Complex underlying surface atmospheric wind field optimization method and system

Abstract

The invention provides a complex underlying surface atmospheric wind field optimization method and system. The method comprises the steps of building a wind tunnel, obtaining flow field data, converting the flow field data into a meteorological station file, assimilating the meteorological station file to a CALMET diagnosis wind field, quantitatively evaluating a wind field interpolation point distribution scheme, and optimizing the CALMET diagnosis wind field. Actual meteorological station data are collected, wind field diagnosis is carried out on the target plant site through the CALMET diagnosis wind field, and a diagnosis result is obtained; establishing an underlying surface model of the target factory site, and extracting wind direction and wind speed data of a point distribution position; and fusing the wind direction and wind speed data of the point distribution position and the actual meteorological station data, inputting the data into a CALMET diagnosis wind field, and obtaining an output. According to the method, three-dimensional NS equation numerical solution is carried out on a local complex underlying surface, and high-resolution flow field data is obtained, so that the problem of how to predict the complex underlying surface more accurately in a nuclear accident emergency scene is solved.

Description

Technical Field

[0001] The present invention belongs to the field of wind field modeling technology, specifically, to a method and system for optimizing atmospheric wind fields over complex underlying surfaces. More specifically, it relates to a method for optimizing atmospheric wind fields over complex underlying surfaces driven by the coupling of forecast and measurement data. The complex underlying surfaces include urban and mountainous areas. Background Art

[0002] One of the most serious consequences of a nuclear accident is the massive leakage of radioactive materials. These radioactive materials usually enter the atmosphere in the form of gas or aerosols and spread rapidly throughout the atmosphere, causing widespread environmental pollution.

[0003] Therefore, it is necessary to accurately understand the propagation path and concentration distribution of radioactive materials. In flat terrain, the wind field is generally relatively stable and uniform, and the plume can be assumed to follow a Gaussian distribution in the horizontal and vertical directions. However, many nuclear facilities are currently located in complex geographical environments, such as mountains, river valleys, and urban buildings. In such terrain, wind flow is hindered, resulting in large velocity gradients and wind direction distortion. In this case, the Gaussian plume model is difficult to accurately predict the nuclide distribution, and refined calculations of the wind field in complex terrain are required.

[0004] Existing nuclear emergency response systems primarily rely on diagnostic wind field models to calculate wind fields over complex terrain. These models estimate the wind field using the principle of mass conservation and incorporate parameterization methods to describe wind field momentum and energy exchange over complex terrain. Representative models include CALMET and SWIFT. This approach, while not solving the complete set of fluid dynamics equations, can rapidly generate wind field data at a high grid resolution, ensuring the timeliness of emergency response. However, the simplified assumptions underlying the governing equations can reduce the model's computational accuracy in complex scenarios. This is particularly true in complex terrain, where simulation accuracy depends heavily on the choice of empirical parameters and the accuracy of the parameterization scheme. Furthermore, diagnostic wind field models utilize data assimilation to interpolate observed data onto continuous grid points using the inverse distance squared weighting (IDW) method, enabling objective analysis of the wind field. However, wind field observations in complex terrain are often scarce and uneven, resulting in poor wind field data assimilation. Therefore, while diagnostic wind field models can ensure the timeliness of emergency response, their computational accuracy for wind fields over complex terrain is limited.

[0005] To improve the accuracy of diagnostic wind field models, they can be coupled with computational fluid dynamics (CFD) models through wind field dynamic downscaling methods. Existing coupling methods use the output wind field data of the diagnostic wind field model as the boundary conditions of the CFD model, and then solve the fluid dynamics equations based on the CFD model to achieve high-precision calculations of the wind field. However, CFD models have difficulty in calculating wind fields under dynamic boundary conditions, such as changing wind direction and wind speed boundary conditions, and therefore their application in nuclear emergency scenarios is limited. At the same time, due to the complexity of the turbulence model and the computational requirements of high-resolution grids, the coupled system has low computational efficiency and is still unable to meet the needs of rapid emergency response.

[0006] Describes the existing implementation scheme that is most similar to the present invention. Document: Al-Khalidy N H. City scale pollutant dispersion modeling utilizing a combination of computational fluiddynamics and standard air quality simulation [J]. International journal of mechanics, 2017, 11: 210-217.

[0007] Specific solutions and existing problems: First, the basic ideas of the existing diagnostic wind field-CFD model coupled wind field calculation method are introduced, and the existing problems are explained:

[0008] The existing method for establishing meteorological fields is to generate the upper-air data required for the CALMET model based on the TAPM forecast model developed by the Commonwealth Scientific and Industrial Research Organization (CSIRO). The three-dimensional meteorological data generated by TAPM is used as the initial guess wind field for the CALMET model, and hourly wind and other meteorological field calculations are performed within the CALMET model's three-dimensional gridded modeling domain.

[0009] Problems: The study did not use data assimilation to drive the CALMET diagnostic wind field calculations, relying solely on TAPM model predictions. Running the model without any observational data assimilation makes the wind field calculations lack objectivity. When calculating wind fields on complex underlying surfaces, the lack of sufficient observational data can lead to greater deviations from actual wind field results.

[0010] The existing approach to pollutant dispersion calculations is to first perform a preliminary CALPUFF atmospheric dispersion simulation based on the meteorological field calculated using the CALMET diagnostic wind model. This preliminary CALPUFF dispersion model study identifies the boundary conditions that lead to the highest predicted pollutant concentrations (e.g., worst-case wind direction, wind speed, and ambient temperature). These boundary conditions are then used to drive the CFD pollutant dispersion calculations.

[0011] The problem is that driving the CALPUFF model diffusion calculation based on the lack of objectivity of CALMET model wind field data will inherit the uncertainty of the wind field results, resulting in significant deviations in the diffusion calculation results. Secondly, the wind field in the actual environment is highly unstable. Using only a single meteorological boundary as the inlet boundary condition of the CFD model will result in poor application of this method under real-world conditions.

[0012] Finally, CFD model calculations require fine grid division and wind field calculations combined with complex turbulence models, which will reduce overall computing efficiency and is not suitable for nuclear emergency scenarios that require rapid response.

[0013] Problems with traditional wind field diagnosis: The impact of complex terrain on the wind field is handled based on the principle of conservation of mass and parameterized terrain schemes. The complete set of fluid mechanics equations is not solved, so the resolution of the terrain is low, making it difficult to reflect the changes in wind field velocity gradients and wind direction distortion caused by the terrain. In addition, the calculation accuracy depends on the accuracy of the observation data and parameterized schemes.

[0014] Problems with traditional wind field coupling methods include: Low computational efficiency: Limited by the coupling method of wind field dynamic downscaling, the diagnostic wind field calculation results must be used as the boundary of the CFD model before CFD calculations can be performed. Furthermore, to achieve refined wind field calculations, the CFD model requires fine meshing and solving complex turbulence models, resulting in low computational efficiency of the coupled model and difficulty in ensuring the timeliness of nuclear emergency response.

[0015] In other words, the traditional method uses interpolation based on the mass conservation equation to solve the wind field based on sparse multi-point meteorological data from meteorological monitoring stations. This method is difficult to accurately predict the wind field when encountering complex underlying surfaces such as buildings, mountains, etc.

[0016] Patent document CN111881630A discloses a system and method for establishing a numerical wind tunnel for simulating the near-field environment of a nuclear facility. This solution includes: a pre-processing module for setting boundary conditions and experimental conditions through a GUI interface, importing and processing the geometric model of the numerical wind tunnel simulation object, setting mesh parameters, and automatically generating the mesh; a CFD solver module for solving steady-state models, isothermal or non-isothermal models, multi-component transport models, and Lagrangian particle models, accounting for the effects of buoyancy and providing multiple turbulence models; a numerical wind tunnel simulation module for selecting the appropriate turbulence model based on the pre-processing module and the CFD solver module, performing numerical wind tunnel simulations of the near-field atmospheric diffusion of the nuclear facility, and obtaining simulation results; and a post-processing module for automatically extracting various data from the simulation results and visualizing them. This solution only provides a CFD numerical wind tunnel and cannot improve the accuracy of wind field and radionuclide atmospheric diffusion predictions, nor does it provide a theoretical basis or technical support for nuclear accident emergency response. Summary of the Invention

[0017] In view of the defects in the prior art, the purpose of the present invention is to provide a method and system for optimizing the atmospheric wind field on a complex underlying surface.

[0018] A method for optimizing the atmospheric wind field on a complex underlying surface provided by the present invention includes:

[0019] Step S1: Build a wind tunnel, optimize and obtain CFD flow field parameters;

[0020] Step S2: Establishing a model of a complex underlying surface, calculating flow field data based on the CFD flow field parameters; generating a wind field interpolation point distribution scheme based on the flow field data; converting the flow field data into an image site file and assimilating it into a CALMET diagnostic wind field, quantitatively evaluating the wind field interpolation point distribution scheme to obtain an evaluation result; selecting a corresponding wind field interpolation point distribution scheme based on the evaluation result to optimize the CALMET diagnostic wind field;

[0021] Step S3: Collect actual weather station data, perform wind field diagnosis on the target plant site using CALMET wind field diagnostics, obtain diagnostic results, and then identify the target area based on the diagnostic results, thereby obtaining a variable wind direction wind speed dataset for the target area;

[0022] Step S4: Establish an underlying surface model of the target plant site, generate a target wind field interpolation layout plan based on the variable wind direction and speed data set of the target area, and then extract the wind direction and speed data at the layout locations through the target wind field interpolation layout plan; fuse the wind direction and speed data at the layout locations with the actual meteorological station data, input them into the CALMET diagnostic wind field, and obtain the output.

