CN-122017793-A - Minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data

CN122017793ACN 122017793 ACN122017793 ACN 122017793ACN-122017793-A

Abstract

The invention discloses a minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data, which relates to the technical field of urban meteorological monitoring and data processing and comprises the following steps of multi-source data acquisition, data processing, optimization model construction, optimization solution and result output; through the self-adaptive sparse scanning strategy, intelligent data acquisition is carried out at the source, key area information is captured preferentially, scanning time can be effectively shortened, minute-level time updating and urban scale three-dimensional wind field real-time inversion with meter-level spatial resolution can be achieved, a unified multi-source collaborative optimization framework is created, the wide area coverage capacity of the laser radar, the precision of a ground weather station and the macroscopic physical consistency of a fluid dynamic numerical simulation model are subjected to deep fusion, and through construction of a data fidelity term and a background field constraint term, advantage complementation is achieved, a complete three-dimensional wind field which is continuous in space, physically consistent and remarkably improved in precision is obtained in the whole target area, and the anti-interference capacity is strong.

Inventors

ZHANG HAORAN
CAO XINGWEN
ZHAO MUHUA
LIU WENZHAO
ZHANG YANZHEN

Assignees

北京大学深圳研究生院

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (10)

1. A minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data is characterized by comprising the following steps: S1, multisource data acquisition, namely deploying at least one Doppler wind measuring laser radar in a target city area, and acquiring sparse radial wind speed observation data of wind fields in the city target area Simultaneously acquiring fixed-point three-dimensional wind speed and direction observation data acquired by ground weather stations in the area And low-resolution background wind field data covering a target area obtained by a numerical weather forecast model ; S2, data processing, namely unifying the acquired multi-source data into the same three-dimensional Cartesian coordinate system, performing time alignment and noise filtering, and collecting low-resolution background wind field data Interpolation to the same high-resolution grid as the wind field to be reconstructed to obtain a high-resolution background wind field ; S3, constructing an optimization model, namely marking a high-resolution three-dimensional wind field vector to be reconstructed as u, and constructing a target optimization function; S4, optimizing and solving, namely adopting an alternate direction multiplier method to iteratively solve the target optimization function, separating a regular term with an L1 norm by introducing auxiliary variables, decomposing the original problem into a plurality of sub-problems, and carrying out alternate updating until convergence conditions are met, so as to obtain the optimal high-resolution three-dimensional wind field solution ; S5, outputting a result, namely outputting and storing the optimal high-resolution three-dimensional wind field solution 。
2. The minute-scale urban microclimate wind farm sparse reconstruction method based on laser radar data of claim 1, wherein in step S1, doppler anemometry laser radar performs a preset adaptive sparse scan with a scan strategy of: and identifying a key area with severe wind field change based on building geographic information system data of a target area and priori wind field knowledge, distributing more scanning time and rays to carry out encryption sampling on the key area in a single scanning period, and carrying out sparse sampling on an area with relatively stable wind field.
3. The minute-scale urban microclimate wind farm sparse reconstruction method based on laser radar data according to claim 1, wherein in the step S2, the specific steps of data processing are as follows: S201, establishing a coordinate system, namely taking a central point of a target city area as an origin O, directing an X axis to the east, directing a Y axis to the north, and vertically upwards directing a Z axis to construct a three-dimensional Cartesian coordinate system as a carrier of a wind field u to be solved; S202, unifying coordinate conversion, namely using known coordinates of the Doppler wind-finding laser radar site ) Converting it into three-dimensional Cartesian coordinate system, and the converted expression is In the following For the range of the radar, For the azimuth angle, Is the elevation angle; The geographic coordinates fixed by each ground station are converted into coordinates in a target coordinate system through map projection calculation; all grid points of the numerical weather forecast model data are unified into a target Cartesian coordinate system through projection and height conversion to form a low-resolution three-dimensional background field; S203, setting a time window, wherein all radial wind speed data points acquired in the time window are reserved as an observation set of the window, the ground station data selects a minute average observation value corresponding to the central moment of the time window as a representative value of the window, and the numerical weather forecast model data selects a forecast closest to the initial or central moment of the reconstruction window in time as a background field of the window; S204, noise filtering and data cleaning, namely calculating the signal-to-noise ratio of each observation point, setting a signal-to-noise ratio threshold of-20 dB, discarding all the observation points lower than the threshold, and removing the areas with weak corresponding signals and dominant noise; S205, interpolation of low-resolution background wind field data, namely inputting the low-resolution background wind field data Each grid point has three components of u, v and w, and a wind field on the high-resolution reconstruction grid is output.
