Wind measurement radar tilt self-calibration method, electronic device and storage medium

Abstract

The present application relates to wind measurement radar signal processing technology field, specifically to a kind of wind measurement radar tilt self-calibration method, electronic equipment and storage medium, collect multi-beam radial velocity, establish observation equation and least square inversion obtain three-dimensional wind vector sequence;In sliding time window, wind vector is de-meaned, and binary regression is constructed according to the vertical and horizontal linear coupling caused by tilt, and the closed-form solution of pitch and roll angle is derived from covariance term as initial value;Subsequently, under the full rotation model, iterative weighted regression of Huber loss is used, the weight of large residual sample is reduced, and the damage of outlier and heavy-tailed noise to angle estimation is suppressed;According to the final angle, the observation matrix is reconstructed and re-inverted, and the corrected three-dimensional wind speed and tilt angle are output.Compared with the prior art, the present application does not need external attitude sensor and parking calibration, is suitable for long-term continuous monitoring under natural wind field disturbance, improves three-dimensional inversion accuracy and reduces system deviation.

Description

Technical Field

[0001] This invention relates to the field of atmospheric wind field detection and wind measurement radar signal processing technology, and in particular to a wind measurement radar tilt self-calibration method and system based on closed-form angle estimation using multi-sample statistical covariance. Background Technology

[0002] Wind speed, wind direction, and their vertical profiles and spatiotemporal distribution information have fundamental value in scenarios such as wind power prediction, yaw and pitch control, turbine load assessment, low-altitude safety assurance, and meteorological monitoring. With the rapid deployment of complex terrain and offshore wind power, engineering sites have placed higher demands on wind measurement systems, requiring them to be "long-term continuous, maintenance-free, and traceable." Wind measurement radars are widely used due to their ability to continuously observe wind field profiles. However, during actual installation and long-term operation, radar platforms inevitably experience slight tilts such as pitch / roll, as well as attitude drift caused by base settlement and tower vibration. These tilts introduce beam pointing deviations, causing horizontal and vertical wind components to intersect, leading to accumulated wind vector inversion errors, especially significantly affecting vertical wind speed and turbulence statistics. Traditional calibration methods relying on manual leveling or external reference equipment are costly, time-consuming, and difficult to update adaptively over time. Therefore, there is an urgent need for a technical solution that can achieve online tilt self-calibration using radar's own observation data and simultaneously complete wind vector correction.

[0003] Existing technologies have attempted to estimate radar attitude deflection using statistical characteristics of wind fields. For example, Chinese patent application CN121276465A proposes constructing an error model based on the ratio of vertical wind speed to horizontal wind speed, establishing a mapping relationship between this ratio and horizontal wind direction, estimating the attitude deflection through data density weighting and robust fitting, and finally performing coordinate transformation on the beam direction and resubmitting it into the inversion model to output the corrected three-dimensional wind speed. However, this approach may still have limitations in engineering applications: First, the method chain explicitly relies on the horizontal wind direction and the "wind speed ratio" characteristic, and usually requires quality control and threshold screening (such as removing missing data and screening data with horizontal wind speeds greater than a set threshold). When the wind speed is low, the wind direction coverage is insufficient, or the data is intermittently missing, the update frequency and stability of the attitude deflection may be limited.

[0004] Therefore, for long-term operational wind-measuring radar engineering applications, there is an urgent need for a wind-measuring radar calibration scheme with online, continuous, and stably updated tilt correction capabilities to meet the requirements of long-term high-reliability monitoring of wind power and meteorology. Summary of the Invention

[0005] The purpose of this invention is to address the problems of existing wind radar calibration technologies, such as the need for quality control and threshold screening, and the limitations in update frequency and stability under conditions of low wind speed, insufficient wind direction coverage, or intermittent data gaps. This invention provides a wind radar tilt self-calibration method based on closed-form angle estimation using multi-sample statistical covariance, along with its electronic design and storage medium.