[0023] Preferably, in step S1, the wind tunnel is constructed according to a size scale of 1:1000 and a speed scale of 1:10, and a proportionally enlarged CFD simulation is performed on the wind tunnel to adjust the CFD flow field parameters; the deviation between the CFD simulation results and the wind tunnel test results corresponding to the CFD flow field parameters is less than 10%;

[0024] The atmospheric boundary layer wind speed contour line of the wind tunnel is determined according to the following mathematical expression:

[0025] v z =v 10 (z / 10) a

[0026] Among them, v z represents the average wind speed at height z, v 10 Indicates the average wind speed at a height of 10m; a is the wind profile index;

[0027] The flow field parameters include: parameters involved in the RANS method or parameters involved in the LES method;

[0028] The parameters involved in the RANS method include: turbulent viscosity, inlet turbulent dissipation rate, inlet turbulent kinetic energy and turbulent Schmidt number;

[0029] The parameters involved in the LES method include: subgrid viscosity, filter width, i.e., grid size and turbulent Schmidt number;

[0030] In step S2, the flow field data includes the flow field data used in the RANS method or the process data used in the LES method;

[0031] The mathematical expressions of the flow field data used in the RANS method are as follows from top to bottom:

[0032]

[0033] Among them, v t represents the turbulent viscosity, C μ is an empirical constant, usually taken as 0.09, k represents the inlet turbulent kinetic energy, ε represents the inlet turbulent dissipation rate, l represents the turbulent length scale, and U avg represents the average inlet velocity, I represents the turbulence intensity, measured by a laser Doppler velocimeter, and Sc t represents the turbulent Schmidt number, D t represents the turbulent diffusion coefficient; the symbol · represents the product;

[0034] The mathematical expressions of the flow field data used in the LES method are as follows from top to bottom:

[0035]

[0036] Among them, ν sgs represents the subgrid viscosity, C s is the Smagorinsky parameter, Δ represents the filter width, S ij represents the strain rate tensor after filtering, Δ represents the filter width, Δx, Δy and Δz represent the size of the grid in the X, Y and Z directions respectively, Sc t represents the turbulent Schmidt number, D sgs represents the sub-grid diffusion coefficient.

[0037] Preferably, the step S2 includes:

[0038] Step S2.1: Build a model of the complex underlying surface;

[0039] The model of the complex underlying surface includes an ideal three-dimensional mountain and an ideal two-dimensional ridge. The mathematical expressions of the ideal three-dimensional mountain and the ideal two-dimensional ridge are as follows from top to bottom:

[0040]

[0041] Where z represents the vertical position of the mountain, x represents the sidewind position of the mountain, and t represents the downwind position of the mountain. For an ideal three-dimensional mountain, H and L represent the mountain height and base radius, respectively. For an ideal two-dimensional ridge, H and L represent the cross-sectional height and half of the base width, respectively.

[0042] Step S2.2: Calculating flow field data based on the CFD flow field parameters;

[0043] Step S2.2: Generate a wind field interpolation point distribution plan based on flow field data;

[0044] Step S2.3: Convert the flow field data into an image site file and assimilate it into the CALEMT diagnostic wind field, quantitatively evaluate the wind field interpolation point distribution scheme, and obtain an evaluation result; the statistical indicators referenced by the evaluation include: FAC2, FB, NMSE and R;

[0045] Step S2.4: Based on the evaluation results, select the corresponding wind field interpolation point layout scheme to optimize the CALMET diagnostic wind field;

[0046] In the step S3, it includes:

[0047] Step S3.1: Decompose the CALMET diagnostic wind field into a velocity scalar field and a wind direction angle field. Fill the contour lines to obtain a wind speed gradient cloud map and a wind direction change rate cloud map. Generate a surface height gradient cloud map based on the underlying surface elevation data. Cross-compare the three types of cloud maps to determine whether the absolute value of the correlation coefficient between the wind speed gradient and the underlying surface height gradient is less than 0.3, or the absolute value of the correlation coefficient between the wind direction change rate and the underlying surface height gradient is less than 0.25, or the absolute value of the correlation coefficient between the wind speed gradient and the wind direction change rate is less than 0.2. If the result is yes, the corresponding area is identified as a low-correlation area with poor correlation and is used as the target area. If the result is no, no processing is performed.

[0048] Step S3.2: Using the CALMET diagnostic wind field as the boundary condition, CFD simulation is used to model the wind field with variable wind direction and wind speed in the local area as the target area, and a variable wind direction and wind speed dataset of the local area is obtained.

[0049] Preferably, the step S2.2 includes:

[0050] Step S2.2.1: Generate a cloud map based on the flow field data, and evenly draw multiple streamlines on the underlying surface starting from the wind speed inlet. Select different numbers of streamlines at even intervals, and then generate multiple streamline selection schemes;

[0051] Step S2.2.2: Using 1 / 3 of the mountain radius as the point spacing, all points are spaced on streamlines obtained through various streamline selection schemes, thereby obtaining various wind field interpolation point distribution schemes with different densities. The wind field interpolation point distribution schemes include uniform velocity gradient encryption at the foot of the windward side, the high wind speed area at the top of the mountain, and the vortex on the leeward side. The uniform velocity gradient encryption increases the number of data points in direct proportion to the rate of change of wind speed.

[0052] In step S2.3, the mathematical expression of FAC2 is:

[0053]

[0054] Among them, FAC2 is expressed as the ratio of CALMET to CFD calculated values, and the proportion of data within the range of two standard deviations;

[0055] The mathematical expression of the FB is:

[0056]

[0057] Wherein, FB represents the CFD and CALMET calculation results, that is, the deviation ratio between the calculated values, and the superscript “—” indicates that the calculated results are averaged;

[0058] The mathematical expression of the NMSE is:

[0059]

[0060] Among them, NMSE represents the standardized error index of the difference between CALMET and CFD calculation results, and the footer "i" represents the ordinal number of the small area after the wind field is evenly divided into 20×20 small areas, a total of 400 small areas, and CFD i and CALMET i They represent the average values ​​of CFD calculation results and CALMET calculation results in the i-th small area respectively;

[0061] The mathematical expression of R is:

[0062]

[0063] Here, R represents the Pearson correlation coefficient.

[0064] Preferably, in step S4, a data assimilation scheme of inverse square interpolation of distance is adopted to fuse the actual weather station data and the location data of the wind field data, assimilate them into the CALMET model, and perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculation;

[0065] The data assimilation scheme of the inverse square distance interpolation is mathematically expressed as follows:

[0066]

[0067] Among them, (u,v)′2 is the component of the second-step wind field at the target location in the x and y directions, (u,v)1 is the first-step wind field component of the target location obtained by considering the terrain dynamics effect, overland flow and terrain blocking effect parameters, (u obs ,v obs ) k is the x- and y-direction components of the observed wind at the kth meteorological station; (u CFD ,v CFD ) i are the wind components in the x and y directions of the i-th interpolation point calculated and transformed by CFD, R, R k With R i are the user-defined first-step wind field weight, the distance from the kth meteorological station to the target location, and the distance from the i-th interpolation point to the target location.

[0068] According to the present invention, a complex underlying surface atmospheric wind field optimization system is provided, comprising:

[0069] Module M1: Build a wind tunnel, optimize and obtain CFD flow field parameters;

[0070] Module M2: Establish a model of the complex underlying surface and calculate flow field data based on the CFD flow field parameters; generate a wind field interpolation point distribution scheme based on the flow field data; convert the flow field data into an image site file and assimilate it into the CAKMET diagnostic wind field, quantitatively evaluate the wind field interpolation point distribution scheme and obtain an evaluation result; select a corresponding wind field interpolation point distribution scheme based on the evaluation result and optimize the CAKMET diagnostic wind field;

[0071] Module M3: Collects actual meteorological station data, performs wind field diagnosis on the target site using CAKMET wind field diagnostics, obtains diagnostic results, and then identifies the target area based on the diagnostic results, thereby obtaining a variable wind direction wind speed dataset for the target area;

[0072] Module M4: Establish the underlying surface model of the target plant site, generate the target wind field interpolation layout plan based on the variable wind direction and speed data set of the target area, and then extract the wind direction and speed data at the layout location through the target wind field interpolation layout plan; fuse the wind direction and speed data at the layout location with the actual meteorological station data, input them into the CALMET diagnostic wind field, and obtain the output.

[0073] Preferably, in the module M1, the wind tunnel is constructed according to a size scale of 1:1000 and a speed scale of 1:10, and a proportionally enlarged CFD simulation is performed on the wind tunnel to adjust the CFD flow field parameters; the deviation between the CFD simulation results and the wind tunnel test results corresponding to the CFD flow field parameters is less than 10%;

[0074] The atmospheric boundary layer wind speed contour line of the wind tunnel is determined according to the following mathematical expression:

[0075] v z =v 10 (z / 10) a

[0076] Among them, v z represents the average wind speed at height z, v 10 Indicates the average wind speed at a height of 10m; a is the wind profile index;

[0077] The flow field parameters include: parameters involved in the RANS method or parameters involved in the LES method;

[0078] The parameters involved in the RANS method include: turbulent viscosity, inlet turbulent dissipation rate, inlet turbulent kinetic energy and turbulent Schmidt number;

[0079] The parameters involved in the LES method include: subgrid viscosity, filter width, i.e., grid size and turbulent Schmidt number;

[0080] In the module M2, the flow field data includes the flow field data used in the RANS method or the process data used in the LES method;

[0081] The mathematical expressions of the flow field data used in the RANS method are as follows from top to bottom:

[0082]

[0083] Among them, v t represents the turbulent viscosity, C μ is an empirical constant, usually taken as 0.09, k represents the inlet turbulent kinetic energy, ε represents the inlet turbulent dissipation rate, l represents the turbulent length scale, and U avg represents the average inlet velocity, I represents the turbulence intensity, measured by a laser Doppler velocimeter, and Sc t represents the turbulent Schmidt number, D t represents the turbulent diffusion coefficient; the symbol · represents the product;

[0084] The mathematical expressions of the flow field data used in the LES method are as follows from top to bottom:

[0085]

[0086] Among them, v sgs represents the subgrid viscosity, C s is the Smagorinsky parameter, Δ represents the filter width, S ij represents the strain rate tensor after filtering, Δ represents the filter width, Δx, Δy and Δz represent the size of the grid in the X, Y and Z directions respectively, Sc t represents the turbulent Schmidt number, D sgs represents the sub-grid diffusion coefficient.

[0087] Preferably, the module M2 includes:

[0088] Module M2.1: Modeling complex underlying surfaces;

[0089] The model of the complex underlying surface includes an ideal three-dimensional mountain and an ideal two-dimensional ridge. The mathematical expressions of the ideal three-dimensional mountain and the ideal two-dimensional ridge are as follows from top to bottom:

[0090]

[0091] Where z represents the vertical position of the mountain, x represents the sidewind position of the mountain, and y represents the downwind position of the mountain. For an ideal three-dimensional mountain, H and L represent the mountain height and base radius, respectively. For an ideal two-dimensional ridge, H and L represent the cross-sectional height and half of the base width, respectively.

