Urban flow field construction design based on fusion of wind lidar and computational fluid mechanics

Abstract

The invention discloses an urban flow field construction design based on wind lidar and computational fluid mechanics fusion, and the design comprises the following steps: S1, employing CFD to construct an initial grid offline based on terrain, building and vegetation; s2, generating a sensing grid based on radar three-dimensional scanning; s3, generating a non-sensing grid based on the sensing grid and the initial grid; and S4, combining the sensing grid and the non-sensing grid, and constructing a micro-scale real-time flow field. Compared with the prior art, the method not only well utilizes high precision and high real-time performance of radar detection, but also completes airspace flow field relation construction through CFD modeling, and fuses the two, thereby achieving the purpose of improving the real-time performance of the micro-scale flow field under the condition of deployment of a small amount of equipment. Micro-scale high-precision flow field detection output can still be achieved, and good economic benefits are achieved.

Description

Technical Field

[0001] The present invention relates to the field of urban flow field construction, and in particular to an urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics. Background Art

[0002] In the field of meteorology, there has always been a technical bottleneck for real-time wind speed detection over a large range and at a microscale, especially for real-time observation of microscale flow fields in complex urban scenes. There is no mature technical solution or mature use case.

[0003] The current urban flow field detection mainly focuses on vertical monitoring and cruise monitoring. Vertical monitoring relies on a fixed in-situ vertical monitoring system built on high-rise buildings. By setting up atmospheric environment monitoring points at different heights, the observation equipment is directly installed in the real environment for long-term continuous online observation. This type of method is only suitable for flow field monitoring in characteristic areas or characteristic points. To achieve micro-scale flow field output at the city level, the equipment deployment cost is huge, and data synchronization and integration are difficult. Cruise monitoring is carried out through a vehicle-mounted detection system. Although it has functions such as automatic sampling and data processing, it is suitable for large-scale, fast-moving monitoring needs, but it is also unable to perform real-time global synchronous output at the city level.

[0004] As for the integrated use of computational fluid dynamics (CFD) and lidar, the current focus is on using mathematical models constructed with CFD to correct the measurement errors caused by radar detection, with the main focus on the accuracy correction dimension of the radar measurement points. Summary of the invention

[0005] The purpose of the present invention is to address the shortcomings of the existing urban flow field detection methods, in which vertical monitoring is only applicable to flow field monitoring of characteristic areas or characteristic points, and the equipment deployment cost is huge, and data synchronization and integration are difficult; cruise monitoring is suitable for large-scale, fast-moving monitoring needs, and cannot perform real-time global synchronous output at the city level. In view of the shortcomings of the above two flow field detection methods, an urban flow field construction design based on the fusion of wind measurement lidar and computational fluid dynamics is proposed.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics includes the following steps: S1. Based on terrain, buildings, and vegetation, the initial mesh is constructed using CFD offline; S2, based on radar 3D scanning, generate perception grid; S3, generating a non-perceived grid based on the perceived grid and the initial grid; S4. Combine the sensing grid and the non-sensing grid to construct a micro-scale real-time flow field.

[0007] As a further preferred embodiment of the present invention, S1 comprises the following specific steps: S11, obtaining terrain data, vegetation data and building data; S12, build STL model based on CFD offline; S13, setting grid boundary conditions; the grid boundary conditions include: velocity conditions at the inlet, no pressure at the outlet, no slip on the static wall of the ground building surface, and a specified shear force of 0 on the sky surface; S14, according to the velocity condition of the incident port, the space of the incident port is equally divided into a number of sectors, each sector corresponding to an incident angle; S15. Solve the STL model based on the standard tool of incompressible flow field in CFD to obtain several groups of spatial flow field relationships at different incident angles; S16. Perform grid division based on downscaling uniform sampling to obtain an initial grid.

[0008] As a further preferred embodiment of the present invention, the grid division comprises the following steps: The areas in the STL model are divided into open areas, uniform areas, densely built areas, densely covered vegetation areas, and no other special areas; All areas are graded; The higher the level, the denser the grid division.

