared with the background noise acoustic energy spectrum component, the coherence of the pure acoustic energy disturbance field with the atmospheric turbulence acoustic energy spectrum component is significantly improved.

[0028] In practice, the acoustic disturbance propagation path is extracted from the pure acoustic disturbance field and physically mapped to the surface deformation feature sequence. Within the pure acoustic disturbance field, the peak point of the acoustic intensity over time is identified, and its movement trajectory in three-dimensional spatiotemporal coordinates is tracked, defining the trajectory as the acoustic disturbance propagation path. The propagation direction and velocity at each point along the propagation path are calculated. The propagation direction is determined by the direction of the line connecting adjacent points, and the propagation velocity is obtained by dividing the distance between points by the time interval. The acoustic disturbance propagation path is superimposed onto the digital elevation model (DEM) corresponding to the surface deformation feature sequence. The DEM has the same spatial resolution as the optical remote sensing data. Based on the DEM, the terrain slope, aspect, and roughness length at each point along the acoustic disturbance propagation path are calculated. The roughness length is assigned a value based on the surface cover type in the surface deformation feature sequence, obtained by looking up a table. Based on the physical model of sound wave refraction and diffraction in a non-uniform atmosphere and complex terrain, and combining the propagation direction, propagation speed, terrain slope, aspect, and roughness length, the equivalent wind speed profile along the sound energy disturbance propagation path is derived. The equivalent wind speed profile represents the average wind speed profile along the path. The equivalent wind speed profile, along with the corresponding terrain slope, aspect, and roughness length, are encoded together into a data structure with geospatial location information. This data structure is the primary wind energy characteristic field containing terrain influence parameters.

[0029] Example 2: See Figure 3By performing spatiotemporal synchronization and physical field coupling processing on atmospheric turbulent acoustic energy spectrum components, surface deformation characteristic sequences, and atmospheric motion trajectory sequences, an acoustic-optical synchronization feature body is constructed. This process includes: establishing a unified spatial geographic coordinate system and timestamp; transforming the acoustic detection point locations of the atmospheric turbulent acoustic energy spectrum component, the pixel locations of the surface deformation feature sequence, and the trajectory point locations of the atmospheric motion trajectory sequence to this unified spatial geographic coordinate system; performing time-aligned interpolation on the atmospheric turbulent acoustic energy spectrum component, the surface deformation feature sequence, and the atmospheric motion trajectory sequence based on the timestamp to ensure that the three data have corresponding spatial descriptions at the same time point; using the acoustic energy intensity of the atmospheric turbulent acoustic energy spectrum component, the deformation intensity of the surface deformation feature sequence, and the motion velocity vector of the atmospheric motion trajectory sequence as coupled physical quantities at the spatiotemporally synchronized data points; establishing a constitutive relationship model between the coupled physical quantities based on the fluid dynamics equations to describe the energy transfer and momentum exchange relationship between acoustic energy fluctuations, surface deformation, and atmospheric motion; and using this constitutive relationship model to jointly solve the spatiotemporally synchronized data to generate a multidimensional data volume containing the acoustic energy field, deformation field, and motion vector field, namely the acoustic-optical synchronization feature volume. After constructing the acoustic-optical synchronization feature, adaptive noise suppression processing is performed on the background noise acoustic energy spectrum component using this feature to generate a clean acoustic energy disturbance field. This process includes: extracting the local deformation intensity and local motion velocity corresponding to the background noise acoustic energy spectrum component in time and space from the acoustic-optical synchronization feature; analyzing the statistical correlation between the local deformation intensity, local motion velocity, and the energy values ​​of the background noise acoustic energy spectrum component; constructing an adaptive filter based on this statistical correlation; dynamically adjusting the transfer function of the adaptive filter by the instantaneous values ​​of the local deformation intensity and local motion velocity; inputting the background noise acoustic energy spectrum component into the adaptive filter; dynamically attenuating random noise energy unrelated to surface or atmospheric motion according to the local deformation intensity and local motion velocity at the current time and space point; performing energy compensation on the filtered signal to retain the acoustic energy components related to the real atmospheric disturbance; and outputting the noise-suppressed signal as the clean acoustic energy disturbance field.