[0092] Module M2.2: Calculate and obtain flow field data based on the CFD flow field parameters;

[0093] Module M2.2: Generate wind field interpolation point distribution plan based on flow field data;

[0094] Module M2.3: Convert the flow field data into image site files and assimilate them into CALMET wind field diagnostics, quantitatively evaluate the wind field interpolation point distribution scheme, and obtain evaluation results; the statistical indicators used in the evaluation include: FAC2, FB, NMSE, and R;

[0095] Module M2.4: Based on the evaluation results, select the appropriate wind field interpolation point layout scheme to optimize the CALMET diagnostic wind field;

[0096] The module M3 includes:

[0097] Module M3.1: Decompose the CALMET diagnostic wind field into a velocity scalar field and a wind direction angle field. Obtain wind speed gradient cloud maps and wind direction change rate cloud maps through contour filling. Generate underlying surface height gradient cloud maps based on underlying surface elevation data. Cross-compare the three types of cloud maps to determine whether the absolute value of the correlation coefficient between the wind speed gradient and the underlying surface height gradient is less than 0.3, or the absolute value of the correlation coefficient between the wind direction change rate and the underlying surface height gradient is less than 0.25, or the absolute value of the correlation coefficient between the wind speed gradient and the wind direction change rate is less than 0.2. If the result is yes, the corresponding area is identified as a low-correlation area with poor correlation and is used as the target area. If the result is no, no processing is performed.

[0098] Module M3.2: Using the CALMET diagnostic wind field as the boundary condition, CFD simulation is used to model the wind field with variable wind direction and wind speed in the local area as the target area, and a data set of variable wind direction and wind speed in the local area is obtained.

[0099] Preferably, the module M2.2 includes:

[0100] Module M2.2.1: Generate a cloud map based on the flow field data, and evenly draw multiple streamlines on the underlying surface starting from the wind speed inlet. Select different numbers of streamlines at even intervals to generate multiple streamline selection schemes.

[0101] Module M2.2.2: Using 1 / 3 of the mountain radius as the point spacing, all points are spaced on streamlines derived from various streamline selection schemes, thereby obtaining wind field interpolation point distribution schemes of varying densities. This wind field interpolation point distribution scheme involves uniformly encrypting velocity gradients at the windward foot of the mountain, the high wind speed area at the top of the mountain, and the vortex on the leeward side. The uniform velocity gradient encryption results in an increase in data points that is directly proportional to the rate of change of wind speed.

[0102] In the module M2.3, the mathematical expression of FAC2 is:

[0103]

[0104] Among them, FAC2 is expressed as the ratio of CALMET to CFD calculated values, and the proportion of data within the range of two standard deviations;

[0105] The mathematical expression of the FB is:

[0106]

[0107] Wherein, FB represents the CFD and CALMET calculation results, that is, the deviation ratio between the calculated values, and the superscript “—” indicates that the calculated results are averaged;

[0108] The mathematical expression of the NMSE is:

[0109]

[0110] Among them, NMSE represents the standardized error index of the difference between CALMET and CFD calculation results, and the footer "i" represents the ordinal number of the small area after the wind field is evenly divided into 20×20 small areas, a total of 400 small areas, and CFD i and CALMET i They represent the average values ​​of CFD calculation results and CALMET calculation results in the i-th small area respectively;

[0111] The mathematical expression of R is:

[0112]

[0113] Here, R represents the Pearson correlation coefficient.

[0114] Preferably, in the module M4, a data assimilation scheme of inverse square interpolation of distance is adopted to fuse the actual weather station data and the location data of the wind field data, assimilate them into the CALMET model, and perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculation;

[0115] The data assimilation scheme of the inverse square distance interpolation is mathematically expressed as follows:

[0116]

[0117] Among them, (u,v)′2 is the component of the second-step wind field at the target location in the x and y directions, (u,v)1 is the first-step wind field component of the target location obtained by considering the terrain dynamics effect, overland flow and terrain blocking effect parameters, (u obs ,v obs ) kis the x- and y-direction components of the observed wind at the kth meteorological station; (u CFD ,v CFD ) i are the wind components in the x and y directions of the i-th interpolation point calculated and transformed by CFD, R, R k With R i are the user-defined first-step wind field weight, the distance from the kth meteorological station to the target location, and the distance from the i-th interpolation point to the target location.

[0118] Compared with the prior art, the present invention has the following beneficial effects:

[0119] 1. The present invention numerically solves the three-dimensional NS equations for local complex underlying surfaces, combines the obtained high-resolution flow field data with multi-point meteorological data from meteorological stations, and performs data assimilation to obtain a higher-resolution wind field. This aims to solve the problem of how to more accurately predict complex underlying surfaces in nuclear accident emergency scenarios.

[0120] 2. The CALMET simulation accuracy of the present invention is low in complex terrain and cannot effectively reflect the wind field's flow, acceleration, and turbulence, resulting in large deviations in plume diffusion. Based on this, the present patented method studies wind field optimization schemes on typical complex underlying surfaces. For three-dimensional mountains, the key wind field disturbance locations are at the foot, top, and leeward side. After optimizing various point placement schemes, significant results were achieved. The wind speed FAC2 between the diagnostic wind field and the CFD simulation results at the same location increased from 0.8102 to 0.9608, the NMSE decreased from 0.2299 to 0.0286, the correlation coefficient increased from -0.7119 to 0.9395, and the absolute value of FB decreased from 0.1358 to 0.0118. Under the premise of minimizing the number of points, the optimal solution is: points on seven streamlines with a velocity gradient of 0.5 m / s. In addition, for two-dimensional ridges, the angle between the ridge and the inlet direction has a greater impact on turbulence. Different optimized point placement schemes were used for parallel ridges and inclined ridges, and both achieved good optimization results. For the underlying surface of the mountain and ridge combination, the wind field changes are similar to those of a single mountain. The optimized CALPUFF model is consistent with the CFD simulation results, verifying the effectiveness of the optimization scheme.

[0121] 3. Aiming at the complex terrain of the actual plant site, the present invention extracts key complex terrain and establishes a CFD flow field database under 16 wind direction boundaries. The corresponding CFD flow field data is extracted under hypothetical accident conditions, and the wind field is optimized. Specifically, the research results show that the optimized wind field can better reflect the influence of complex terrain, especially in areas with complex terrain, where the deformation and trajectory deflection of smoke diffusion are more obvious. Under different terrain conditions, the smoke diffusion results before and after optimization are significantly different, and the demand for wind field optimization in areas with complex terrain is more prominent.

[0122] 4. By integrating CFD high-resolution flow field data and meteorological station monitoring data, the present invention significantly improves the prediction accuracy of wind fields and atmospheric diffusion of radioactive nuclides without reducing the timeliness of calculations, providing a theoretical basis and technical support for nuclear accident emergency response. BRIEF DESCRIPTION OF THE DRAWINGS

[0123] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings:

[0124] Figure 1 A schematic diagram showing a comparison of improvements to the method provided by the present invention;

[0125] Figure 2 Schematic diagram of the typical complex underlying surface wind field assimilation optimization method provided by the present invention;

[0126] Figure 3 Schematic diagram of the hourly resolution wind field data assimilation optimization method provided by the present invention in the nuclear accident emergency response situation;

[0127] Figure 4 Schematic diagram comparing the wind tunnel experiment provided by the present invention with the CFD-RANS and CFD-LES models;

[0128] Figure 5 Schematic diagram of ideal three-dimensional mountain flow field distribution provided by the present invention with H = 400m and L = 3000m; H and L represent the mountain height and bottom radius respectively;

[0129] Figure 6 The ideal three-dimensional mountain wind vector distribution and wind speed and direction difference heat map provided by the present invention with H = 400m and L = 3000m, where H and L represent the mountain height and bottom radius respectively;

[0130] Figure 7 Schematic diagram of an ideal three-dimensional mountain point distribution scheme provided by the present invention with H = 400m and L = 3000m, where H and L represent the mountain height and bottom radius respectively;

[0131] Figure 8 Schematic diagram of the diagnostic wind field vector distribution after optimization of eight schemes for an ideal three-dimensional mountain with H = 400m and L = 3000m provided by the present invention, where H and L represent the mountain height and bottom radius respectively;

[0132] Figure 9 The CALMET and CFD wind speed and direction difference heat map after optimization of the ideal three-dimensional mountain schemes 1-4 with H = 400m and L = 3000m provided by the present invention, where H and L represent the mountain height and bottom radius respectively;

[0133] Figure 10 The CALMET and CFD wind speed and direction difference heat map after optimization of the ideal three-dimensional mountain scheme 5-8 with H = 400m and L = 3000m provided by the present invention, where H and L represent the mountain height and bottom radius respectively;

[0134] Figure 11 A schematic diagram of the key complex terrain in the northwest direction of the ITER site provided by the present invention;

[0135] Figure 12 Schematic diagram of the boundary setting of the local area CFD model provided by the present invention;

[0136] Figure 13 Schematic diagram of wind direction at the southern boundaries of regions 1 and 2 based on the preliminary simulation of the CALMET diagnostic wind field on January 15, 2022, provided by the present invention;

[0137] Figure 14 Schematic diagram of CFD flow field distribution in region 1 below the typical inlet wind direction boundary provided by the present invention;

[0138] Figure 15 A schematic diagram showing the wind vector comparison before and after optimization of wind farms in regions 1 and 2 from 15:00 to 18:00 on January 15, 2022, provided by the present invention;

[0139] Figure 16 This is a schematic diagram comparing the diffusion of the smoke puff provided by the present invention when it moves toward the northwest wind direction and when it passes through areas 1 and 2 before and after optimization. DETAILED DESCRIPTION

[0140] The present invention will be described in detail below with reference to specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those skilled in the art, several changes and improvements can be made without departing from the scope of the present invention. These all fall within the scope of protection of the present invention.

[0141] Related Glossary: ​​Gaussian plume model: A theoretical model used to describe the diffusion of pollutants in the atmosphere. It assumes that the horizontal and vertical diffusion of pollutants follows a normal, or Gaussian, distribution. It is applicable to flat terrain and simple atmospheric flow conditions.