4. The method for sparse reconstruction of minute-scale urban microclimate wind farm based on laser radar data according to claim 3, wherein in step S205, the specific steps of interpolation of low-resolution background wind farm data are as follows: Adopting a tri-linear interpolation method to find a minimum cube unit containing each target point in a low-resolution background wind field grid for each target point on a high-resolution target grid; Independent scalar interpolation is respectively carried out on three wind speed components of u, v and w, namely, in the u direction, bilinear interpolation is firstly carried out on a vertical layer of a low-resolution grid twice to obtain u values at two virtual horizontal positions right above and right below a target point, then linear interpolation is carried out between the two values to obtain u component values at the target point, and the process is repeated on the v and w components; For points exceeding the grid range of the low-resolution background wind field in the target grid, adopting nearest neighbor interpolation or extrapolation according to boundary layer theory, and obtaining a high-resolution background wind field priori completely consistent with the u dimension of the wind field to be reconstructed after interpolation 。
5. The minute-scale urban microclimate wind farm sparse reconstruction method based on laser radar data according to claim 1, wherein in the step S3, the specific steps of constructing an optimization model are as follows: s301, constructing a laser radar data fidelity term, wherein the expression is as follows In the following The method comprises the steps that an observation operator for projecting a three-dimensional wind field to a laser radar scanning direction is shown, the observation operator is a large sparse matrix for collecting projection relations of all observed points in the whole wind field u, and L2 norm square represents least square fitting under Gaussian noise; s302, constructing a ground weather station data fidelity item for ensuring that the position of the reconstructed wind field at the ground weather station is consistent with an actual observed value, wherein the expression is as follows In the following The sampling operator is a matrix formed by extracting wind speed values of grid points corresponding to the positions of the meteorological stations from the whole wind field u, Setting the weight coefficient according to the measurement precision and the representativeness of the ground station; s303, constructing a fidelity term of numerical weather forecast model data, wherein the expression is as follows For providing a physically reasonable default guess in areas where the data is extremely sparse or where there is no data; S304, constructing a multiple sparse regularization term, wherein the expression is as follows In the following Is a linear difference operator, is used for calculating the space difference of the wind field u in the three directions of x, y and z, is used for representing the characteristic of segment smoothing of the wind field, Representing the amount of change in wind speed in three directions at each grid point, For a linear operator of the transformation of the wind field from physical space to feature space, representing the structured nature of turbulence, And Regularization parameters for controlling the two sparsity intensities are respectively represented; s305, integrating to obtain a target optimization function, wherein the expression is 。
6. The minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data of claim 5, wherein the observer is characterized by For each laser radar radial wind speed observation, determining all high-resolution grid cells through which scanning rays pass; for each grid cell traversed, according to its central coordinates And the azimuth angle of the scanning ray And elevation angle Calculating radial projection coefficients As the corresponding wind field component in the observation equation The coefficients corresponding to all observations are arranged in rows to form a sparse matrix 。
7. The minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data of claim 5, wherein the linear operator The method is trained from historical high-fidelity computational fluid dynamics simulation data or historical high-density laser radar scanning data.
8. The minute-scale urban microclimate wind farm sparse reconstruction method based on laser radar data according to claim 5, wherein in the step S4, the specific steps of optimizing and solving are as follows: S401, introducing auxiliary variables And Equivalently converting the target optimization function into the following constraint optimization problem, wherein the expression is as follows , , ; S402, constructing an augmented Lagrangian function of the constraint optimization problem, and initializing wind field variables Auxiliary variable 、 And dual variables And ; S403, performing iterative update, in the first step In the multiple iterations: Fixing , , And Updating wind field variables by solving a system of linear equations ; Fixing , And Updating auxiliary variables with soft threshold functions , ; Updating dual variables , ; S404, checking whether the original residual error and the dual residual error are smaller than a preset threshold value, if yes, stopping iteration and outputting As a means of If not, let And returns to step S403 to continue the iteration.
9. The method for sparse reconstruction of minute-scale urban microclimate wind farm based on lidar data of claim 8, wherein in step S403, the coefficient matrix of the linear equation set is In the following And And solving by adopting a pretreatment conjugate gradient method for penalty parameters of an alternate direction multiplier method.
10. The minute-scale urban microclimate wind farm sparse reconstruction method based on lidar data of claim 9, wherein in step S5, after iterative convergence, The method comprises the steps that a server files wind field data, marks a time stamp, and pushes the data to authorized subscribing users through a RESTful API interface, wherein the wind field data is the optimal high-resolution three-dimensional wind field reconstructed in the current 2-minute period.