[0006] The objective of this invention can be achieved through the following technical solutions: As a first aspect of the present invention, a method for tilt self-calibration of a wind-measuring radar is provided, the method comprising the following steps: The observation values ​​of multibeam radial velocity from the wind-measuring radar at the same measurement time are obtained. An observation matrix is ​​constructed based on the nominal pointing unit vector of each beam. The observation values ​​of multibeam radial velocity are inverted within a preset time window to obtain an initial three-dimensional wind vector sequence. The initial three-dimensional wind vector sequence within the time window is processed to remove the mean, and a linear coupled regression model between the vertical and horizontal components caused by tilt is established. The covariance term between the horizontal and vertical components after removing the mean is calculated. Based on the covariance term, the closed-form solution of the regression coefficients of the linear coupled regression model is obtained to obtain the tilt coupling parameters, and the tilt coupling parameters are mapped to the initial value of the tilt angle, which includes the pitch angle and / or roll angle. Under the full rotation model, candidate tilt angles are constructed based on the initial tilt angle and a rotation matrix is ​​established. The initial three-dimensional wind vector sequence is rotated and corrected using the rotation matrix to obtain a corrected wind vector sequence. A linear relationship between the vertical component and the horizontal component is established for the corrected wind vector sequence. Iterative weighted regression is used to fit the linear relationship between the vertical component and the horizontal component to obtain the coefficient of the linear relationship. The residual linear dependence of the horizontal component on the vertical component, represented by the coefficient, is used as the objective function to iteratively update the candidate tilt angle until convergence, and the final tilt angle estimate is obtained. The observation matrix is ​​corrected and reconstructed based on the final tilt angle, and the multibeam radial velocity observations are re-inverted to output the corrected three-dimensional wind vector and the final tilt angle.

[0007] As a preferred technical solution, the process of obtaining the initial three-dimensional wind vector sequence is as follows: Within a preset time window, pulse compression and range gating are performed on each beam to obtain a complex echo sequence; Doppler spectrum estimation is performed on the complex echo sequence to obtain the main peak Doppler frequency shift; based on the wavelength of the wind radar transmitted signal, the main peak Doppler frequency shift is converted into radial velocity, thus obtaining the radial velocity sequence: An observation matrix is ​​constructed based on the nominal pointing unit vector. Least squares inversion is performed on the multibeam radial velocity sequence at each measurement time to obtain the initial three-dimensional wind vector sequence.

[0008] As a preferred technical solution, the linear coupling regression model between the vertical and horizontal components caused by tilt is expressed as follows: in, 、 For tilt angle parameters, This is the error term; , , These are the horizontal latitudinal component, horizontal longitudinal component, and vertical component after removing the mean.

[0009] As a preferred technical solution, the closed-form solution for solving the regression coefficients of the linear coupled regression model based on the covariance term is specifically implemented as follows: Calculate the discriminant: When discriminant When the value is greater than or equal to the discriminant threshold and the number of samples within the window is not less than the minimum effective sample number threshold, the initial value of the tilt angle parameter is obtained using the closed-form covariance solution: in, This is the initial value for the longitudinal tilt angle parameter; This is the initial value for the lateral tilt angle parameter; The variance of the horizontal latitudinal component; The variance of the horizontal meridional component; This represents the covariance between the horizontal latitudinal component and the horizontal longitudinal component; The covariance between the horizontal latitudinal component and the vertical component; This represents the covariance between the horizontal meridional component and the vertical component.

[0010] As a preferred technical solution, when the discriminant is less than the discriminant threshold or the number of samples in the window is insufficient, at least one of the following processes is performed to ensure the solvability and numerical stability of the closed angle estimation: expanding the time window, merging adjacent windows, delaying the update and using the tilt angle estimate from the previous moment, or marking the current tilt angle update result as invalid and not participating in subsequent matrix reconstruction.