[0142] Diagnostic wind field models: Based on known meteorological observation data, combined with the physical laws of atmospheric motion, such as conservation of mass and momentum, and empirical relationships, a specific algorithm is used to calculate and analyze the magnitude, direction, and spatial distribution of wind in the atmosphere at the current moment. Meteorological observation data includes air pressure and temperature.

[0143] The Navier-Stokes equations are an expression of Newton's second law for incompressible viscous flow. These equations describe the conservation of momentum for viscous, incompressible fluids. Together with the conservation of mass and energy equations, they form a complete set of fluid dynamics equations used to solve a variety of complex fluid flow problems.

[0144] Computational Fluid Dynamics (CFD) is a mathematical method used to solve fluid flow problems. Based on the fundamental conservation laws of fluid mechanics—namely, the laws of conservation of mass, momentum, and energy—it describes the motion of fluids by establishing a system of partial differential equations. Numerical methods are then used to discretize the continuous fluid region into a finite number of grid cells. The partial differential equations are then converted into algebraic equations and solved, yielding numerical solutions for physical quantities such as velocity, pressure, and temperature at each point in the flow field.

[0145] Wind field dynamic downscaling: The process of using the output information of the large-scale meteorological field as the boundary conditions of the small-scale model and refining the wind field prediction in the small-scale model is called wind field dynamic downscaling. This method can improve the spatial resolution of the large-scale model and drive the wind field calculation of the small-scale model in the absence of observational data, overcoming the dependence on observational data. The large-scale meteorological field includes global or regional climate models; the small-scale model includes local or block models.

[0146] Data assimilation: Data assimilation is a technique that combines observational data with numerical models. The goal is to obtain a more accurate estimate and description of the system state by integrating information from both. Based on the principles of Bayesian statistics, data assimilation comprehensively considers both observational data errors and model errors to provide an optimal estimate of the system state. This estimate is consistent with both the observed data and the physical laws and dynamic characteristics described by the numerical model.

[0147] Turbulence Model: A turbulence model is a set of methods and equations used to simplify and model the pulsating characteristics of turbulent flow through certain assumptions and mathematical methods. This method transforms the turbulence problem into a solvable mathematical problem. The goal is to reduce solution complexity and computational cost while maintaining a certain level of accuracy.

[0148] The innovation of this invention is to use CFD to recalculate the local area, that is, the area caused by the inaccurate wind field diagnostic method, and then put the CFD calculation results into the CALMET model, combined with the original meteorological data points, to recalculate a more accurate wind field.

[0149] A method for optimizing the atmospheric wind field on a complex underlying surface provided by the present invention includes:

[0150] Step S1: Optimize CFD calculation parameters based on wind tunnel experiments;

[0151] The CFD flow field parameters are CFD calculation parameters;

[0152] According to the mission objectives, the appropriate turbulence simulation method is determined, including: Reynolds-averaged Navier-Stokes method, namely RANS method and large eddy simulation method, namely LES method.

[0153] If the RANS method is used, the parameters that need to be determined include: turbulent viscosity, inlet turbulent dissipation rate, inlet turbulent kinetic energy, and turbulent Schmidt number;

[0154] If the LES method is used, the parameters that need to be determined include: subgrid viscosity, filter width, that is, grid size, and turbulent Schmidt number.

[0155] Step S2: Establish an ideal model of a typical complex underlying surface, calculate the flow field on the typical underlying surface using the CFD method to obtain flow field data, and generate multiple wind field interpolation point distribution schemes based on the flow field data; convert the flow field data into a meteorological station file and assimilate it into the CALMET diagnostic wind field, and then quantitatively evaluate the multiple wind field interpolation point distribution schemes to obtain evaluation results; select a corresponding scheme based on the evaluation results to optimize the wind field;

[0156] Step S3: Based on the actual weather station data, CALMET is used to diagnose the wind field at the target site under all wind directions and speeds. Based on the simulation results, local areas with large velocity gradients and wind direction distortion are identified. Using the CALMET-diagnosed wind field as the boundary condition, CFD methods are used to further model the wind field in the local area with varying wind directions and speeds, generating a localized variable wind direction and speed dataset.

[0157] Step S4: Based on the wind field conditions at the actual emission time of the target plant site, the wind direction and speed data of the distribution point locations are extracted from the local area variable wind direction and speed dataset in step S3 according to the distribution plan in step S2, and formatted into a meteorological station time series file that can be recognized by CALMET. Finally, the actual meteorological station data and distribution point location data are integrated to perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculations.

[0158] In step S1, CFD calculation parameters are optimized according to wind tunnel experiments;

[0159] Specifically, a wind tunnel experimental environment was constructed using a 0.5% CO2 mixture as a tracer, with a scale factor of 1:1000 and a velocity factor of 1:10. Wind tunnel diffusion experiments were conducted for different wind inlet profiles, chimney outlet velocities, and underlying surface types, collecting CO concentrations at a height of 1 cm from 50 to 200 cm downwind. CFD simulations were then performed on the wind tunnel area, scaling it up according to the scale factor and velocity factor. The CFD turbulence model, turbulent Schmidt number, wind inlet turbulent kinetic energy, and turbulent dissipation rate were adjusted to ensure that the CFD simulation results deviated from the wind tunnel experimental results by less than 10%.

[0160] The optimized CFD calculation parameters obtained by the wind tunnel experiment in step S1 are logically related to step S2 as follows: the optimized CFD calculation parameters obtained by the wind tunnel experiment are input parameters for calculating the flow field on a typical underlying surface using the CFD method;

[0161] In step S2, multiple wind field interpolation point distribution schemes are generated through flow field data;

[0162] Several wind field interpolation point placement schemes involve drawing streamlines within the CFD-calculated flow field and placing all points along these streamlines. The density of the points is controlled by varying the number of streamlines, the spacing between points on each streamline, and the degree of density. The goal is to maximize the flow field optimization while minimizing the amount of point placement work.

[0163] First, streamlines were evenly arranged on the underlying surface, using four schemes: 17, 12, 7, and 3. Since it was necessary to at least reflect the main characteristics of the flow field, which is flowing around the two sides of the mountain and accelerating at the top of the mountain, at least 3 streamlines were arranged.

[0164] Because the flow field decelerates on the windward side and accelerates with increasing mountain height, forming vortices on the leeward side, the velocity gradient is large in these areas. Therefore, more data points are needed to accurately reflect the flow characteristics. Therefore, we considered adding data points at the foot of the windward side, the high wind speed area at the top of the mountain, and the vortex on the leeward side. At these three key locations, a data point is added for each certain change in velocity, using two different velocity changes: 0.5 m / s and 1 m / s.

[0165] Specifically, the process of generating a wind field interpolation point distribution plan using flow field data includes:

[0166] To accurately reflect the flow field characteristics, post-processing software was used to plot the flow field data obtained from CFD calculations into a cloud map. A total of 20 streamlines were evenly drawn on the underlying surface starting from the wind speed inlet. Different numbers of streamlines were selected at even intervals, following the selection principles of 85%, 60%, 35%, and 15%. Because the main characteristics of the flow field, which flows around the sides of the mountain and accelerates at the top, must be reflected, a minimum of three streamlines were arranged, resulting in four options: 17, 12, 7, and 3 streamlines. All points were then placed on the streamlines, with a spacing of 1 / 3 of the mountain radius.

[0167] Because the flow field slows down on the windward side and accelerates as the mountain height increases, forming vortices on the leeward side, the velocity gradient is large in these areas, requiring more data points to accurately reflect the flow characteristics. Therefore, we considered intensifying the data at the foot of the windward side, the high wind speed area at the top of the mountain, and the vortex on the leeward side. At these three key locations, a data point is added for each change in velocity by a certain value, ultimately resulting in a variety of wind field interpolation point schemes with different densities.

[0168] The flow field data includes RANS method and LES method;

[0169] The mathematical expressions of the flow field data used in the RANS method are as follows from top to bottom:

[0170]

[0171] Among them, v t represents the turbulent viscosity, C μ is an empirical constant, usually taken as 0.09, k represents the inlet turbulent kinetic energy, ε represents the inlet turbulent dissipation rate, l represents the turbulent length scale, and U avg represents the average inlet velocity, I represents the turbulence intensity, measured by a laser Doppler velocimeter, and Sc t represents the turbulent Schmidt number, D t represents the turbulent diffusion coefficient;

[0172] The mathematical expressions of the flow field data used in the LES method are as follows from top to bottom:

[0173]

[0174] Among them, v sgs represents the subgrid viscosity, C s is the Smagorinsky parameter, Δ represents the filter width, S ij represents the strain rate tensor after filtering, Δ represents the filter width, Δx, Δy, Δz represent the size of the grid in three directions, Sc t represents the turbulent Schmidt number, D sgs represents the subgrid diffusion coefficient;

[0175] In the step S2, a plurality of wind field interpolation point distribution schemes are quantitatively evaluated to obtain an evaluation result; a corresponding scheme is selected based on the evaluation result; the quantitative evaluation value and the corresponding wind field interpolation point distribution scheme are specifically associated as follows: FAC2, FB, NMSE, and R are four commonly used statistical indicators, especially in meteorology, fluid mechanics, and model evaluation, which are used to measure the deviation and consistency between simulated data and actual observation data. Therefore, these indicators will be used to quantitatively analyze the deviation and correlation of the wind field before and after CFD and CALMET optimization. Since the time consumption of the CALMET method based on the wind field diagnosis method is related to the number of interpolation points used, even if adding more interpolation points can theoretically make the wind field diagnosis more accurate, it is still necessary to obtain the optimal solution between the calculation time and calculation accuracy; therefore, the FAC2, FB, NMSE, R of different wind field interpolation point distribution schemes and the wind direction and wind speed interpolation heat maps before and after optimization are comprehensively analyzed to obtain the optimal distribution scheme;

[0176] In step S4, the key complex terrain area boundary includes: a turbulent area and a bypass area;

[0177] In step S4, the meteorological station time series file consists of hourly changing wind direction and wind speed data at the boundary of the key complex terrain area, and hourly changing CFD flow field data of the wind direction boundary;

[0178] The meteorological station time series file consists of: the hourly wind direction and wind speed of the optimized wind field interpolation point evaluated in step S2 and its change with height. Figure 7 For example, in each sub-graph of this figure, the black points distributed around the mountain are interpolation points, and the wind direction and speed at these locations change over time. Because this is a bird's-eye view, the change in height at each point is not shown. In fact, the wind direction and speed at each point at different heights also change over time, which is the time series file of the meteorological station. The wind direction and speed data for the "critical complex terrain area boundary" you mentioned is only used to provide the inlet boundary conditions for the CFD calculation.