[0009] As a further preferred embodiment of the present invention, S2 comprises the following specific steps: S21, obtaining the radial wind speed of the measuring point through the laser radar, and inverting the wind speed and wind direction; S22, based on the carrier-to-noise ratio lower limit threshold filtering mechanism, the carrier-to-noise ratio threshold is screened for wind speed and wind direction; S23, setting a threshold range of radial wind speed, and screening the radial wind speed by speed range; S24, performing a time consistency check on the wind speed and wind direction at the same distance layer, eliminating data with sudden changes in the time series, and screening the wind speed and wind direction for spatiotemporal consistency; the sudden change data is set to be data that differs from the median of the data in the local range where the data is located by more than n times the absolute deviation of the local median; S25, calculating the coordinates of the measuring points based on the data information of the laser radar measuring points; the data information includes the range gate, the pitch angle, the azimuth angle and the wind speed data; the wind speed data includes the wind speed and the wind direction; S26, using the wind speed data of the measuring point as a sensing input according to the grid corresponding to the measuring point coordinates; S27, performing interval matching between the actually measured wind speed of each grid, i.e., the wind speed of the measuring point after inversion, and the sector; S28, selecting an initial grid based on the sector, and calculating the weight of the perceived input data and the data corresponding to the initial grid ; S29, combining the perception input and the weight of the data corresponding to the perception input and the initial grid to obtain a perception grid.

[0010] As a further preferred embodiment of the present invention, S3 comprises the following specific steps: S31, using the KNN nearest neighbor algorithm to obtain k perception grids that are closest to the perception grid; S32, based on the inverse distance weighted method, the wind speed is derived by fusing k sensing grids; S33, after decomposing the wind speed, obtain the wind speed components of the non-sensing grid according to the wind speed inverse distance weighting method; S34. Synthesize the wind speed component vectors of the non-perceiving grid into the wind direction of the non-perceiving grid.

[0011] The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics proposed in the present invention has the following beneficial effects compared with the prior art: 1. The present invention not only makes good use of the high precision and high real-time performance of radar detection, but also completes the construction of airspace flow field relationship through CFD modeling, and integrates the two, so that even with a small number of equipment deployed, micro-scale high-precision flow field detection output can still be achieved, which has good economic benefits; 2. By dividing the feature space into grids, the representation of wind field information is discretized and digitized to facilitate subsequent data fusion processing 3. Through the high-precision real-time perception capability of the wind laser radar, a large amount of spatial wind information can be quickly output within the line of sight, providing as many basic data sets as possible for model fusion construction; 4. The nearest neighbor algorithm is used to efficiently search for the nearest perception grid, and combined with the CFD flow field relationship, non-perception grid flow field information beyond direct perception is constructed with relatively small computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 It is a schematic diagram of the process of establishing the initial CFD grid; Figure 3 is a schematic diagram of the fused grid. DETAILED DESCRIPTION

[0013] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0014] The present invention proposes a method based on radar perception detection and CFD fusion, which obtains high-precision real-time detection perception of part of the airspace through wind measurement laser radar, builds a high-resolution physical model in the airspace according to CFD, and obtains large-scale, micro-scale, high-precision, and high-real-time process results through numerical deduction through the relationship between perception and spatial models.

[0015] Embodiment 1: Combination Figure 1-2 , the urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics includes the following steps: S1. Based on terrain, buildings, and vegetation, the initial mesh is constructed offline using CFD.

[0016] The present invention will perform fine grid division in the airspace based on data such as terrain, buildings, vegetation, etc., and construct the fluid relationship of the grid in the entire airspace through a specific mathematical method as one of the inputs of this data fusion.

[0017] The CFD algorithm mainly considers factors such as terrain, vegetation coverage and building distribution. The terrain data and vegetation coverage map of the characteristic area can be obtained through the geospatial data cloud.

[0018] Regarding the urban building part, the urban buildings and their geometric shapes are fully modeled in the CFD solution, and all the building layout and shape details are constructed and refined to better solve the wind flow simulation in the complex urban environment. This technology has been widely used in urban urban building design and planning.

[0019] The specific steps include: S11. Acquire terrain data, vegetation data, and building data.

[0020] S12. Build STL model based on CFD offline.