Description

Minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data Technical Field The invention relates to the technical field of urban meteorological monitoring and data processing, in particular to a minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data. Background With the acceleration of the urban process, the urban microclimate effect has increasingly prominent influence on the aspects of life, air quality, low-altitude flight safety and the like of residents. The space-time evolution of complex wind field structures such as urban building group canyon wind, corner wind and the like is the core and difficulty of microclimate research, and how to accurately acquire urban three-dimensional wind field information in real time has a crucial strategic significance for smart city management, disaster early warning and low-altitude economic development. At present, urban wind field information is mainly obtained through network observation and computational fluid dynamics numerical simulation of ground weather stations, and the following problems still exist in the practical application of the traditional modes: The traditional ground meteorological station network observation data are affected by the cost and physical space of the meteorological stations, the site layout is quite sparse, the sparse observation network can provide single-point meteorological data, but cannot effectively capture local phenomena such as complex wind field distortion, vortex and mutation caused by dense building groups, the data updating frequency is low, the representativeness is not enough, the high-fidelity three-dimensional wind field distribution can be provided based on a physical equation through computational fluid dynamics numerical simulation, complex geometric modeling, high-quality grid dissection and the like are needed, the whole calculation time is huge, and the real-time monitoring requirement is difficult to meet. In recent years, wind lidar technology has become an important tool for atmospheric wind field detection due to the advantages of high accuracy and high space-time resolution. The device can actively emit laser beams, and inverts the wind speed in the sight line direction by detecting the Doppler frequency shift of a backward scattering signal of aerosol particles in the atmosphere, so that sparse wind field observation along a laser path is provided. However, the inversion method generally assumes that the wind field is horizontally uniform within the scan volume, which is not true in urban environments where terrain and construction are complex. To obtain a three-dimensional wind field, a time-consuming volume scan is typically required, often taking several tens of minutes or even longer to complete a complete, high-density scan of a larger urban area. In summary, the traditional urban microclimate wind field sparse reconstruction method has the problems that a sparse prior model is single, multiple sparse characteristics of urban wind field segmentation smoothing and structured turbulence are not fully combined, a multi-source data fusion mode is simple, deep synergy of laser radar, ground stations and NWP data is not realized in a unified optimization framework, optimization of a scanning strategy is not fully considered to maximize information acquisition efficiency, algorithm instantaneity is insufficient, and business operation requirements are difficult to meet. Disclosure of Invention The invention aims to provide a minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data, which aims to solve the problems of sparse observation and low calculation efficiency in the traditional wind field monitoring technology proposed by the background technology. Therefore, the invention provides a minute-scale urban microclimate wind field sparse reconstruction method based on laser radar data, which comprises the following steps: S1, multisource data acquisition, namely deploying at least one Doppler wind measuring laser radar in a target city area, and acquiring sparse radial wind speed observation data of wind fields in the city target area Simultaneously acquiring fixed-point three-dimensional wind speed and direction observation data acquired by ground weather stations in the areaAnd low-resolution background wind field data covering a target area obtained by a numerical weather forecast model; S2, data processing, namely unifying the acquired multi-source data into the same three-dimensional Cartesian coordinate system, performing time alignment and noise filtering, and collecting low-resolution background wind field dataInterpolation to the same high-resolution grid as the wind field to be reconstructed to obtain a high-resolution background wind field; S3, constructing an optimization model, namely marking a high-resolution three-dimensional wind field vector to be reconstructed as u,