[0179] Based on the wind field conditions at the target plant site at the time of actual emissions, the wind direction and speed data for the target site locations are extracted from the local area variable wind direction and speed dataset in step S3 according to the site placement plan in step S2. This data is then formatted into a weather station time series file recognizable by CALMET. Finally, the actual weather station data and site location data are integrated to perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculations. The calculation method for this process is built into CALMET, namely the wind field diagnosis method.

[0180] In other words, in step S4, a data assimilation scheme of inverse square interpolation of distance is adopted to fuse the actual weather station data and the distribution point location data to perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculation;

[0181] The method for fusing actual weather station data and location data is specifically as follows: combining the wind field data generated by the CFD model with the actual weather station data, assimilating them into the CALMET model, correcting and filling the deficiencies in the observation data, and making the wind field assimilation results closer to the actual state of the system.

[0182] The specific implementation method is: use the data assimilation scheme of inverse square distance interpolation, and its mathematical expression is:

[0183]

[0184] Among them, (u,v)′2 is the component of the second-step wind field at the target location in the x and y directions, (u,v)1 is the first-step wind field component of the target location obtained by considering the terrain dynamics effect, overland flow and terrain blocking effect parameters, (u obs ,v obs ) k is the x- and y-direction components of the observed wind at the kth meteorological station; (u CFD ,v CFD ) i are the wind components in the x and y directions at the i-th interpolation point calculated and transformed by CFD, R, R k 、R i are the user-defined first-step wind field weight, the distance from the kth meteorological station to the target location, and the distance from the ith interpolation point to the target location; the core idea of ​​this method is inverse distance weighting, that is, weighted correction of wind speed according to the distance of the observation point.

[0185] Specifically, the observation point or interpolation point that is closer to the grid point, that is, R k or R i Smaller values ​​have a greater impact on wind field corrections and are therefore assigned a higher weight. The weighting parameter R adjusts the relative weight between the wind field and the observational data in the first step. By adjusting R, the model's reliance on observational data and calculated results can be flexibly controlled. Applying this method to hourly-resolution wind field diagnosis and atmospheric diffusion simulations completes the entire interpolation optimization process.

[0186] 1) Steps

[0187] Step 1: CFD model reliability verification;

[0188] The basic idea is to verify the reliability of the turbulence model used by the CFD method in predicting the diffusion behavior of pollutants in the atmosphere by comparing and analyzing the deviations between CFD results and wind tunnel experiments using different turbulence models under a flat underlying surface.

[0189] Steps: Determine the ratio of the geometric scale and velocity scale between the actual environment and the wind tunnel experiment based on the similarity criterion. The wind speed contour line of the atmospheric boundary layer satisfies:

[0190] v z =v 10 (z / 10) a

[0191] Among them, v z represents the average wind speed at height z, v 10 represents the average wind speed at a height of 10 m; a is the wind profile index, which is fitted to the wind profile equation by adjusting the fan speed of the wind tunnel apparatus. The ratio of the downwind concentration C to the source term Q is defined as the atmospheric dispersion factor C / Q, and used as an indicator to compare and analyze the model and experimental prediction results.

[0192] Step 2: Optimization of wind field assimilation data on typical complex underlying surfaces;

[0193] The basic concept is to model a typical complex underlying surface, then perform CFD flow field calculations within the study area. Based on the reliability of the CFD model verified by wind tunnel testing, CFD flow field data from key areas of flow field variation is extracted and used as input meteorological data for the CALMET diagnostic wind field model, thereby improving the diagnostic wind field model's accuracy for complex underlying surface wind fields.

[0194] Steps: First, according to the cosine square cross-section formula shown in equation (1), an ideal model of a typical complex underlying surface is established. Two typical complex underlying surfaces, mountain and ridge, are considered: for an ideal three-dimensional mountain, H and L represent the mountain height and bottom radius, respectively; for an ideal two-dimensional ridge, H and L represent the cross-section height and 1 / 2 of the bottom width, respectively.

[0195]

[0196] CFD methods are used to calculate the flow field over a typical underlying surface. Through comprehensive simulation of the wind field, topography, and wind speed, detailed flow field distribution characteristics are obtained. Particular attention is paid to key locations that significantly influence the wind field, such as turbulent areas and flow-around regions. Based on the wind field data extracted from the CFD simulation results, various wind field interpolation point placement schemes are designed.

[0197] The extracted CFD flow field data is formatted into a weather station file recognizable by the CALMET model. After data conversion, the CFD flow field data is assimilated into the CALMET diagnostic wind field using the distance inverse method. The optimization effect can be quantified by comparing and analyzing the differences in wind field data between the CALMET and CFD models before and after interpolation.

[0198] The deviation and correlation of the wind field before and after CFD and CALMET optimization were quantitatively analyzed based on four indicators: FAC2, FB, NMSE, and R.

[0199] FAC2 is defined as the proportion of data for which the ratio of the CALMET-calculated value to the CFD-calculated value is within two standard deviations. The closer the FAC2 value is to 1, the closer the CALMET-calculated value to the CFD-calculated value.

[0200]

[0201] FB is defined as the deviation ratio between the CFD and CALMET calculation results. If FB = 0, the two calculations are exactly the same. If FB > 0, the CALMET calculation is too large, and vice versa.

[0202]

[0203] NMSE is defined as the normalized error metric of the difference between CALMET and CFD. It measures the error magnitude using the mean square error, or MSE, while also eliminating the effect of data size by normalizing the variance of CALMET data. - Indicates that the calculation results are averaged;

[0204]

[0205] R is the Pearson Correlation Coefficient, which is used for the linear correlation between CALMET and CFD, and its value range is -1 to 1. The closer R is to 1, the stronger the positive linear relationship between the two.

[0206]

[0207] Based on the above statistical indicators, the optimization effect of each scheme is quantitatively evaluated to determine the optimal interpolation and assimilation point distribution scheme suitable for typical complex underlying surfaces. The specific flow chart is as follows: Figure 2 As shown; Figure 2 The wind field assimilation optimization method for typical complex underlying surfaces is demonstrated.

[0208] Statistical indicators are used to quantitatively analyze the deviation and correlation of the wind fields before and after CFD and CALMET optimization, taking into account calculation time and accuracy, and then the optimal solution is obtained. Among the above statistical indicators, the closer the FAC2 value is to 100%, the closer the CALMET and CFD calculated values ​​are, and the better the optimization effect is; the closer the FB and NMSE values ​​are to 0, the smaller the deviation between the CALMET and CFD calculated values ​​is, and the better the optimization effect is; the closer the R value is to 1, the higher the positive correlation between the CALMET and CFD calculated values ​​is, and the better the optimization effect is.

[0209] Step 3: Optimization and analysis of wind field assimilation data on complex underlying surfaces with variable wind direction;

[0210] Steps: First, for a specific nuclear power plant site, the CALPUFF model was used to diagnose the wind field and simulate pollutant dispersion within the study area. Based on these simulation results, the complex local terrain affecting the wind field was identified. Elevation data for this terrain region was then extracted, and CFD methods were used to refine the wind field model. Analysis of the CALPUFF model's wind field diagnostic results revealed that wind speed and direction were more stable in areas with relatively flat terrain.

[0211] Therefore, the flat terrain side of the study area served as the input boundary for wind speed and direction. Based on the x- and y-axis components of the wind speed in this area, CFD simulations were performed under 16 different wind direction conditions. Wind field data at key points were extracted using the data extraction scheme developed for wind field optimization on typical complex underlying surfaces. This data will serve as a complex terrain wind field assimilation database to support subsequent wind field optimization and diagnosis.

[0212] When an accident occurs, the wind field distribution in the area is first calculated using the diagnostic wind field model, and the hourly changing wind direction and wind speed data of the key complex terrain area boundaries are obtained. Next, the CFD flow field data of the hourly changing wind direction boundary is extracted and formatted into a meteorological station time series file that can be recognized by CALMET. Finally, based on these data, hourly resolution diagnostic wind field and atmospheric diffusion optimization calculations are performed. The process of the optimization method is as follows: Figure 3 shown.

[0213] In other words, step S3.1: identifying the target area based on the diagnostic results is specifically achieved as follows: decomposing the CALMET diagnostic wind field into a velocity scalar field and a wind direction angle field, and using the contour filling method to generate a wind speed gradient cloud map and a wind direction change rate cloud map to highlight the local variation characteristics of the wind field;

[0214] Based on the underlying surface elevation data, the underlying surface height gradient cloud map is generated by terrain slope calculation. Based on the principles of wind field dynamics, the wind speed / direction gradient should have a significant correlation with the terrain gradient. By cross-comparing the three maps, low-correlation areas with poor correlation are identified. These areas require interpolation optimization because the CALMET diagnostic results differ greatly from the actual results due to insufficient initial wind field data or terrain parameterization errors. These are the "target areas" mentioned in this article.

[0215] 5.1 CFD model reliability verification

[0216] Wind tunnel experiments validated the reliability of the turbulence model used in CFD methods to predict tritium's atmospheric diffusion behavior. The deviations between CFD results and wind tunnel experiments were compared and analyzed for different turbulence models applied to a flat underlying surface. The model scale was determined based on empirical operating conditions and the diffusion range, and the geometric and velocity scale ratios were defined as 1:1000 and 1:10, respectively, based on the Mauve criterion.

[0217] Assuming that only density, a gas's inherent properties, determine its diffusion behavior in a wind tunnel, we used carbon monoxide (CO) as a tracer, with a density similar to that of gaseous tritiated water (HTO). The actual experiments used a mixture of 0.5% carbon monoxide (CO) and 99.5% nitrogen (N2) by volume. Results from CFD-RANS models with turbulent Schmidt numbers of 0.2 and 0.7, as well as results from a CFD-LES model with a turbulent Schmidt number of 0.7, were compared with the wind tunnel experiments.