[0011] As a preferred technical solution, the establishment of the linear relationship between the vertical component and the horizontal component is specifically implemented as follows: Candidate tilt angles are constructed based on the initial tilt angle, and a rotation matrix is ​​established. ; The corrected wind vector sequence is obtained by rotating and correcting the initial three-dimensional wind vector sequence within the time window. ); Establish a linear relationship between the vertical component and the horizontal component within the time window: in, The first The horizontal latitudinal component, horizontal longitudinal component, and vertical component of each sample after correction; The linear fitting coefficients represent the residual linear dependence of the vertical component on the horizontal component after correction. For the intercept term; This is the residual.

[0012] As a preferred technical solution, the linear relationship between the vertical component and the horizontal component is achieved by fitting the residuals using the Huber loss function and solving for the fitting coefficients through iterative weighted regression. ; Among them, iterative weighted regression is based on residuals The samples are weighted, and when the absolute value of the residual is less than or equal to the set residual threshold, squared loss is used; when the absolute value of the residual is greater than the set residual threshold, linear loss is used to reduce the weight of outlier samples; the iteration stops when the number of iterations reaches the preset upper limit. The obtained fitting coefficients Characterize the residual linear dependence of the corrected vertical component on the horizontal component, and construct the objective function: For candidate tilt angle Perform iterative updates to improve the objective function. The angle is gradually reduced until the preset convergence condition is met, and the final estimated value of the tilt angle is obtained.

[0013] As a preferred technical solution, the corrected three-dimensional wind vector is obtained through the following process: A rotation matrix is ​​constructed based on the final tilt angle estimate, and the tilt-corrected observation matrix is ​​formed by combining it with the nominal pointing unit vector; Using the tilt-corrected observation matrix, the radial velocity observation vector at any given time is inverted by least squares to obtain the corrected three-dimensional wind vector.

[0014] As a second aspect of the present invention, an electronic device is provided, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the wind measurement radar tilt self-calibration method as described above.

[0015] As a third aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the wind-measuring radar tilt self-calibration method as described above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1) This invention proposes an online tilt self-calibration and wind vector correction scheme for the attitude drift problem of long-term operation of wind measuring radar. The linear coupling of the vertical component to the horizontal component introduced by tilt is regarded as a statistically identifiable feature. The initial value of the tilt angle is obtained through covariance closed derivation within the time window. Furthermore, the large residual samples are weighted to reduce the damage of outliers and heavy-tailed noise to the estimated angle. Finally, the estimated angle is substituted back to reconstruct the observation matrix and the corrected wind field result is re-inverted and output. Thus, the continuous update and consistent presentation from the estimated angle to the final output of the correction are completed in the same processing link, which significantly improves the stability and engineering applicability of tilt correction.

[0017] 2) This invention proposes a tilt self-calibration mechanism based on closed-form angle estimation using multi-sample covariance: within a sliding time window, the mean of the three-dimensional wind vector sequence obtained from the initial inversion is removed, a binary regression model of the vertical and horizontal components is established, and the closed-form solution of the tilt angle parameter is derived from the covariance / variance term as the initial value for angle estimation, so that the tilt estimation does not depend on external benchmarks or manual calibration and can be automatically updated over time.

[0018] 3) This invention proposes an integrated process for angle estimation refinement and wind vector reconstruction in complex noise environments: under a fully rotating model, iterative weighted fitting with Huber loss is used to reduce the weight of large residual samples, suppressing the interference of clutter residue, abnormal radial velocity and heavy-tailed noise on tilt estimation; and the final tilt angle is used to reconstruct the tilt correction observation matrix, and the radial velocity is re-inverted by least squares, outputting the corrected three-dimensional wind speed vector and horizontal wind speed amplitude, realizing the synchronous update of tilt correction and wind field output under a unified model and the same time axis, improving the accuracy of three-dimensional inversion and reducing systematic bias. Attached Figure Description

[0019] Figure 1 This is a flowchart of a wind-measuring radar tilt self-calibration method based on covariance closed-form angle estimation according to the present invention.

[0020] Figure 2 This is a top view of the radar beam of the present invention.