[0030] In practical implementation, the atmospheric turbulent acoustic energy spectrum components, surface deformation feature sequences, and atmospheric motion trajectory sequences are spatiotemporally synchronized and coupled with physical fields to construct an acoustic-optical synchronization feature body. This process is implemented as follows: A unified spatial geographic coordinate system and timestamps are established. The spatial geographic coordinate system adopts the UTM projection coordinate system, and the timestamps are second-level timestamps counted from the start of data acquisition. The acoustic detection point locations of the atmospheric turbulent acoustic energy spectrum components, the pixel locations of the surface deformation feature sequences, and the trajectory point locations of the atmospheric motion trajectory sequences are all transformed to the unified spatial geographic coordinate system through coordinate projection transformation. Using second-level timestamps as a benchmark, time-aligned interpolation was performed on the atmospheric turbulent acoustic energy spectrum component, the surface deformation characteristic sequence, and the atmospheric motion trajectory sequence. The time sampling rate of the atmospheric turbulent acoustic energy spectrum component was 10 Hz, the time interval of the surface deformation characteristic sequence was 1 hour, and the atmospheric motion trajectory sequence was composed of optical flow vectors connected every minute. A linear interpolation method was used to uniformly resample all data to a 1 Hz time point, ensuring that at each 1 Hz time point, the atmospheric turbulent acoustic energy spectrum component, the surface deformation characteristic sequence, and the atmospheric motion trajectory sequence all had corresponding spatial descriptions. At the data points where spatiotemporal synchronization was achieved, the acoustic energy intensity of the atmospheric turbulent acoustic energy spectrum component, the deformation intensity of the surface deformation characteristic sequence, and the motion velocity vector of the atmospheric motion trajectory sequence were used as coupled physical quantities. Based on the simplified forms of the Navier-Stokes equations and the acoustic wave equations, a constitutive relationship model between the coupled physical quantities was established. The constitutive relationship model is specifically expressed as a set of simultaneous equations describing the energy transfer and momentum exchange relationships between acoustic energy fluctuations, surface deformation, and atmospheric motion. The constitutive relation model is used to jointly solve the spatiotemporally synchronized data. An iterative optimization algorithm is used to solve the simultaneous equations to generate a multidimensional data volume containing the acoustic energy field, deformation field and motion vector field. This multidimensional data volume is the acoustic-optical synchronization feature volume. Each data point in the acoustic-optical synchronization feature volume is associated with geographic coordinates, time, acoustic energy intensity, deformation intensity and three-dimensional motion velocity vector.

[0031] In specific implementation, the acoustic-optical synchronization feature body is used to perform adaptive noise suppression processing on the background noise acoustic energy spectrum component and generate a pure acoustic energy perturbation field. This step is implemented as follows: Local deformation intensity and local motion velocity, which are completely spatiotemporally corresponding to the background noise acoustic energy spectrum component, are extracted from the acoustic-optical synchronization feature body. The statistical correlation between the local deformation intensity, local motion velocity, and the energy values ​​of the background noise acoustic energy spectrum component is analyzed, and the nonlinear dependence among the three is quantified by calculating mutual information entropy. Based on the statistical correlation analysis results, an adaptive filter is constructed. The transfer function of the adaptive filter is dynamically adjusted by the instantaneous values ​​of the local deformation intensity and local motion velocity. The dynamic adjustment rule of the transfer function can be described as follows: , in: This represents the filter adjustment factor at time n. This represents the local deformation intensity at time n. This represents the local velocity vector at time n. For its modulus length, It is a reference value with velocity dimensions. and The dimensionless weighting coefficients are pre-calibrated through correlation analysis. The background noise acoustic energy spectrum components are input into the adaptive filter, which is adjusted based on the local deformation intensity and local motion velocity at the current spatiotemporal point. The filter's cutoff frequency and attenuation amplitude are dynamically adjusted to attenuate random noise energy unrelated to surface or atmospheric motion. Energy compensation is then applied to the filtered signal, with the gain value and adjustment factor used for energy compensation. The mutual information entropy is inversely proportional to the noise level, thus preserving the acoustic energy components related to the actual atmospheric disturbances. The output signal, after noise suppression and energy compensation, is the pure acoustic energy disturbance field, which exhibits clearer characteristic spectral lines related to atmospheric physical processes. In some embodiments, the mutual information entropy can be calculated using histogram estimation, based on the discretized joint probability distribution of local deformation intensity, local velocity modulus, and the energy values ​​of the background noise acoustic energy spectrum components. In some embodiments, the spatiotemporal synchronization process has specific requirements for computational resources; the time alignment interpolation step can be executed in parallel on a distributed computing cluster, with each computing node responsible for data synchronization within a specific spatiotemporal subdomain. In specific implementations, the establishment of the constitutive model depends on specific fluid assumptions; for the near-surface boundary layer, the constitutive model can adopt a simplified form suitable for neutral atmospheric stratification.

[0032] Example 3: Extract the acoustic disturbance propagation path from the pure acoustic disturbance field and perform physical spatial mapping with the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters. This step is implemented as follows: trace the propagation trajectory of the peak acoustic intensity in spacetime within the pure acoustic disturbance field, define this trajectory as the acoustic disturbance propagation path, calculate the propagation direction and velocity at each point on the propagation path, superimpose the acoustic disturbance propagation path onto the digital elevation model corresponding to the surface deformation feature sequence, calculate the topographic slope, aspect, and roughness length at each point on the path, and, based on the physical model of sound wave propagation in a non-uniform atmosphere and complex terrain, combine the obtained propagation direction, propagation velocity, topographic slope, aspect, and roughness length to inversely derive the equivalent wind speed profile along the acoustic disturbance propagation path. Encode this equivalent wind speed profile and the corresponding topographic slope, aspect, and roughness length together into a data structure with spatial location information; this data structure is the primary wind energy feature field.