[0218] The results show that the CFD-RANS model based on a turbulent Schmidt number of 0.7 over-predicts the downwind concentration. Therefore, the CFD-RANS simulation results were optimized by reducing the turbulent Schmidt number to 0.2. At this time, the downwind concentrations at different altitudes were more consistent with the wind tunnel test results. The CFD-LES model with a turbulent Schmidt number of 0.7 was in good agreement with the wind tunnel test at 90m in the actual environment. However, at 50m in the actual environment, the prediction results were not as good as the CFD-RANS model with a turbulent Schmidt number of 0.2. At 70m in the actual environment, there was a certain deviation between the prediction results of both models and the wind tunnel test. Therefore, considering computational efficiency, the flow field data of the CFD-RANS model with a turbulent Schmidt number of 0.2 was used for the subsequent wind field data assimilation optimization of the CALMET model.

[0219] 5.2 Optimization of wind field assimilation data on typical complex underlying surfaces;

[0220] In the 5km×5km calculation domain, an ideal three-dimensional symmetrical mountain underlying surface is established. The mountain parameters are L=3000m, H=400m, and the mountain contour conforms to the cosine square cross-section formula shown in formula (1). The flow field is calculated using the CFD model, and the calculation results are as follows: Figure 5 shown.

[0221] CFD simulation results show that the mountain's influence on the flow field is significant. As air flows over the mountain, it first encounters obstruction on the windward side, significantly reducing wind speed. Simultaneously, some air flows around the sides of the mountain. As the air rises up the mountain, the wind speed accelerates, particularly as the mountain's height increases. On the leeward side of the mountain, the air converges, forming a distinct vortex structure.

[0222] The same underlying surface and computational domain are established in the CALMET model, and the resulting flow field vector distribution is as follows: Figure 6 As shown in the lower left. Compared with the CFD results, the CALMET flow field is evenly distributed, which makes it difficult to reflect the change in velocity direction when the flow field passes through the mountain. The wind direction and speed values ​​calculated by CFD and CALMET are subtracted to obtain the difference heat map before optimization, as shown in Figure 6 Right. Significant differences in wind speed are observed in key areas of flow field variation, such as at the foot and summit of the mountain on either side. Significant differences in wind direction are also observed in the vortex formed at the foot of the leeward mountain. Wind direction differences also exist where the flow circulates around the mountain on either side. Flow field distribution is crucial for diffusion. CALMET's low accuracy in calculating flow fields for typical complex underlying surfaces leads to significant uncertainty in concentration diffusion results. Therefore, it is necessary to optimize CALMET's wind field diagnostics.

[0223] Determining the placement of points in the optimization scheme is crucial for optimizing the wind field. To accurately reflect the flow field characteristics, streamlines are drawn from the CFD-calculated flow field, and all points are placed along these streamlines. The density of the points is controlled by varying the number of streamlines, the spacing between points on each streamline, and the degree of density. The goal is to maximize the flow field optimization while minimizing the amount of data placement. First, streamlines are evenly distributed across the underlying surface. Four schemes, 17, 12, 7, and 3, were adopted. Since the primary characteristic of the flow field, which flows around the mountain and accelerates at the summit, must be captured, a minimum of three streamlines are required. On each streamline passing through the mountain, points are placed at a distance of 1 / 3 of the mountain's radius, which in this case is 1000 meters. Because the flow field decelerates on the windward side and accelerates with increasing mountain height, forming vortices on the leeward side, the velocity gradient is large in these areas. Therefore, more data points are needed to accurately reflect the flow characteristics. Therefore, density is considered at the foot of the windward side, at the high wind speed area at the summit, and at the vortices on the leeward side. At these three key locations, a data point is added every time the speed changes by a certain value, and two encryption schemes with different speed changes of 0.5m / s and 1m / s are adopted. Based on the above ideas, 8 point distribution schemes with different densities are proposed for this typical underlying surface. Figure 7 and Table 1.

[0224] Table 1: Layout parameters of ideal three-dimensional mountain schemes 1 to 8 with H = 400m and L = 3000m

[0225]

[0226]

[0227] Based on the above 8 layout schemes, the CFD wind direction and wind speed calculated values ​​of the corresponding locations were extracted and compiled into CALMET, that is, a meteorological station file that can be recognized by the CALMET model, to optimize the wind field. After CALMET optimization, the wind vector distribution under the 8 schemes is as follows Figure 4-15 As shown. The optimization scheme of CFD data assimilation has significantly improved the CALMET calculation results, especially in the key areas where the flow field changes, the wind direction and wind speed have changed significantly. For example, at the top of the mountain, the wind speed increased significantly, and a vortex flow phenomenon appeared at the foot of the mountain on the leeward side of the mountain. The optimization effects under different point distribution schemes vary greatly. Schemes 1-4 have a large number of points and the points at the top of the mountain are denser, so they are more outstanding in optimizing the acceleration effect at the top of the mountain. However, as the number of points decreases, the high-speed wind vector at the top of the mountain decreases. It should be noted that arranging too many interpolation points at the top of the mountain may affect the wind field in the low-speed area. Therefore, the optimization results need to be further quantitatively analyzed and adjusted to ensure the accuracy and rationality of the overall wind field.

[0228] Furthermore, by calculating the difference between the optimized CALMET results and the CFD model results, the wind direction and wind speed difference thermal coefficients under the eight optimization schemes were obtained. Figure 9 and Figure 10 Compared to pre-optimization, the wind direction and wind speed differences for all optimization schemes were significantly reduced in key areas, particularly in areas with significant wind field variations. The wind direction differences remained largely unchanged across the eight optimization schemes, but as the number of deployment points decreased, the wind speed differences for schemes 5-8 were significantly greater than those for schemes 1-4. Maximizing the optimization effect while minimizing the number of deployment points is key to selecting the optimal deployment solution. Therefore, further quantitative analysis of the optimization results is necessary.

[0229] Based on four statistical indicators, a quantitative analysis of wind speed and direction before and after optimization was conducted, with the results shown in Tables 2 and 3. Before optimization, the CALMET wind speed and the calculated CFD wind speed showed a negative linear correlation. After optimization, the two showed a strong positive linear relationship, with the correlation coefficient optimized from -0.7119 to a maximum of 0.9395.

[0230] Under the eight optimization schemes, FAC2 and NMSE were optimized to varying degrees, with FAC2 improving from 0.8102 to a maximum of 0.9608, and NMSE improving from 0.2299 to a minimum of 0.0286. Meanwhile, the absolute value of FB decreased in optimization schemes 1 to 5, from 0.1358 to a minimum of 0.0118, while optimization schemes 6 to 8 resulted in increased deviations in some areas. This may be because, when the number of points is small, the data in the critical flow area has a greater impact on other areas, and the lack of sufficient points in other areas to optimize the flow field at that location increases the deviation of the results.

[0231] Table 2 Wind speed statistical index values ​​before optimization and after optimization of eight schemes for an ideal three-dimensional mountain with H = 400m and L = 3000m

[0232]

[0233]

[0234] Good optimization results were also achieved for wind direction. Before optimization, CALMET and CFD wind direction showed a weak negative correlation, with poor correlation. After optimization, the correlation reached a maximum of 0.5249, and FAC2 increased from 0.7755 to a maximum of 0.8634. The absolute value of FB decreased from 0.2397 to a minimum of 0.0128, while NMSE decreased from 0.1846 to a minimum of 0.1189. In schemes 6-8, NMSE increased slightly compared to before optimization. Figure 7It can be seen that the smaller number of points on the leeward side results in a poor response to the dramatic wind direction changes caused by turbulence in that area. Overall, Scheme 5 achieves better optimization results with a smaller number of points, and all four statistical indicators show significant improvements.

[0235] Therefore, Scheme 5 is selected to represent the diagnostic wind field optimization scheme of the ideal three-dimensional mountain underlying surface and is used for the subsequent combined mountain optimization layout scheme.

[0236] Table 3 Wind direction statistical index values ​​before optimization and after optimization of eight schemes for an ideal three-dimensional mountain with H = 400m and L = 3000m

[0237]

[0238] 5.3 Optimization and analysis of wind field assimilation data on complex underlying surfaces with variable wind direction

[0239] Taking the International Thermonuclear Experimental Reactor (ITER) as an example, the wind field is optimized and analyzed based on the optimization scheme obtained in Section 5.2 for the case where puffs move in different directions. The trajectory and distribution characteristics of puffs driven by the wind field before and after optimization are compared and analyzed to verify the feasibility of applying the wind field data assimilation optimization method based on CFD flow field data on actual terrain.

[0240] In view of the smoke cloud spreading to the northwest, two key complex terrains in the northwest of ITER were extracted, such as Figure 11 As shown in Figure 1, Terrain 1 is generally ridge-like and can be abstracted as an ideal ridge underlying surface. The corresponding optimized point distribution scheme is then applied to extract CFD flow field data. The southern portion of Terrain 1 is flat, and the wind field there can be used as the wind direction boundary for CFD modeling. Region 2 can be approximated as an ideal three-dimensional symmetrical mountain. The CALMET diagnostic wind field in the southern flat region is used as the input boundary for the CFD model.

[0241] A CFD model is established based on the terrain elevation data of regions 1 and 2, and a refined wind field modeling is performed on the local complex terrain. Taking the 150km regional coordinates centered on ITER as the benchmark, the x-direction range of region 1 is [-34000, -14000], the y-direction range is [7000, 17000], the model x, y plane size is 20km×10km, and the height is 3km. The x, y direction ranges of region 2 are [-35000, -17000] and [24000, 38000] respectively, the x, y plane size is 25km×10km, and the height is 3km. In the CFD calculation, the velocity inlet and pressure outlet are set, the underlying surface is set as the wall, and the remaining surfaces are set as outflow boundaries, such as Figure 12Among them, the CFD velocity inlet boundary realizes different wind direction inputs by setting the velocity magnitude and the x and y components of the velocity. Based on the above model and boundary condition settings, CFD flow field calculations are performed in areas 1 and 2 under 16 wind direction boundary conditions. According to the wind field data assimilation optimization method of typical complex underlying surfaces, wind field data at key locations are extracted and formatted into CALMET meteorological station files as the CFD model flow field database.

[0242] The 10 hours after 14:00 on January 15, 2022 are selected as the time series under the hypothetical accident scenario. Taking the CALMET diagnosis of the wind field at 14:00 on January 15 as an example, Figure 13 As shown, the wind direction in the south of key areas 1 and 2 in the northwest direction is consistent at this moment, at 282°, which is closest to 292° among the 16 wind direction intervals. Therefore, CFD flow field data with an inlet wind direction of 292° at this moment are extracted for wind field data assimilation optimization. Similarly, during the 10 hours from 2:00 PM to 11:00 PM on January 15th, the wind directions for areas 1 and 2, as initially simulated by CALMET, were read and flow field data corresponding to the CFD wind direction were extracted hourly. The corresponding wind directions are shown in Table 4, achieving flow field data assimilation optimization under the variable meteorological conditions during this period.