[0021] Figure 3 This is a front view of the radar beam of the present invention.

[0022] Figure 4 This is a comparison chart of the deviation wind speed and the corrected wind speed according to the present invention. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0024] Example 1 This invention proposes an online tilt self-calibration and wind vector correction scheme for addressing the attitude drift problem of long-term wind-measuring radar operation. By relying solely on multi-beam radial velocity observations and nominal beam geometry, it achieves a closed-loop process for tilt angle estimation and three-dimensional wind vector correction without the need for external attitude sensors. This avoids engineering problems associated with traditional calibration, such as reliance on manual labor and external benchmarks, low update frequency, and difficulty in adaptive tracking over time. Its core improvement lies in treating the linear coupling of the vertical component to the horizontal component introduced by tilt as a statistically identifiable feature. Within a time window, the initial tilt angle value is obtained through closed-form covariance derivation. Furthermore, large residual samples are weighted to reduce the impact of outliers and heavy-tailed noise on the estimated angle. Finally, the estimated angle is substituted back to reconstruct the observation matrix and inverted to output the corrected wind field result. This allows for continuous updating and consistent presentation from angle estimation to the final corrected output within the same processing link, significantly improving the stability and engineering applicability of tilt correction.

[0025] like Figure 1 As shown, the specific implementation steps of this method are as follows: S1 acquires the multi-beam radial velocity observations of the wind-measuring radar at the same measurement moment, and acquires the nominal pointing unit vector of each beam; S2 constructs an observation matrix based on the nominal pointing unit vector, performs least-squares inversion on the multibeam radial velocity observations, and obtains the initial three-dimensional wind vector at that moment; repeats the above process within a preset time window to obtain the initial three-dimensional wind vector sequence; S3 performs mean-removal processing on the initial three-dimensional wind vector sequence within the time window, establishes a linear coupled regression model between the vertical and horizontal components caused by tilt, and calculates the covariance term between the mean-removed horizontal and vertical components. S4 solves the closed-form solution of the regression coefficients of the linear regression model based on the covariance term, obtains the initial value of the tilt coupling parameter, and maps the initial value of the tilt coupling parameter to the initial value of the tilt angle, which includes the pitch angle and / or roll angle. In the full rotation model, S5 constructs a rotation matrix based on candidate tilt angles, performs rotation correction on the initial three-dimensional wind vector sequence to obtain a corrected wind vector sequence, and establishes a linear relationship between the vertical component and the horizontal component of the corrected wind vector sequence. Iterative weighted regression based on Huber loss is used to fit the linear relationship, so that large residual samples are reduced in weight during the fitting process to obtain the coefficients of the linear relationship. The residual linear dependence of the horizontal component on the vertical component, represented by the coefficients, is used as the objective function to iteratively update the candidate tilt angles until convergence, and the final tilt angle is obtained. S6 corrects and reconstructs the observation matrix based on the final tilt angle, performs re-inversion on the multibeam radial velocity observations, and outputs the corrected three-dimensional wind vector and the final tilt angle.

[0026] Furthermore, in step S3, a time window containing samples from multiple consecutive time points is selected. The time window can be a sliding window or a fixed window, and the number of samples within the window is not less than a preset minimum number of samples. ;Calculate the sample mean of the horizontal and vertical components of the initial three-dimensional wind vector sequence obtained in step S2 within the window, and perform mean removal processing to obtain: in, , , These are the mean-free horizontal latitudinal component, horizontal longitudinal component, and vertical component, respectively. , , This is the original inverted wind vector; , , This is the sample mean.

[0027] Based on the linear coupling relationship between the vertical and horizontal components caused by tilt, a binary regression model is constructed: in, 、 For tilt angle parameters, The error term is represented; and the regression model is expressed in matrix form. ,in, .