[0033] In practical implementation, the propagation path of acoustic disturbance is extracted from the pure acoustic disturbance field and physically mapped to the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters. The specific implementation process of this series of operations is as follows: In the pure acoustic disturbance field, local extreme points where the acoustic intensity exceeds a preset threshold are identified in each time slice. These extreme points are used as tracking targets. By comparing the spatial correlation of extreme points on consecutive time slices, multiple trajectories moving continuously in the spatiotemporal coordinate system are formed. Each trajectory is defined as an acoustic disturbance propagation path. The propagation direction and propagation speed of each path point on the acoustic disturbance propagation path are calculated. The propagation direction is determined by the projection angle of the line connecting the current path point and the previous path point on the horizontal plane. The propagation speed is obtained by dividing the three-dimensional spatial straight-line distance between the current path point and the previous path point by the time difference between the two points. In practical implementation, taking a mountainous area as an example, the tracking of the acoustic disturbance propagation path is carried out within a 6-hour data window with a time resolution of 1 second. Spatially, several clear trajectories moving from the valley towards the ridge are tracked.

[0034] In practice, the acoustic disturbance propagation path is superimposed onto the digital elevation model (DEM) corresponding to the surface deformation feature sequence. The DEM has the same 30-meter spatial grid resolution as the optical remote sensing data. For each path point on each acoustic disturbance propagation path, the elevation value at the center coordinates of the path point is obtained from the DEM using bilinear interpolation. Based on the elevation values ​​of the path point's location and its adjacent grids, the terrain slope, aspect, and roughness length at the path point's location are calculated. The terrain slope is obtained by fitting the elevation surface within a 3x3 window and calculating the maximum slope value. The aspect is the horizontal azimuth angle of the direction of maximum slope decrease. The roughness length is assigned based on the surface cover type identified in the surface deformation feature sequence at the same location, by referring to a preset typical surface roughness length lookup table. For example, the roughness length corresponding to forest is 0.8 meters, and that of grassland is 0.03 meters. In some embodiments, the terrain slope can be calculated by directly estimating the east-west and north-south elevation gradient components using the finite difference method.

[0035] In practical implementation, based on the physical model of sound wave propagation in a non-uniform atmosphere and complex terrain, and considering the propagation direction, propagation speed, terrain slope, aspect, and roughness length, the equivalent wind speed profile along the sound energy disturbance propagation path is inverted. The sound wave propagation physical model used takes into account the wind speed acceleration effect caused by terrain and the correction of the near-surface wind speed profile by surface roughness. The core relationship for the equivalent wind speed profile inversion can be expressed as: , in: The equivalent wind speed represented by the inverse calculation, Represents the observed propagation speed of acoustic energy disturbance. Represents the slope of the terrain. Represents the slope aspect of the terrain. Represents roughness length, function It encapsulates the physical relationship between sound wave propagation speed and meteorological wind speed, topographic dynamic lifting correction, and logarithmic wind speed profile adjustment based on roughness length. The equivalent wind speed profile, along with the corresponding topographic slope, aspect, and roughness length at each point, are encoded into a data structure with spatial location information. This data structure is the primary wind energy feature field. In this example, the primary wind energy feature field is stored as geospatial vector line features. The geometry of each line feature represents the sound energy disturbance propagation path, and the attribute table sequentially records the equivalent wind speed, topographic slope, aspect, and roughness length at each path point. In some embodiments, the function... The specific form can be calibrated by conducting simultaneous acoustic and meteorological tower observations on typical terrain.