[0021] The 3D mesh was generated using the snappyHexMesh tool of the open source CFD software Openfoam. The terrain DEM (Digital Elevation Model) data was converted into a smooth STL (STereoLithography) model; the vegetation information was mapped to the surface roughness; the building was placed on the surface of the terrain according to the coordinates, and the height was adjusted in proportion.

[0022] S13. Set grid boundary conditions; the grid boundary conditions include: velocity conditions at the inlet, no pressure at the outlet, no slip on the stationary wall of the ground building surface, and a specified shear force of 0 on the sky surface. The grid boundary conditions are the initial assumptions for solving the fluid relationship between spatial grids.

[0023] S14. According to the velocity condition of the incident port, the space of the incident port is equally divided into several sectors, each sector corresponding to an incident angle; the spatial grid of the incident port is divided into equally spaced sectors according to the angle, so as to obtain a richer fluid relationship between grids later.

[0024] According to the change of the incident port conditions, the space is divided into 36 sectors, one sector every 10 degrees. The model calculates the initial incident wind speed of the 36 sectors respectively, and the wind speed is set to 10m / s.

[0025] S15. The STL model is solved based on the standard tool of incompressible flow field in CFD to obtain several sets of spatial flow field relationships at different incident angles.

[0026] The model was solved using simpleFoam, a standard tool for incompressible flow fields of the open source CFD tool Openfoam. According to each set of inlet conditions, the corresponding vector wind speeds of all grids in the space were iteratively calculated to form a set of flow field relationships; 36 sets of spatial flow field relationships at different incident angles were obtained.

[0027] S16. Perform grid division based on downscaling uniform sampling to obtain an initial grid.

[0028] Specifically include: The areas in the STL model are divided into open areas, uniform areas, densely built areas, densely covered vegetation areas, and no other special areas.

[0029] All areas are graded.

[0030] The higher the level, the denser the grid division.

[0031] Since the characteristic area will form a finer grid division according to the differences in terrain and building texture, this increases the computing power pressure for the overall construction of a large range; the present invention adopts downscaling uniform sampling to reasonably sparse and uniform the grid entering the model fusion, making it the CFD initial grid, which greatly reduces the subsequent computing power requirements.

[0032] For example, the grid is encrypted in the building area of ​​key concern with a resolution of 10m; for the relatively mild and uniform area, a relatively sparse 30m resolution is used to save computing resources.

[0033] S2. Generate a perception grid based on radar 3D scanning.

[0034] The present invention adopts a three-dimensional scanning wind measurement laser radar to perform rapid real-time wind speed measurement in the airspace, providing basic true value data support for the data fusion process.

[0035] Specifically include the following: S21. Obtain the radial wind speed at the measuring point through LiDAR, and invert the wind speed and wind direction.

[0036] The wind speed and direction obtained by LiDAR inversion are used as the true value (perception grid) of the fusion model, and their quality control is particularly critical. Data can be filtered through S22-S24 to obtain more accurate data.

[0037] S22. Based on the carrier-to-noise ratio lower limit threshold filtering mechanism, the carrier-to-noise ratio threshold is screened for wind speed and wind direction.

[0038] The signal detected by the coherent Doppler lidar contains useful echo signals and noise signals. The carrier-to-noise ratio (CNR) is an important indicator to measure the detection capability and noise resistance of the system. Low carrier-to-noise ratio may cause false alarms of non-real wind speed signals, which is the key to quality control. Through a large number of experimental comparisons, this paper adopts a carrier-to-noise ratio lower limit threshold filtering mechanism to exclude these interference signals.

[0039] S23, setting a threshold range of radial wind speed, and screening the radial wind speed by speed range.

[0040] For Molas 3D, the radial wind speed can be expressed as: ; in, is the radar radial wind speed, is the horizontal wind speed in the wind field, is the azimuth, is the angle between the horizontal wind direction and the north direction. is the elevation angle. When nearby, Approximately 0, the radial wind speed is a small value. When it is substituted into the wind speed inversion operation, it may not converge, resulting in a large deviation in the inverted wind speed and direction. Therefore, the lower limit threshold of the radial wind speed can be set to ensure the overall inversion accuracy.