[0028] Furthermore, in step S4, the covariance / variance term is calculated based on the mean-removed samples within the time window: in, The variance of the horizontal latitudinal component; The variance of the horizontal meridional component; This represents the covariance between the horizontal latitudinal component and the horizontal longitudinal component; The covariance between the horizontal latitudinal component and the vertical component; This represents the covariance between the horizontal meridional component and the vertical component.

[0029] Calculate the discriminant: When discriminant Discriminant threshold And the number of samples within the window is not less than the minimum effective sample number threshold. At that time, the initial value of the tilt angle parameter is obtained by using the covariance closed-form solution: in, , where is the initial value of the longitudinal tilt angle parameter, corresponding to the influence coefficient of radar pitch attitude on the vertical component; This is the initial value of the lateral (roll) tilt angle parameter, corresponding to the influence coefficient of the radar roll attitude on the vertical component.

[0030] The tilt coupling parameter in the above formula is the equivalent parameter of the tilt angle under small-angle linearization, which represents the crosstalk intensity from the horizontal wind component to the vertical wind component. The regression coefficient of this invention and the tilt angle parameter in the corresponding direction are the corresponding linearization relationship. Therefore, this application uses the initial values ​​of the longitudinal tilt angle parameter and the initial values ​​of the lateral tilt angle parameter based on the closed solution of the covariance term as the initial estimates of the corresponding tilt angle.

[0031] when If the number of samples in the window is insufficient, perform at least one of the following processes to ensure the solvability and numerical stability of the closed-form angle estimation: expand the time window, merge adjacent windows, delay the update and use the tilt angle estimate from the previous time step, or mark the current tilt angle update result as invalid and not participate in subsequent matrix reconstruction.

[0032] Furthermore, in step S5, candidate tilt angles are constructed based on the initial tilt angle value, and a rotation matrix is ​​established. ， The corrected wind vector sequence is obtained by rotating and correcting the initial three-dimensional wind vector sequence within the time window. Establish a linear relationship between the vertical component and the horizontal component within the time window: in, The first The horizontal latitudinal component, horizontal longitudinal component, and vertical component of each sample after correction; The linear fitting coefficients represent the residual linear dependence of the vertical component on the horizontal component after correction. For the intercept term; This is the residual.

[0033] The Huber loss function is used to fit the residuals, and the fitting coefficients are solved by iterative weighted regression. This causes samples with large residuals to be weighted less in the fitting process. The Huber loss function is: Among them, iterative weighted regression is based on residuals Weighting the samples, when the residuals When the absolute value of the residual is less than or equal to the set residual threshold (small residual, normal sample), squared loss is used, which is equivalent to ordinary least squares, and sample weights are preserved; when the residual... When the absolute value of the residual exceeds the set residual threshold (large residual, outlier sample), linear loss is used to reduce the weight of the outlier sample; when the number of iterations reaches the preset upper limit... Stop iterating when the time comes.

[0034] The obtained fitting coefficients Characterize the residual linear dependence of the corrected vertical component on the horizontal component, and construct the objective function: And the candidate tilt angle Perform iterative updates to make the objective function The angle is gradually reduced until the preset convergence condition is met, thus obtaining the final tilt angle.

[0035] Furthermore, in step S6, the final tilt angle estimate is obtained. Then, construct the rotation matrix. And based on the nominal pointing unit vector The observation matrix after tilt correction is formed: in, This is the observation matrix after tilt correction; For each nominal pointing unit vector of the radar; This is the corrected beam vector.

[0036] The corrected wind vector is obtained by performing a least-squares re-inversion on the radial velocity observation vector at any given time. in, The corrected wind vector; This is the radial velocity observation vector; The inverse of the coefficient matrix; This is the final corrected three-dimensional wind vector.