[0036] Example 4: A layered wind energy contribution assessment is performed on the primary wind energy characteristic field to distinguish between the topographically dominated wind energy component and the atmospheric turbulence-dominated wind energy component. This assessment process includes: separating the parameter set representing the topographic influence and the parameter representing the equivalent wind speed from the primary wind energy characteristic field; pre-setting a topographic influence weighting function, which calculates the contribution coefficient of the topography to the local wind speed using topographic slope, aspect, and roughness length as inputs; multiplying the equivalent wind speed by the contribution coefficient to obtain the topographically modulated wind speed; defining the wind energy corresponding to the topographically modulated wind speed as the topographically dominated wind energy component; subtracting the topographically dominated wind energy component from the total wind energy in the primary wind energy characteristic field, and defining the remaining wind energy as the atmospheric turbulence-dominated wind energy component; and statistically analyzing the spatial distribution characteristics of the topographically dominated wind energy component and the atmospheric turbulence-dominated wind energy component. Subsequently, the topographically dominated wind energy component and the atmospheric turbulence-dominated wind energy component are fused, and external measured atmospheric profile data are introduced as constraints to perform a constrained three-dimensional wind field reconstruction to generate a three-dimensional wind vector field. The reconstruction process includes: weighting and superimposing the wind energy components dominated by topography and wind energy components dominated by atmospheric turbulence in the spatially distributed area according to their energy values ​​to obtain a fused two-dimensional wind energy field; acquiring measured atmospheric profile data provided by meteorological towers or radiosondes in the target area, which includes wind speed and direction at different altitudes; using the fused two-dimensional wind energy field as the prior distribution of horizontal wind field intensity and the measured atmospheric profile data as the hard constraint condition in the vertical direction; establishing a three-dimensional wind field reconstruction variational model; the objective function of the variational model includes making the reconstructed three-dimensional wind field as close as possible to the prior distribution in the horizontal direction, and passing through the constraint points of the measured atmospheric profile data in the vertical section, while the entire wind field satisfies fluid continuity and boundary layer physical laws; solving the three-dimensional wind field reconstruction variational model to obtain the wind speed vector at each three-dimensional grid point in space, thus forming a three-dimensional wind vector field.

[0037] In practice, a layered wind energy contribution assessment is performed on the primary wind energy characteristic field to distinguish between the topographically dominated wind energy component and the atmospheric turbulence-dominated wind energy component. This assessment process is implemented as follows: In the data structure of the primary wind energy characteristic field, a set of parameters representing the topographic influence and a set of parameters representing the equivalent wind speed are separated. The set of parameters representing the topographic influence includes topographic slope, aspect, and roughness length. The parameter representing the equivalent wind speed is the equivalent wind speed obtained from the inversion of the primary wind energy characteristic field. A pre-defined topographic influence weighting function is used, taking topographic slope, aspect, and roughness length as inputs, to calculate the contribution coefficient of the topography to the local wind speed. The contribution coefficient is a dimensionless number between 0 and 1. The equivalent wind speed in the primary wind energy characteristic field is multiplied by the calculated contribution coefficient to obtain the topographically modulated wind speed. The topographically modulated wind speed represents the portion of the wind speed explained solely by topographic influence, and the wind energy corresponding to the topographically modulated wind speed is defined as the topographically dominated wind energy component. From the total wind energy calculated based on equivalent wind speed from the primary wind energy characteristic field, the wind energy component dominated by topography is subtracted, and the remaining wind energy is defined as the wind energy component dominated by atmospheric turbulence. The spatial distribution characteristics of the wind energy components dominated by topography and those dominated by atmospheric turbulence are statistically analyzed separately, for example, calculating the spatial autocorrelation lengths of the two components in the east-west and north-south directions.

[0038] It is understandable that the specific form of the terrain influence weighting function can be defined based on fluid dynamics simulations or empirical relationships. An exemplary mathematical expression of the terrain influence weighting function is as follows: , in: Indicates the contribution coefficient. Indicates the slope of the terrain. Indicates the slope aspect of the terrain. Indicates the prevailing wind direction. Indicates the roughness length. Indicates reference altitude. and The parameters are for calibration. The parameterization of the terrain influence weighting function can be referenced in typical terrain cases. See Table 1, which shows a simplified correspondence between terrain parameters and contribution coefficients.

[0039] Table 1: Correspondence between Topographic Parameters and Contribution Coefficients

[0040] In some embodiments, prevailing wind direction It can be extracted from long-term climate statistics or the atmospheric motion trajectory sequence observed in this study. Optionally, the contribution coefficient can be further calculated by taking into account topographic curvature parameters.

[0041] In practice, the wind energy components dominated by topography and those dominated by atmospheric turbulence are fused, and external measured atmospheric profile data are introduced as constraints to perform a constrained three-dimensional wind field reconstruction to generate a three-dimensional wind vector field. The specific implementation of this reconstruction process is as follows: The spatially discrete wind energy components dominated by topography and those dominated by atmospheric turbulence are weighted and superimposed according to their respective energy values. The weights can be allocated based on data reliability or spatial representativeness. After weighted superposition, spatial interpolation yields a fused two-dimensional wind energy field covering the entire target area with a regular grid. Measured atmospheric profile data provided by meteorological towers or radiosondes within the target area are acquired. This measured atmospheric profile data includes wind speed and direction measurements at at least three different altitude levels. The fused two-dimensional wind energy field is used as the prior distribution of horizontal wind field intensity, and the measured atmospheric profile data is used as a hard constraint in the vertical direction. A variational model for reconstructing a three-dimensional wind field is established. The objective function of the variational model for reconstructing a three-dimensional wind field consists of three main parts. The first part is to make the wind speed distribution of the reconstructed three-dimensional wind field at each horizontal layer as close as possible to the prior distribution indicated by the fused two-dimensional wind energy field. The second part is to force the reconstructed three-dimensional wind field to strictly pass through the wind speed and direction constraint points provided by the measured atmospheric profile data at the location of the weather tower or radiosonde. The third part is to make the entire three-dimensional wind vector field satisfy the fluid continuity equation and conform to the vertical wind shear relationship defined by the physical laws of the atmospheric boundary layer.