[0041] S24, performing a time consistency check on the wind speed and wind direction at the same distance layer, eliminating the sudden change data in the time series, and screening the wind speed and wind direction for spatiotemporal consistency. The sudden change data is set to be data that differs from the median of the data within the local range where the data is located by more than n times the absolute deviation of the local median.

[0042] Temporal and spatial consistency screening: The wind speed and direction at the same distance layer are checked for temporal consistency, and data with obvious mutations in the time series are eliminated. Mutation data is defined as data that differs from the local median by more than n times the local MAD within the specified window length (the multiple relationship can be adjusted according to actual conditions). MAD represents the absolute deviation of the median. A finite-length vector of scalar observations , Defined as: ;in, .

[0043] S25. Calculate the coordinates of the measuring points based on the data information of the laser radar measuring points; the data information includes the range gate, the pitch angle, the azimuth angle and the wind speed data; the wind speed data includes the wind speed and the wind direction.

[0044] S26. According to the grid corresponding to the coordinates of the measuring point, the wind speed data of the measuring point is used as the perception input.

[0045] S27, performing interval matching between the actually measured wind speed of each grid, that is, the wind speed of the measuring point after inversion, and the sector.

[0046] S28, selecting an initial grid based on the sector, and calculating the weight of the perceived input data and the data corresponding to the initial grid .

[0047] Select the initial CFD grid based on the sector and calculate the weight of the sensed input data and the CFD data of the grid , the formula is: ; in, Representation Perception Grid , Representation Grid Radar wind speed, Representation Grid The relationship between radar observation data and CFD is transferred through weights.

[0048] S29, combining the perception input and the weight of the data corresponding to the perception input and the initial grid to obtain a perception grid.

[0049] There may be multiple radar detection data in a grid, so it is necessary to remove duplicates and keep only one to enter the model calculation; after deduplication, the wind field data directly sensed by the radar and the weight relationship between these data and the initial grid constructed by CFD offline can be filled into the perception grid structure through perception data and weights. However, this is only a small part of the entire three-dimensional flow field. Most of the data still needs to be formed through subsequent data fusion, such as Figure 3 As shown, 3D is a three-dimensional radar, B300 is a ground-based radar, and the closer to red, the stronger the wind speed.

[0050] S3. Generate a non-perceived grid based on the perceived grid and the initial grid.

[0051] Specifically include the following: S31, use KNN nearest neighbor algorithm to obtain the closest to the perception grid A perception grid.

[0052] For non-perceptual grids, the KNN (K-Nearest Neighbors) nearest neighbor algorithm is used to find the closest This step uses the principle of spatial proximity to ensure that the non-sensing grid can obtain reasonable wind speed and direction estimates. Then the Inverse Distance Weight (IDW) method is used to fuse The wind speed value deduced by a sensing grid is as follows: , ; in Represents a non-aware grid The wind speed, Indicates the distance from the grid point The most recent Grid , Represents a non-aware grid CFD wind speed, Indicated by The grid weight and wind speed derived from the current CFD grid, Indicates Grids and grid points The Euclidean distance between .

[0053] S32, based on the inverse distance weighted method, fusion The wind speed is deduced from the sensing grid.

[0054] The deduction method for wind direction is similar to that for wind speed, and the inverse distance weighting method is also used to interpolate the wind direction of the non-perceptual grid. The difference from the deduction process of wind speed lies in two points: first, the wind direction is not transmitted according to the weight to ensure the wind direction change characteristics in the flow field model; second, the weighting of wind direction needs to be in the form of vector weighting, and the wind speed is first decomposed and then synthesized after weighting.

[0055] The steps for vector weighting of wind direction are: S33, after decomposing the wind speed, the wind speed components of the non-perceiving grid are obtained according to the wind speed inverse distance weighting method.

[0056] Assume that the CFD grid wind speed is , wind direction is , then the horizontal component of wind speed and the vertical component It can be calculated using the following formula: , ; According to the wind speed inverse distance weighted formula, the non-perceptual grid is obtained after weighting. The wind speed component and : , , in Indicates the distance from the grid point The most recent Grid , and Respectively represent The wind speed components of the grid.