[0037] The proposed wind-measuring radar tilt self-calibration method directly utilizes the statistical coupling characteristics of the wind vector sequence obtained by multi-beam inversion within a time window. It quickly provides the initial value of the tilt angle through closed-form covariance calculation and further employs iterative weighted fitting to suppress the influence of outliers and heavy-tailed noise on the angle estimation. Finally, the estimated tilt angle is substituted back to reconstruct the observation matrix and least-squares re-inversion is performed to output the corrected three-dimensional wind vector and horizontal wind speed amplitude. Thus, without adding external benchmarks or relying on wind direction-ratio mapping, it achieves online, continuous, and stably updated tilt correction capabilities, meeting the urgent need for long-term high-reliability monitoring of wind power and meteorology.

[0038] Example 2 As one specific embodiment of the present invention, this embodiment uses a four-beam wind-measuring radar, and the radar parameters and wind field settings are as follows: Design the pitch angle (angle between beam and vertical): ; Pulse repetition period: ; Pulse width: .

[0039] S1. Beam geometry and observation model construction: like Figure 2 As shown, let the four nominal directions (north / east / south / west) of the radar under ideal installation be unit vectors. .

[0040] like Figure 3 As shown, assume that the beams in all four directions are... Leaning away from vertical, causing Then we can define: When there is a small tilt (pitch / roll), the actual beam pointing is determined by the rotation matrix. Applying the action to the nominal pointer yields: S2, Four-beam radial velocity observations satisfy the linear observation equation: Radial velocity acquisition: Within each observation period, the following processing is performed on each of the four beams to obtain the radial velocity sequence: Pulse compression and range gate selection: Matched filtering (or equivalent pulse compression) is performed on the echo I / Q data to obtain the complex echo sequence of the range gate in the range dimension.

[0041] Doppler spectrum estimation: Perform Doppler FFT or spectral peak tracking on the pulse sequence within the coherent processing time (CPI) to obtain the Doppler frequency shift of the main peak.

[0042] Radial velocity conversion: Get the first At any moment, beam Radial velocity. Threshold screening / reweighting can be applied to samples with low SNR or abnormal quality indicators.

[0043] Initial inversion: Three-dimensional wind vector estimation ignoring tilt.

[0044] Using nominal matrix Perform least-squares inversion at each time step: Obtain the initial wind vector sequence .

[0045] S3, multi-sample covariance closed-form angle estimation: A sliding time window of length M is selected, containing samples from multiple consecutive time points. The time window can be a sliding window or a fixed window, and the number of samples within the window is not less than a preset minimum number of samples. ;Calculate the sample mean of the horizontal and vertical components of the initial three-dimensional wind vector sequence obtained within the window, and perform mean removal processing to obtain: Based on the linear coupling relationship between the vertical and horizontal components caused by tilt, a binary regression model is constructed: in, , For tilt angle parameters, The error term is represented by the regression model in matrix form. ,in .

[0046] Calculate the covariance / variance term based on the mean-removed samples within the time window: And calculate the discriminant: S4, when the discriminant is... And the number of samples within the window is not less than At that time, the initial value of the tilt angle parameter is obtained by using the covariance closed-form solution: when If the number of samples in the window is insufficient, perform at least one of the following processes to ensure the solvability and numerical stability of the closed-form angle estimation: expand the time window, merge adjacent windows, delay the update and use the tilt angle estimate from the previous time step, or mark the current tilt angle update result as invalid and not participate in subsequent matrix reconstruction.

[0047] S5, Angle Refinement under Full Rotation Model: Candidate tilt angles are constructed based on the initial tilt angle, and a rotation matrix is ​​established. The corrected wind vector sequence is obtained by rotating and correcting the initial three-dimensional wind vector sequence within the time window. Establish a linear relationship between the vertical component and the horizontal component within the time window: in, The residuals are used; the Huber loss function is applied to fit the residuals. The Huber loss function is: The fitting coefficients were solved by iterative weighted regression. This causes large residual samples to be weighted less during fitting.

[0048] The obtained fitting coefficients Characterize the residual linear dependence of the corrected vertical component on the horizontal component, and construct the objective function: And the candidate tilt angle Perform iterative updates to make the objective function The slope is gradually reduced until the preset convergence condition is met, thus obtaining the final tilt angle. In the iterative weighted regression, the slope is determined based on the residuals. Weighting samples to satisfy The samples are weighted less during the fitting process; iteration stops when any of the following conditions are met: the number of iterations reaches a preset upper limit. .