[0243] Within a selected 10-hour period, the CALMET wind direction boundaries were mapped to the 16 wind direction zones of the CFD model. Typical wind directions of 292°, 180°, 157.5°, 135°, and 112.5° appeared in the 10-hour time series. The wind directions in zones 1 and 2 were similar, but due to the higher altitude of zone 2, the wind direction deflection was more pronounced. Therefore, there was some difference in wind direction between the two zones within the same timeframe.

[0244] Table 4 CALMET and CFD inlet wind directions during the 10-hour period when the smoke puff spread to the northwest

[0245]

[0246] During the 10 hours that the smoke puff moved northwestward, it mainly experienced flow field distribution under four different wind direction conditions, as follows: Figure 14 As shown in Figure 1, the flow field distribution in region 1 is similar to that of an ideal ridge underlying surface. On the windward side of the ridge, the wind speed initially decreases, then gradually increases with increasing ridge height, reaching a maximum at the ridge crest. On the leeward side of the ridge, an irregular turbulent structure forms. Therefore, it is reasonable to apply the optimization scheme for the ideal underlying surface to actual terrain.

[0247] Based on the CFD flow field data of the northwest region 1 and region 2, this paper optimizes the CALMET wind field data by assimilation, and compares the wind vector changes within four hours from 15:00 to 18:00 on January 15, 2022, before and after optimization. Figure 15 . At 15:00 and 16:00 on January 15, the wind direction and wind speed in the 150km study area before optimization were relatively stable, and no significant terrain influence was shown. However, after optimization of Area 1 and Area 2, the wind field showed obvious deflection in these two places. At 17:00 and 18:00 on January 15, the wind field before optimization showed large fluctuations due to the influence of terrain, while after optimization, the wind direction deflection angles in Area 1 and Area 2 increased further. At the same time, other areas were not affected by the optimization, and the wind field still reflected the terrain characteristics well.

[0248] Starting at 3:00 PM on January 15, 2022, the smoke puff moved northwestward within 10 hours. Due to the increased wind speed in the optimized wind field, the smoke puff passed through Area 1 and Area 2 at different times before and after optimization. Therefore, the moments when the smoke puff passed through key complex terrain were selected for comparative analysis.

[0249] Figure 16 The figure shows the average concentration of smoke puffs over a 10-hour period before and after optimization, as well as the diffusion of smoke puffs as they passed through Areas 1 and 2. Before optimization, the smoke puffs had a uniform, regular elliptical distribution and were unaffected by the terrain as they passed through Areas 1 and 2, with minimal morphological changes.

[0250] The average concentration of smoke plumes within 10 hours was mainly concentrated around the chimney, with a limited diffusion range and a peak concentration of 115,000 Bq / m 3 After optimization, the smoke puff underwent significant deformation when passing through areas 1 and 2 due to increased wind speed and a significant shift in wind direction. Compared with before optimization, the diffusion of the smoke puff was enhanced, and the average peak concentration within 10 hours dropped to 29,000 Bq / m 3 , concentrations were not only concentrated around the chimney but also had a significant impact on a larger area to the northwest. In summary, the imprecise response of the wind field to complex terrain before optimization led to significant uncertainty in the results; however, after optimization, smoke puff diffusion was enhanced and responded more effectively to complex terrain, highlighting the importance of complex terrain in nuclear emergency response and the need to develop emergency strategies on a larger scale.

[0251] The present invention also provides a complex underlying surface atmospheric wind field optimization system, which can be implemented by executing the process steps of the complex underlying surface atmospheric wind field optimization method. That is, those skilled in the art can understand the complex underlying surface atmospheric wind field optimization method as a preferred implementation of the complex underlying surface atmospheric wind field optimization system.

[0252] Those skilled in the art will appreciate that, in addition to implementing the system and its various devices, modules, and units provided by the present invention in purely computer-readable program code, it is entirely possible to implement the same functions of the system and its various devices, modules, and units provided by the present invention in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; the devices, modules, and units for implementing various functions can also be considered as both software modules implementing the method and structures within the hardware component.

[0253] In the description of this application, it should be understood that the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application.

[0254] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. The embodiments of this application and the features in the embodiments may be combined with each other in any manner unless there is a conflict.

Claims

1. A method for optimizing the atmospheric wind field on a complex underlying surface, characterized in that: include: Step S1: Build a wind tunnel, optimize and obtain CFD flow field parameters; Step S2: Establish a model of the complex underlying surface, calculate and obtain flow field data based on the CFD flow field parameters; and generate a wind field interpolation point distribution plan based on the flow field data; Converting the flow field data into an image site file and assimilating it into a CALMET diagnostic wind field, quantitatively evaluating the wind field interpolation point distribution scheme to obtain an evaluation result; selecting a corresponding wind field interpolation point distribution scheme based on the evaluation result to optimize the CALMET diagnostic wind field; Step S3: Collect actual weather station data, perform wind field diagnosis on the target plant site using CALMET wind field diagnostics, obtain diagnostic results, and then identify the target area based on the diagnostic results, thereby obtaining a variable wind direction wind speed dataset for the target area; Step S4: Establish an underlying surface model of the target plant site, generate a target wind field interpolation layout plan based on the variable wind direction and speed data set of the target area, and then extract the wind direction and speed data at the layout location based on the target wind field interpolation layout plan; The wind direction and speed data at the distribution point locations are integrated with the actual weather station data, input into the CALMET wind field diagnosis, and output is obtained.

2. The method for optimizing the atmospheric wind field on a complex underlying surface according to claim 1, characterized in that: In step S1, the wind tunnel is constructed according to a size scale of 1:1000 and a speed scale of 1:10, and a proportionally enlarged CFD simulation is performed on the wind tunnel to adjust CFD flow field parameters; the deviation between the CFD simulation results and the wind tunnel test results corresponding to the CFD flow field parameters is less than 10%; The atmospheric boundary layer wind speed contour line of the wind tunnel is determined according to the following mathematical expression: in z =in 10 (with / 10) a Among them, v z represents the average wind speed at height z, v 10 Indicates the average wind speed at a height of 10m; a is the wind profile index; The flow field parameters include: parameters involved in the RANS method or parameters involved in the LES method; The parameters involved in the RANS method include: turbulent viscosity, inlet turbulent dissipation rate, inlet turbulent kinetic energy and turbulent Schmidt number; The parameters involved in the LES method include: subgrid viscosity, filter width, i.e., grid size and turbulent Schmidt number; In step S2, the flow field data includes the flow field data used in the RANS method or the process data used in the LES method; The mathematical expressions of the flow field data used in the RANS method are as follows from top to bottom: Among them, c t represents the turbulent viscosity, C μ is an empirical constant, usually taken as 0.09, k represents the inlet turbulent kinetic energy, ε represents the inlet turbulent dissipation rate, l represents the turbulent length scale, and U avg represents the average inlet velocity, I represents the turbulence intensity, measured by a laser Doppler velocimeter, and Sc t represents the turbulent Schmidt number, D t represents the turbulent diffusion coefficient; the symbol · represents the product; The mathematical expressions of the flow field data used in the LES method are as follows from top to bottom: Among them, v sgs represents the subgrid viscosity, C s is the Smagorinsky parameter, Δ represents the filter width, S ij represents the strain rate tensor after filtering, Δ represents the filter width, Δx, Δy and Δz represent the size of the grid in the X, Y and Z directions respectively, Sc t represents the turbulent Schmidt number, D sgs represents the sub-grid diffusion coefficient.

3. The method for optimizing the atmospheric wind field on a complex underlying surface according to claim 2, characterized in that: In the step S2, it includes: Step S2.1: Build a model of the complex underlying surface; The model of the complex underlying surface includes an ideal three-dimensional mountain and an ideal two-dimensional ridge. The mathematical expressions of the ideal three-dimensional mountain and the ideal two-dimensional ridge are as follows from top to bottom: Where z represents the vertical position of the mountain, x represents the sidewind position of the mountain, and y represents the downwind position of the mountain. For an ideal three-dimensional mountain, H and L represent the mountain height and base radius, respectively. For an ideal two-dimensional ridge, H and L represent the cross-sectional height and half of the base width, respectively. Step S2.2: Calculating flow field data based on the CFD flow field parameters; Step S2.2: Generate a wind field interpolation point distribution plan based on flow field data; Step S2.3: Convert the flow field data into an image site file and assimilate it into the CALMET diagnostic wind field, quantitatively evaluate the wind field interpolation point distribution scheme, and obtain an evaluation result; the statistical indicators used in the evaluation include: FAC2, FB, NMSE and R; Step S2.4: Based on the evaluation results, select the corresponding wind field interpolation point layout scheme to optimize the CALMET diagnostic wind field; In the step S3, it includes: Step S3.1: Decompose the CALMET diagnostic wind field into a velocity scalar field and a wind direction angle field. Fill the contour lines to obtain a wind speed gradient cloud map and a wind direction change rate cloud map. Generate a surface height gradient cloud map based on the underlying surface elevation data. Cross-compare the three types of cloud maps to determine whether the absolute value of the correlation coefficient between the wind speed gradient and the underlying surface height gradient is less than 0.3, or the absolute value of the correlation coefficient between the wind direction change rate and the underlying surface height gradient is less than 0.25, or the absolute value of the correlation coefficient between the wind speed gradient and the wind direction change rate is less than 0.

2. If the result is yes, the corresponding area is identified as a low-correlation area with poor correlation and is used as the target area. If the result is no, no processing is performed. Step S3.2: Using the CALMET diagnostic wind field as the boundary condition, CFD simulation is used to model the wind field with variable wind direction and wind speed in the local area as the target area, and a variable wind direction and wind speed dataset of the local area is obtained.