[0057] S34. Synthesize the wind speed component vectors of the non-perceiving grid into the wind direction of the non-perceiving grid.

[0058] Wind speed component and Synthesize into a non-aware mesh Wind direction: , in Represents a non-aware grid wind direction.

[0059] S4. Combine the sensing grid and the non-sensing grid to construct a micro-scale real-time flow field.

[0060] The perception grid directly obtained by the wind-measuring lidar and the non-perception grid obtained by fusing the perception grid with the CFD model together constitute the three-dimensional wind field information in the feature space, providing an efficient construction method for urban-level micro-scale real-time meteorological scenes.

[0061] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any form, and any technical solution obtained by equivalent replacement or equivalent transformation falls within the protection scope of the present invention.

Claims

1. The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics is characterized by: The following steps are involved: S1. Based on terrain, buildings, and vegetation, the initial mesh is constructed using CFD offline; S2, based on radar 3D scanning, generate perception grid; S3, generating a non-perceived grid based on the perceived grid and the initial grid; S4. Combine the sensing grid and the non-sensing grid to construct a micro-scale real-time flow field.

2. The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics according to claim 1 is characterized in that: The S1 comprises the following specific steps: S11, obtaining terrain data, vegetation data and building data; S12, build STL model based on CFD offline; S13, setting grid boundary conditions; the grid boundary conditions include: velocity conditions at the inlet, no pressure at the outlet, no slip on the static wall of the ground building surface, and a specified shear force of 0 on the sky surface; S14, according to the velocity condition of the incident port, the space of the incident port is equally divided into a number of sectors, each sector corresponding to an incident angle; S15. Solve the STL model based on the standard tool of incompressible flow field in CFD to obtain several groups of spatial flow field relationships at different incident angles; S16. Perform grid division based on downscaling uniform sampling to obtain an initial grid.

3. The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics according to claim 2 is characterized in that: The grid division comprises the following steps: The areas in the STL model are divided into open areas, uniform areas, densely built areas, densely covered vegetation areas, and no other special areas; All areas are graded; The higher the level, the denser the grid division.

4. The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics according to claim 2 is characterized in that: The S2 comprises the following specific steps: S21, obtaining the radial wind speed of the measuring point through the laser radar, and inverting the wind speed and wind direction; S22, based on the carrier-to-noise ratio lower limit threshold filtering mechanism, the carrier-to-noise ratio threshold is screened for wind speed and wind direction; S23, setting a threshold range of radial wind speed, and screening the radial wind speed by speed range; S24, performing a time consistency check on the wind speed and wind direction at the same distance layer, eliminating data with sudden changes in the time series, and screening the wind speed and wind direction for spatiotemporal consistency; the sudden change data is set to be data that differs from the median of the data in the local range where the data is located by more than n times the absolute deviation of the local median; S25, calculating the coordinates of the measuring points based on the data information of the laser radar measuring points; the data information includes the range gate, the pitch angle, the azimuth angle and the wind speed data; the wind speed data includes the wind speed and the wind direction; S26, using the wind speed data of the measuring point as a sensing input according to the grid corresponding to the measuring point coordinates; S27, performing interval matching between the actually measured wind speed of each grid, i.e., the wind speed of the measuring point after inversion, and the sector; S28, selecting an initial grid based on the sector, and calculating the weight of the perceived input data and the data corresponding to the initial grid ; S29, combining the perception input and the weight of the data corresponding to the perception input and the initial grid to obtain a perception grid.

5. The urban flow field construction design based on the fusion of wind laser radar and computational fluid dynamics according to claim 4 is characterized in that: The S3 includes the following specific steps: S31, use KNN nearest neighbor algorithm to obtain the closest to the perception grid a perception grid; S32, based on the inverse distance weighted method, fusion The wind speed is derived from the sensing grids; S33, after decomposing the wind speed, obtain the wind speed components of the non-sensing grid according to the wind speed inverse distance weighting method; S34. Synthesize the wind speed component vectors of the non-perceiving grid into the wind direction of the non-perceiving grid.

Images