[0049] S6. Obtain the final tilt angle estimate. Then, construct the rotation matrix. And based on the nominal pointing unit vector The observation matrix after tilt correction is formed: A The least-squares re-inversion is performed on the radial velocity observation vector at any given time to obtain the corrected three-dimensional wind vector: This invention can directly utilize the statistical coupling characteristics of the wind vector sequence obtained by multibeam inversion within a time window, quickly provide the initial value of the tilt angle through closed-form covariance calculation, and further employ iterative weighted fitting to suppress the influence of outliers and heavy-tailed noise on the angle estimation. Finally, the estimated tilt angle is substituted back to reconstruct the observation matrix and least-squares re-inversion is performed to output the corrected three-dimensional wind vector and horizontal wind speed amplitude. Figure 4 As shown, this invention achieves online, continuous, and stably updated tilt correction capabilities without adding external benchmarks or relying on wind direction-ratio mapping, meeting the urgent need for long-term, highly reliable monitoring of wind power and meteorology.

[0050] Example 3 This application also provides an electronic device, including: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-described wind radar tilt self-calibration method based on covariance closed-form angle estimation. In addition to the processors, memory, and interfaces described above, any data processing device in the embodiments may also include other hardware depending on the actual function of the data processing device, which will not be elaborated further.

[0051] Example 4 This application also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the aforementioned wind radar tilt self-calibration method based on covariance closed-form angle estimation. The computer-readable storage medium can be an internal storage unit of any data-processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data-processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data-processing device, and can also be used to temporarily store data that has been output or will be output.

[0052] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for tilt self-calibration of a wind-measuring radar, characterized in that, The method steps include: The observation values ​​of multibeam radial velocity from the wind-measuring radar at the same measurement time are obtained. An observation matrix is ​​constructed based on the nominal pointing unit vector of each beam. The observation values ​​of multibeam radial velocity are inverted within a preset time window to obtain an initial three-dimensional wind vector sequence. The initial three-dimensional wind vector sequence within the time window is processed to remove the mean, and a linear coupled regression model between the vertical and horizontal components caused by tilt is established. The covariance term between the horizontal and vertical components after removing the mean is calculated. Based on the covariance term, the closed-form solution of the regression coefficients of the linear coupled regression model is obtained to obtain the tilt coupling parameters, and the tilt coupling parameters are mapped to the initial value of the tilt angle, which includes the pitch angle and / or roll angle. Under the full rotation model, candidate tilt angles are constructed based on the initial tilt angle and a rotation matrix is ​​established. The initial three-dimensional wind vector sequence is rotated and corrected using the rotation matrix to obtain a corrected wind vector sequence. A linear relationship between the vertical component and the horizontal component is established for the corrected wind vector sequence. Iterative weighted regression is used to fit the linear relationship between the vertical component and the horizontal component to obtain the coefficient of the linear relationship. The residual linear dependence of the horizontal component on the vertical component, represented by the coefficient, is used as the objective function to iteratively update the candidate tilt angle until convergence, and the final tilt angle estimate is obtained. The observation matrix is ​​corrected and reconstructed based on the final tilt angle, and the multibeam radial velocity observations are re-inverted to output the corrected three-dimensional wind vector and the final tilt angle.

2. The wind-measuring radar tilt self-calibration method according to claim 1, characterized in that, The process of obtaining the initial three-dimensional wind vector sequence is as follows: Within a preset time window, pulse compression and range gating are performed on each beam to obtain a complex echo sequence; Doppler spectrum estimation is performed on the complex echo sequence to obtain the main peak Doppler frequency shift; based on the wavelength of the wind radar transmitted signal, the main peak Doppler frequency shift is converted into radial velocity, thus obtaining the radial velocity sequence: An observation matrix is ​​constructed based on the nominal pointing unit vector. Least squares inversion is performed on the multibeam radial velocity sequence at each measurement time to obtain the initial three-dimensional wind vector sequence.