4. The method for optimizing the atmospheric wind field on a complex underlying surface according to claim 3, characterized in that: In the step S2.2, it includes: Step S2.2.1: Generate a cloud map based on the flow field data, and evenly draw multiple streamlines on the underlying surface starting from the wind speed inlet. Select different numbers of streamlines at even intervals, and then generate multiple streamline selection schemes; Step S2.2.2: Using 1 / 3 of the mountain radius as the point spacing, all points are spaced on streamlines obtained through various streamline selection schemes, thereby obtaining various wind field interpolation point distribution schemes with different densities. The wind field interpolation point distribution schemes include uniform velocity gradient encryption at the foot of the windward side, the high wind speed area at the top of the mountain, and the vortex on the leeward side. The uniform velocity gradient encryption increases the number of data points in direct proportion to the rate of change of wind speed. In step S2.3, the mathematical expression of FAC2 is: Among them, FAC2 is expressed as the ratio of CALMET to CFD calculated values, and the proportion of data within the range of two standard deviations; The mathematical expression of FB is: Wherein, FB represents the CFD and CALMET calculation results, that is, the deviation ratio between the calculated values, and the superscript "—" indicates that the calculated results are averaged; The mathematical expression of the NMSE is: Among them, NMSE represents the standardized error index of the difference between CALMET and CFD calculation results. The footer "i" represents the ordinal number of the small area after the wind field is evenly divided into 20×20 small areas, a total of 400 small areas. i and CALMET i They represent the average values ​​of CFD calculation results and CALMET calculation results in the i-th small area respectively; The mathematical expression of R is: Where R represents the Pearson correlation coefficient.

5. The method for optimizing the atmospheric wind field on a complex underlying surface according to claim 1, characterized in that: In step S4, a data assimilation scheme of inverse square interpolation of distance is used to fuse the actual weather station data and the location data of the wind field data, and assimilate them into the CALMET model to perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculation; The data assimilation scheme of the inverse square distance interpolation is mathematically expressed as follows: Among them, (u,v)′2 is the component of the second-step wind field at the target location in the x and y directions, (u,v)1 is the first-step wind field component of the target location obtained by considering the terrain dynamics effect, overland flow and terrain blocking effect parameters, (u obs ,v obs ) k is the x- and y-direction components of the observed wind at the kth meteorological station; (u CFD ,v CFD ) i are the wind components in the x and y directions of the i-th interpolation point calculated and transformed by CFD, R, R k With R i are the user-defined first-step wind field weight, the distance from the kth meteorological station to the target location, and the distance from the i-th interpolation point to the target location.

6. A complex underlying surface atmospheric wind field optimization system, characterized in that: include: Module M1: Build a wind tunnel, optimize and obtain CFD flow field parameters; Module M2: Establish a model of the complex underlying surface and calculate flow field data based on the CFD flow field parameters; generate a wind field interpolation point distribution scheme based on the flow field data; convert the flow field data into an image site file and assimilate it into the CALMET diagnostic wind field, quantitatively evaluate the wind field interpolation point distribution scheme and obtain an evaluation result; select a corresponding wind field interpolation point distribution scheme based on the evaluation result and optimize the CALMET diagnostic wind field; Module M3: Collects actual weather station data, performs wind field diagnosis on the target site using CALMET wind field diagnostics, obtains diagnostic results, and then identifies the target area based on the diagnostic results, thereby obtaining a variable wind direction wind speed dataset for the target area; Module M4: Establish the underlying surface model of the target plant site, generate the target wind field interpolation layout plan based on the variable wind direction and speed data set of the target area, and then extract the wind direction and speed data at the layout location through the target wind field interpolation layout plan; The wind direction and speed data at the distribution point locations are integrated with the actual weather station data, input into the CALMET wind field diagnosis, and output is obtained.

7. The complex underlying surface atmospheric wind field optimization system according to claim 6, characterized in that: In the module M1, the wind tunnel is constructed according to a scaled-down scale of 1:1000 and a speed scaled-down scale of 1:10, and a proportionally scaled CFD simulation is performed on the wind tunnel to adjust the CFD flow field parameters; the deviation between the CFD simulation results and the wind tunnel test results corresponding to the CFD flow field parameters is less than 10%; The atmospheric boundary layer wind speed contour line of the wind tunnel is determined according to the following mathematical expression: in z =in 10 (with / 10) a Among them, v z represents the average wind speed at height z, v 10 Indicates the average wind speed at a height of 10m; a is the wind profile index; The flow field parameters include: parameters involved in the RANS method or parameters involved in the LES method; The parameters involved in the RANS method include: turbulent viscosity, inlet turbulent dissipation rate, inlet turbulent kinetic energy and turbulent Schmidt number; The parameters involved in the LES method include: subgrid viscosity, filter width, i.e., grid size and turbulent Schmidt number; In the module M2, the flow field data includes the flow field data used in the RANS method or the process data used in the LES method; The mathematical expressions of the flow field data used in the RANS method are as follows from top to bottom: Among them, v t represents the turbulent viscosity, C μ is an empirical constant, usually taken as 0.09, k represents the inlet turbulent kinetic energy, ε represents the inlet turbulent dissipation rate, l represents the turbulent length scale, and U avg represents the average inlet velocity, I represents the turbulence intensity, measured by a laser Doppler velocimeter, and Sc t represents the turbulent Schmidt number, D t represents the turbulent diffusion coefficient; the symbol · represents the product; The mathematical expressions of the flow field data used in the LES method are as follows from top to bottom: Among them, v sgs represents the subgrid viscosity, C s is the Smagorinsky parameter, Δ represents the filter width, S ij represents the strain rate tensor after filtering, Δ represents the filter width, Δx, Δy and Δz represent the size of the grid in the X, Y and Z directions respectively, Sc t represents the turbulent Schmidt number, D sgs represents the sub-grid diffusion coefficient.

8. The complex underlying surface atmospheric wind field optimization system according to claim 7, characterized in that: The module M2 includes: Module M2.1: Modeling complex underlying surfaces; The model of the complex underlying surface includes an ideal three-dimensional mountain and an ideal two-dimensional ridge. The mathematical expressions of the ideal three-dimensional mountain and the ideal two-dimensional ridge are as follows from top to bottom: Where z represents the vertical position of the mountain, x represents the sidewind position of the mountain, and y represents the downwind position of the mountain. For an ideal three-dimensional mountain, H and L represent the mountain height and base radius, respectively. For an ideal two-dimensional ridge, H and L represent the cross-sectional height and half of the base width, respectively. Module M2.2: Calculate and obtain flow field data based on the CFD flow field parameters; Module M2.2: Generate wind field interpolation point distribution plan based on flow field data; Module M2.3: Convert the flow field data into image site files and assimilate them into CALMET wind field diagnostics, quantitatively evaluate the wind field interpolation point distribution scheme, and obtain evaluation results; the statistical indicators used in the evaluation include: FAC2, FB, NMSE, and R; Module M2.4: Based on the evaluation results, select the appropriate wind field interpolation point layout scheme to optimize the CALMET diagnostic wind field; The module M3 includes: Module M3.1: Decompose the CALMET diagnostic wind field into a velocity scalar field and a wind direction angle field. Obtain wind speed gradient cloud maps and wind direction change rate cloud maps through contour filling. Generate underlying surface height gradient cloud maps based on underlying surface elevation data. Cross-compare the three types of cloud maps to determine whether the absolute value of the correlation coefficient between the wind speed gradient and the underlying surface height gradient is less than 0.3, or the absolute value of the correlation coefficient between the wind direction change rate and the underlying surface height gradient is less than 0.25, or the absolute value of the correlation coefficient between the wind speed gradient and the wind direction change rate is less than 0.

2. If the result is yes, the corresponding area is identified as a low-correlation area with poor correlation and is used as the target area. If the result is no, no processing is performed. Module M3.2: Using the CALMET diagnostic wind field as the boundary condition, CFD simulation is used to model the wind field with variable wind direction and wind speed in the local area as the target area, and a data set of variable wind direction and wind speed in the local area is obtained.

9. The complex underlying surface atmospheric wind field optimization system according to claim 8, characterized in that: The module M2.2 includes: Module M2.2.1: Generate a cloud map based on the flow field data, and evenly draw multiple streamlines on the underlying surface starting from the wind speed inlet. Select different numbers of streamlines at even intervals to generate multiple streamline selection schemes. Module M2.2.2: Using 1 / 3 of the mountain radius as the point spacing, all points are spaced on streamlines derived from various streamline selection schemes, thereby obtaining wind field interpolation point distribution schemes of varying densities. This wind field interpolation point distribution scheme involves uniformly encrypting velocity gradients at the windward foot of the mountain, the high wind speed area at the top of the mountain, and the vortex on the leeward side. The uniform velocity gradient encryption results in an increase in data points that is directly proportional to the rate of change of wind speed. In the module M2.3, the mathematical expression of FAC2 is: Among them, FAC2 is expressed as the ratio of CALMET to CFD calculated values, and the proportion of data within the range of two standard deviations; The mathematical expression of FB is: Wherein, FB represents the CFD and CALMET calculation results, that is, the deviation ratio between the calculated values, and the superscript "—" indicates that the calculated results are averaged; The mathematical expression of the NMSE is: Among them, NMSE represents the standardized error index of the difference between CALMET and CFD calculation results. The footer "i" represents the ordinal number of the small area after the wind field is evenly divided into 20×20 small areas, a total of 400 small areas. i and CALMET i They represent the average values ​​of CFD calculation results and CALMET calculation results in the i-th small area respectively; The mathematical expression of R is: Where R represents the Pearson correlation coefficient.

10. The complex underlying surface atmospheric wind field optimization system according to claim 6, characterized in that: In the module M4, a data assimilation scheme of inverse square interpolation of distance is used to integrate the location data of actual weather station data and wind field data, assimilate them into the CALMET model, and perform hourly resolution wind field diagnosis and atmospheric diffusion optimization calculations; The data assimilation scheme of the inverse square distance interpolation is mathematically expressed as follows: Among them, (u,v)′2 is the component of the second-step wind field at the target location in the x and y directions, (u,v)1 is the first-step wind field component of the target location obtained by considering the terrain dynamics effect, overland flow and terrain blocking effect parameters, (u obs ,v obs ) k is the x- and y-direction components of the observed wind at the kth meteorological station; (u CFD ,v CFD ) i are the wind components in the x and y directions of the i-th interpolation point calculated and transformed by CFD, R, R k With R i are the user-defined first-step wind field weight, the distance from the kth meteorological station to the target location, and the distance from the i-th interpolation point to the target location.

Images