[0042] Solving the variational model for reconstructing the 3D wind field, numerical solutions are obtained using the adjoint method or optimization algorithm. This yields the east-west, north-south, and vertical wind speed components at each 3D grid point, which together constitute the 3D wind vector field. In an example scenario, the target area is divided into 100m x 100m horizontal grids, with 10 vertical layers extending from the ground to a height of 300m. The final 3D wind vector field contains 3D wind speed information from tens of thousands of grid points.

[0043] See Figure 4In the micro-site selection of wind turbines during the application phase of wind energy resource mapping, this map presents a high-resolution wind power density distribution field of the target area in an XY plane (unit: km) (with polygonal areas of different hues representing spatial differences in wind energy resources), and marks the finally optimized wind turbine locations (red dots). In its implementation, this layout scheme is based on the full-area wind energy resource map. By identifying potential areas where wind power density exceeds the engineering development threshold, and combining the hub height and rotor diameter corresponding to the wind turbine model, it iteratively arranges the locations with the goal of maximizing total power generation. During the iteration process, the wake influence between locations based on the prevailing wind direction needs to be calculated and power generation losses deducted until the increment is lower than a preset threshold. The hue gradient in the map reflects the spatial heterogeneity of wind energy resources, while the distribution of wind turbine locations reflects the balance between utilizing high wind energy resource areas and avoiding wake interference.

[0044] Example 5: Based on the regional wind energy resource map, a micro-site selection simulation of wind turbines is performed to generate a wind turbine location layout scheme and a wake impact assessment report. The simulation process includes: identifying potential turbine locations on the regional wind energy resource map where the wind power density exceeds the engineering development threshold; determining the hub height and rotor diameter of the turbines within these potential locations based on preset single-unit capacity and turbine model; performing an automatic iterative simulation of turbine location layout within these potential locations with the goal of maximizing total power generation; calculating the wake impact between any two locations based on the prevailing wind direction in each iteration and deducting the power generation loss caused by the wake; stopping the iteration when the total power generation increase is less than a preset threshold, thus obtaining the final set of spatial coordinates of the wind turbine locations and forming a wind turbine location layout scheme; and simulating the wake superposition effect of all wind turbines under typical wind conditions based on the wind turbine location layout scheme and a three-dimensional wind vector field computational fluid dynamics wake model, outputting the increase in turbulence intensity and the annual power generation reduction coefficient for each wind turbine to form a wake impact assessment report.

[0045] In practice, a micro-site selection simulation of wind turbines is performed based on the regional wind energy resource map to generate turbine location plans and wake impact assessment reports. This simulation process is implemented as follows: On the regional wind energy resource map, a wind power density engineering development threshold is set, for example, 300 watts per square meter. Through spatial query and regional growth algorithms, all continuous geographical areas with wind power densities exceeding this threshold are identified and marked as potential turbine location areas. Within these potential location areas, the hub height and rotor diameter of the wind turbines are determined according to the pre-set single-unit capacity and selected turbine model. For example, a 3 MW wind turbine corresponds to a hub height of 100 meters and a rotor diameter of 120 meters. With the goal of maximizing total power generation, an iterative simulation of automatic wind turbine placement is conducted within a potential turbine location area. Initially, a set of initial wind turbine location spatial coordinates is generated randomly or according to a regular grid within the potential turbine location area. In each iteration, based on the wind direction frequency distribution and wind speed Weibull distribution parameters provided by the full-area wind energy resource map, the wake effect experienced by any two wind turbine locations downstream of the prevailing wind direction is calculated. An engineering wake model is used to calculate and deduct the power generation loss caused by the wake effect, thus obtaining the net total power generation under the current placement. In specific implementation, for a potential turbine location area of ​​10 km x 10 km, 50 candidate turbine locations are initially placed, and their positions are adjusted through iterative optimization. The calculation and optimization objective of total power generation can be expressed as: , in: This represents the total net power generation of the entire site. This indicates the total number of wind turbines. Indicates the free-flowing wind speed Next The theoretical power generation of a typhoon turbine is obtained by convolving the turbine's power curve with the wind speed frequency distribution. Indicates that it is located at the th The collection of all wind turbines upwind of the typhoon. Indicates due to the upwind direction The wake generated by the typhoon caused the first The power generation loss caused by typhoons, the loss value It is a function of the distance between the two aircraft, wind direction, turbulence intensity, and terrain parameters.