3. The wind-measuring radar tilt self-calibration method according to claim 1, characterized in that, The linear coupling regression model between the vertical and horizontal components caused by tilt is expressed as follows: in, 、 For tilt angle parameters, This is the error term; , , These are the horizontal latitudinal component, horizontal longitudinal component, and vertical component after removing the mean.

4. The wind-measuring radar tilt self-calibration method according to claim 1, characterized in that, The closed-form solution for the regression coefficients of the linear coupled regression model is obtained based on the covariance term, and is implemented as follows: Calculate the discriminant : in, The variance of the horizontal latitudinal component; The variance of the horizontal meridional component; This represents the covariance between the horizontal latitudinal component and the horizontal longitudinal component; When discriminant When the discriminant threshold is greater than or equal to the threshold and the number of samples within the window is not less than the minimum effective sample number threshold, the initial value of the tilt angle parameter is obtained using the closed-form covariance solution: in, This is the initial value for the longitudinal tilt angle parameter; This is the initial value for the lateral tilt angle parameter; The covariance between the horizontal latitudinal component and the vertical component; This represents the covariance between the horizontal meridional component and the vertical component.

5. The wind-measuring radar tilt self-calibration method according to claim 4, characterized in that, When the discriminant is less than the discriminant threshold or the number of samples in the window is insufficient, at least one of the following processes is performed to ensure the solvability and numerical stability of the closed angle estimation: expanding the time window, merging adjacent windows, delaying the update and using the tilt angle estimate from the previous moment, or marking the current tilt angle update result as invalid and not participating in subsequent matrix reconstruction.

6. The wind-measuring radar tilt self-calibration method according to claim 1, characterized in that, The establishment of a linear relationship between the vertical component and the horizontal component is specifically implemented as follows: Candidate tilt angles are constructed based on the initial tilt angle, and a rotation matrix is ​​established. ; The corrected wind vector sequence is obtained by rotating and correcting the initial three-dimensional wind vector sequence within the time window. ); Establish a linear relationship between the vertical component and the horizontal component within the time window: in, The first The horizontal latitudinal component, horizontal longitudinal component, and vertical component of each sample after correction; The linear fitting coefficients represent the residual linear dependence of the vertical component on the horizontal component after correction. For the intercept term; It represents the residual.

7. The wind-measuring radar tilt self-calibration method according to claim 6, characterized in that, The linear relationship between the vertical component and the horizontal component is described by fitting the residuals using the Huber loss function, and the linear fitting coefficients are solved through iterative weighted regression. and intercept term ; Among them, iterative weighted regression is based on residuals The samples are weighted, and when the absolute value of the residual is less than or equal to the set residual threshold, squared loss is used; when the absolute value of the residual is greater than the set residual threshold, linear loss is used to reduce the weight of outlier samples; the iteration stops when the number of iterations reaches the preset upper limit. The obtained fitting coefficients Characterize the residual linear dependence of the corrected vertical component on the horizontal component, and construct the objective function: For candidate tilt angle Perform iterative updates to improve the objective function. The angle is gradually reduced until the preset convergence condition is met, and the final estimated value of the tilt angle is obtained.

8. The wind-measuring radar tilt self-calibration method according to claim 1, characterized in that, The corrected three-dimensional wind vector is obtained as follows: a rotation matrix is ​​constructed based on the final tilt angle estimate, and a tilt-corrected observation matrix is ​​formed by combining it with the nominal pointing unit vector; using the tilt-corrected observation matrix, the radial velocity observation vector at any time is subjected to least-squares re-inversion to obtain the corrected three-dimensional wind vector.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the wind-measuring radar tilt self-calibration method as described in any one of claims 1-8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the wind-measuring radar tilt self-calibration method as described in any one of claims 1-8.

Images