[0046] In practical implementation, when the total power generation in multiple consecutive iterations When the increment is less than a preset threshold, such as less than 0.1%, the iterative optimization process stops. At this point, the set of spatial coordinates of the wind turbine locations is determined as the final scheme, forming a wind turbine location layout scheme. The wind turbine location layout scheme records the latitude and longitude coordinates, hub height, and model of each wind turbine in the form of a file. Based on the wind turbine location layout scheme and the three-dimensional wind vector field generated in the previous steps, a computational fluid dynamics wake model is run. The computational fluid dynamics wake model uses the three-dimensional wind vector field as the inlet boundary condition, solves the Reynolds-averaged Navier-Stokes equations in the simulation domain, and uses an actuator disk or actuator line model to characterize the effect of the wind turbine rotor on the airflow, simulating the wake superposition effect of all wind turbines under multiple typical wind conditions.

[0047] In some embodiments, the turbulence model of the computational fluid dynamics wake model can adopt the k-epsilon model or its modified form. It is understood that the selection of typical wind conditions should cover the dominant wind direction sectors and typical wind speed sections in the entire region's wind energy resource map. After the simulation is completed, the increase in turbulence intensity at the hub height of each wind turbine and the annual power generation reduction factor considering wake loss are output. The increase in turbulence intensity is calculated from the turbulent kinetic energy of the flow field behind the wind turbine, and the annual power generation reduction factor is obtained by comparing the theoretical annual power generation of the wind turbine under free flow with the actual annual power generation calculated by the simulation. These results are compiled into a structured report to form a wake impact assessment report. The wake impact assessment report may include the average turbulence intensity at each turbine location, the annual equivalent full-load hours, and the percentage of power generation loss due to wake. Optionally, the automatic arrangement and iterative simulation of wind turbine locations can be performed using heuristic algorithms such as genetic algorithms and particle swarm optimization.

[0048] See Figure 5 In the wake impact assessment of wind turbine micro-situation, this figure presents the distribution of wake superposition effect coefficients for five wind turbines under five typical wind conditions (including wind direction and wind speed parameters). Specifically, the vertical axis of the figure represents typical wind conditions including wind direction (e.g., NW, NE) and wind speed (e.g., 8 m / s, 10 m / s), and the horizontal axis represents the turbine number. The cell values ​​and colors correspond to the wake superposition effect coefficients (the darker the color, the higher the coefficient). From the data characteristic analysis: under condition 4 (W, 12 m / s), the wake superposition effect coefficient of turbine 3 reaches 3.5, which is the maximum value in the figure, indicating that under this wind condition, this turbine is most affected by the wake superposition effect of the upwind turbine. Under condition 3 (SW, 6 m / s), the coefficients of each turbine are generally lower than 1.2, reflecting that the wake superposition effect is weaker under low wind speed and specific wind conditions. The coefficient distribution in this figure can directly support the quantitative analysis of "power generation loss caused by wake at each turbine location" in the wake impact assessment report. Its data comes from the simulation results of typical wind conditions by the computational fluid dynamics wake model and is one of the core reference bases for optimizing the wind turbine location layout scheme.

[0049] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for wind energy resource map inversion driven by acoustic-optical fusion, characterized in that, The method includes: Collect raw acoustic and optical remote sensing data within the target geographic area; The original acoustic data and original optical remote sensing data are processed to generate a primary wind energy feature field that includes terrain influence parameters. A stratified wind energy contribution assessment was performed on the primary wind energy characteristic field to distinguish between the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence. By integrating the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence, and introducing external measured atmospheric profile data as constraints, a constrained three-dimensional wind field reconstruction is performed to generate a three-dimensional wind vector field. Based on the three-dimensional wind vector field, the wind power density of each grid cell is calculated, and spatial diffusion correction is performed in combination with the surface roughness parameter to generate a high-resolution wind power density distribution field. The high-resolution wind power density distribution field is divided into multi-wind-direction sectors, and the wind frequency and average wind speed in each sector are statistically analyzed to draw a wind energy resource map of the entire region. Based on the wind energy resource map of the entire region, a micro-site selection simulation of wind turbine units is performed to generate a wind turbine location layout plan and a wake impact assessment report.

2. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 1, characterized in that, The process of processing the original acoustic data and the original optical remote sensing data to generate a primary wind energy feature field containing terrain influence parameters includes: Multi-scale acoustic energy spectrum decomposition processing is performed on the original acoustic data to decompose it into atmospheric turbulence acoustic energy spectrum components and background noise acoustic energy spectrum components. Temporal image difference enhancement processing is performed on the original optical remote sensing data to obtain the surface deformation feature sequence and the atmospheric motion trajectory sequence. The atmospheric turbulent acoustic energy spectrum component, the surface deformation characteristic sequence, and the atmospheric motion trajectory sequence are spatiotemporally synchronized and coupled with physical fields to construct an acoustic-optical synchronization feature body. Using the aforementioned acoustic-optical synchronization feature, adaptive noise suppression processing is performed on the background noise acoustic energy spectrum components to generate a pure acoustic energy perturbation field; The acoustic disturbance propagation path is extracted from the pure acoustic disturbance field and physically mapped to the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters.

3. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 2, characterized in that, The process of performing multi-scale acoustic spectrum decomposition on the original acoustic data, decomposing it into atmospheric turbulence acoustic spectrum components and background noise acoustic spectrum components, includes: Perform a Fourier transform on the original acoustic wave data to obtain its frequency domain representation; A turbulence characteristic frequency range that matches the atmospheric turbulence physical model is preset, and frequency components falling within the turbulence characteristic frequency range and their corresponding amplitudes and phases are extracted in the frequency domain representation; The frequency components and their corresponding amplitudes and phases are subjected to inverse Fourier transform to reconstruct the atmospheric turbulent acoustic energy spectrum components; The frequency components corresponding to the atmospheric turbulent acoustic energy spectrum components are removed from the frequency domain representation of the original acoustic data to obtain the remaining frequency domain components. Wavelet threshold denoising is performed on the remaining frequency domain components to filter out high-frequency random noise and obtain smooth remaining frequency domain components. Perform an inverse Fourier transform on the smoothed residual frequency domain components to reconstruct the background noise acoustic energy spectrum components; The step of performing time-series image differential enhancement processing on the original optical remote sensing data to obtain a land surface deformation feature sequence and an atmospheric motion trajectory sequence includes: Optical remote sensing images of the target geographic area at multiple consecutive times are acquired to form the original image time series; Radiometric calibration and atmospheric correction are performed on each optical remote sensing image in the original image time series to obtain a standard surface reflectance image; In the standard surface reflectance image sequence, the reflectance difference value at the same pixel position in two consecutive images is calculated; A threshold is set for the reflectivity difference value, and pixels whose absolute difference value exceeds the threshold are marked as potentially changed pixels; Morphological filtering and region growing are performed on the potential changed pixels to extract connected regions as surface deformation patches, and the surface deformation feature sequence is composed of the surface deformation patches at all times. Meanwhile, in the standard surface reflectance image sequence, images of specific bands are selected, and the motion vectors of atmospheric clouds or aerosols between adjacent images are calculated using the optical flow method. The motion vectors at multiple consecutive moments are connected to form the atmospheric motion trajectory sequence.

4. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 2, characterized in that, The process of spatiotemporally synchronizing and coupling the atmospheric turbulent acoustic energy spectrum components, the surface deformation characteristic sequence, and the atmospheric motion trajectory sequence to construct an acoustic-optical synchronization feature body includes: Establish a unified spatial geographic coordinate system and timestamp, and transform the acoustic detection point positions of the atmospheric turbulence acoustic energy spectrum components, the pixel positions of the surface deformation feature sequences, and the trajectory point positions of the atmospheric motion trajectory sequences to the unified spatial geographic coordinate system. Based on the timestamp, time-aligned interpolation is performed on the atmospheric turbulence acoustic energy spectrum component, the surface deformation feature sequence, and the atmospheric motion trajectory sequence to ensure that the three data have corresponding spatial descriptions at the same time point; At the data points where spatiotemporal synchronization is completed, the acoustic energy intensity of the atmospheric turbulent acoustic energy spectrum component, the deformation intensity of the surface deformation characteristic sequence, and the motion velocity vector of the atmospheric motion trajectory sequence are used as coupled physical quantities. Based on the fluid dynamics equations, a constitutive model of the coupled physical quantities is established. This constitutive model is used to describe the energy transfer and momentum exchange relationships between acoustic energy fluctuations, surface deformation, and atmospheric motion. Using the constitutive relation model, the spatiotemporally synchronized data is jointly solved to generate a multidimensional data volume containing acoustic energy field, deformation field and motion vector field. This multidimensional data volume is the acoustic-optical synchronization feature volume.

5. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 2, characterized in that, The step of using the aforementioned acoustic-optical synchronization feature to perform adaptive noise suppression processing on the background noise acoustic energy spectrum components to generate a pure acoustic energy perturbation field includes: Extract the local deformation intensity and local motion velocity that correspond in time and space to the background noise acoustic energy spectrum components from the acoustic-optical synchronization feature body; Analyze the statistical correlation between the local deformation intensity, local motion velocity, and the energy values ​​of the background noise acoustic energy spectrum components; Based on the statistical correlation, an adaptive filter is constructed, and the transfer function of the adaptive filter is dynamically adjusted by the instantaneous values ​​of the local deformation intensity and the local motion velocity. The background noise acoustic energy spectrum component is input into the adaptive filter, and the adaptive filter dynamically attenuates random noise energy that is unrelated to surface or atmospheric motion based on the local deformation intensity and local motion velocity at the current spatiotemporal point. Energy compensation is performed on the filtered signal to retain the acoustic energy components related to the actual atmospheric disturbances, and the output signal after noise suppression is the pure acoustic energy disturbance field.

6. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 2, characterized in that, The step of extracting the acoustic disturbance propagation path from the pure acoustic disturbance field and mapping it physically to the surface deformation feature sequence to generate a primary wind energy feature field containing topographic influence parameters includes: In the pure acoustic energy disturbance field, the propagation trajectory of the peak acoustic energy intensity in space and time is traced, and the propagation trajectory is defined as the acoustic energy disturbance propagation path; Calculate the propagation direction and speed at each point along the propagation path of the acoustic disturbance; The acoustic disturbance propagation path is superimposed onto the digital elevation model corresponding to the surface deformation feature sequence, and the terrain slope, aspect and roughness length at each point on the path are calculated. Based on the physical model of sound wave propagation in non-uniform atmosphere and complex terrain, and combined with the propagation direction, propagation speed, terrain slope, aspect and roughness length, the equivalent wind speed profile along the sound energy disturbance propagation path is inverted. The equivalent wind speed profile, along with the corresponding terrain slope, aspect, and roughness length, are encoded together into a data structure with spatial location information. This data structure is the primary wind energy feature field.

7. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 1, characterized in that, The step of performing a stratified wind energy contribution assessment on the primary wind energy characteristic field, distinguishing between the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence, includes: In the primary wind energy characteristic field, the set of parameters characterizing the influence of topography and the parameter characterizing the equivalent wind speed are separated. A preset terrain influence weighting function is used, which takes terrain slope, aspect and roughness length as inputs to calculate the contribution coefficient of terrain to local wind speed. Multiply the equivalent wind speed by the contribution coefficient to obtain the terrain-modulated wind speed, and define the wind energy corresponding to the terrain-modulated wind speed as the terrain-dominated wind energy component. From the total wind energy of the primary wind energy characteristic field, the wind energy component dominated by topography is subtracted, and the remaining wind energy is defined as the wind energy component dominated by atmospheric turbulence. The spatial distribution characteristics of the wind energy component dominated by topography and the wind energy component dominated by atmospheric turbulence were statistically analyzed separately.

8. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 1, characterized in that, The process involves fusing the topographically dominated wind energy component and the atmospheric turbulence-dominated wind energy component, and incorporating external measured atmospheric profile data as constraints to perform constrained three-dimensional wind field reconstruction, generating a three-dimensional wind vector field, including: The wind energy components dominated by topography and wind energy components dominated by atmospheric turbulence in space are weighted and superimposed according to their energy values ​​to obtain a fused two-dimensional wind energy field. Acquire measured atmospheric profile data provided by meteorological towers or radiosondes within the target area, wherein the measured atmospheric profile data includes wind speed and wind direction at different altitude levels; The fused two-dimensional wind energy field is used as the prior distribution of the horizontal wind field intensity, and the measured atmospheric profile data is used as the hard constraint condition in the vertical direction. A three-dimensional wind field reconstruction variational model is established. The objective function of the variational model includes: making the reconstructed three-dimensional wind field as close as possible to the prior distribution in the horizontal direction, and passing through the constraint points of the measured atmospheric profile data in the vertical section, while the entire wind field satisfies the fluid continuity and boundary layer physical laws. Solving the three-dimensional wind field reconstruction variational model yields the wind speed vector at each three-dimensional grid point in space, thus forming the three-dimensional wind vector field.

9. The method for wind energy resource map inversion driven by acoustic-optical fusion according to claim 1, characterized in that, Based on the aforementioned regional wind energy resource map, a micro-site selection simulation of wind turbines is performed to generate a wind turbine location layout plan and a wake impact assessment report, including: On the entire region's wind energy resource map, potential turbine locations with wind power density exceeding the engineering development threshold were identified; Within the potential machine location area, the hub height and rotor diameter of the fan are determined according to the preset single-unit capacity and fan model; With the goal of maximizing total power generation, an iterative simulation of automatic wind turbine location layout is performed within the potential turbine location area. In each iteration, the wake effect between any two locations based on the prevailing wind direction is calculated, and the power generation loss caused by the wake is deducted. When the increase in total power generation is less than a preset threshold, the iteration stops, and the final set of spatial coordinates of the wind turbine locations is obtained, forming the wind turbine location layout scheme. Based on the wind turbine location layout scheme and the three-dimensional wind vector field, a computational fluid dynamics wake model is run to simulate the wake superposition effect of all wind turbines under typical wind conditions. The increase in turbulence intensity and the annual power generation reduction coefficient of each wind turbine are output to form the wake impact assessment report.

10. A wind energy resource map inversion system driven by acoustic-optical fusion, characterized in that, The system includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the acoustic-optical fusion-driven wind energy resource map inversion method as described in any one of claims 1 to 9.

Images