Method for correcting wind field measurement error of laser radar during rainfall

Abstract

The invention discloses a method for correcting wind field measurement errors of a laser radar during rainfall, and the method comprises the following steps: obtaining original spectrum data collected by the laser radar during rainfall, and removing background noise; fitting the power spectrum with the double-peak structure by adopting a double-Gaussian model to obtain an aerosol Doppler frequency and a raindrop Doppler frequency; after the power spectrum of the single-peak structure is identified as a rainfall spectrum based on spectral width and skewness parameters, aerosol and raindrop signals are separated by adopting a double-Gaussian model; calculating a radial wind speed by using the separated aerosol Doppler frequency and inverting three-dimensional wind field information; carrying out raindrop / aerosol signal separation and inversion on the original spectrum data of the plurality of rainfall periods recorded by the laser radar, and carrying out comparison verification with gradient tower synchronous data; turbulent flow parameters of the stable fluctuating wind are corrected; time-varying average wind speed correction turbulence parameters are extracted from the non-stationary fluctuating wind; and correcting the characteristic parameters of the downwind pulsating wind, and verifying the consistency of the corrected data and the three-dimensional ultrasonic anemograph.

Description

Technical Field

[0001] The present invention relates to the field of meteorological measurement technology, and in particular to a method for correcting errors in wind field measurement by a laser radar during rainfall. Background Art

[0002] With the increasing demand for accuracy of wind field measured data, lidar, as an advanced remote sensing measurement technology, has been widely used in the field of wind field measurement due to its advantages such as small size, wide measurement range and easy portability.

[0003] LiDAR (LiDAR) uses the Doppler shift principle to measure wind speed and direction by emitting laser pulses and receiving light signals scattered back from particles in the atmosphere. This provides a non-contact, continuous, and real-time measurement method. Although LiDAR can provide highly accurate wind field data under clear sky conditions, its measurement reliability is severely affected in rainy conditions. During rainfall, LiDAR receives both aerosol backscatter and raindrop scattered signals, resulting in a complex bimodal power spectrum. Because the intensity of raindrop signals can be greater than that of aerosol signals, LiDAR may mistake the raindrop signals for aerosol signals. Consequently, the data obtained after subsequent inversion reflects the falling raindrop velocity rather than the actual wind speed. This error is particularly pronounced in vertical wind speed measurements.

[0004] To solve this problem, a method for correcting lidar measurement errors during rainfall is proposed, aiming to improve the accuracy of lidar wind speed measurements during rainfall. Summary of the Invention

[0005] In response to the above problems, the present invention provides a method for correcting laser radar wind field measurement errors during rainfall.

[0006] The purpose of the present invention is achieved by adopting the following technical solutions:

[0007] A method for correcting errors in laser radar wind field measurements during rainfall, comprising the following steps:

[0008] S1. Obtain the original spectrum data of the lidar during rainfall and remove its background noise;

[0009] S2. determining the structure type of the power spectrum in the original spectrum data;

[0010] If it is a double-peak structure, the aerosol Doppler frequency and raindrop Doppler frequency are obtained by fitting the double-Gaussian model;

[0011] If it is a single-peak structure, determine whether it is a rainfall spectrum based on the preset spectral width and skewness parameters. If it is a rainfall spectrum, separate the aerosol Doppler frequency and raindrop Doppler frequency using a double Gaussian model.

[0012] S3, calculate the radial wind speed by the aerosol Püller frequency, and obtain the three-dimensional wind field information by inversion using the DBS four-beam wind field inversion algorithm;

[0013] S4, executing S2-S4 on the raw spectrum data of the same laser radar in multiple different rainfall periods;

[0014] S5. Determine the wind field type during different rainfall periods;

[0015] If the wind is steady and pulsating, the turbulence parameters are corrected by fitting the low-frequency band of the measured wind spectrum through the Von Karman spectrum;

[0016] If the wind is non-stationary and fluctuating, the time-varying average wind speed is extracted by discrete wavelet transform, and the turbulence parameters are corrected based on the non-stationary model.

[0017] As a preferred embodiment, in S3, the calculation formula of the radial wind speed is:

[0018]

[0019] Where λ is the laser wavelength, f D It is the difference between the frequency at the maximum power spectrum intensity and the zero frequency point 49.26;

[0020] The DBS four-beam wind field inversion algorithm calculates the three-dimensional wind speed components using the following formula:

[0021]

[0022] Among them, V losN 、V losE 、V losS 、V losW are the radial wind speeds in four directions respectively; θ is the elevation angle of the laser beam.

[0023] As a preferred embodiment, the expression of the double Gaussian model is:

[0024]

[0025] Among them, I a , I r are the signal strengths of aerosols and raindrops, respectively; σ a , σ r are the spectral widths of the two respectively.

[0026] As a preferred embodiment, in the non-stationary wind speed model, the turbulence intensity is defined as:

[0027]

[0028] The gust factor is calculated using the following formula:

[0029]

[0030] Among them, t g is the duration of the gust.

[0031] As a preferred embodiment, the preset conditions for identifying the unimodal structured rainfall spectrum are:

[0032] Spectral width greater than 8 or skewness greater than 0.2 or less than -0.1;

[0033] When fitting the unimodal rainfall spectrum, the Doppler frequency fr of raindrops at adjacent moments at the same height is fixed to reduce overfitting.

[0034] As a preferred method, the discrete wavelet transform adopts the db10 wavelet basis, the maximum decomposition layer number n is log2(N), where N is the length of the wind speed time history, and the optimal decomposition layer is determined by wavelet energy mutation to extract the time-varying average wind speed.

[0035] As a preferred embodiment, the step S5 further includes the following steps:

[0036] Obtain the preset number of rainfall times recorded by the lidar and determine the abnormal sudden changes in horizontal and vertical wind speeds;

[0037] The data of the three-dimensional ultrasonic anemometer is obtained, and the wind speed, wind direction and turbulence parameters before and after correction are obtained after removing the interference of the wake effect.

[0038] As a preferred embodiment, in the non-stationary wind speed model, the turbulence intensity is defined as:

[0039]

[0040] The gust factor is calculated using the following formula:

[0041]

[0042] Among them, t g is the duration of the gust.

[0043] As a preferred method, the output data of the lidar includes Level 0 raw spectrum data, Level 1 instantaneous wind speed data and Level 2 average wind profile data; wherein, the raw spectrum data is stored in CSV format, the first 300 points are background noise, each distance library corresponds to 100 points, and the distance resolution is 15m.

[0044] The beneficial effects of the present invention are as follows: in the prior art, wind speed inversion is usually performed only by single-peak power under clear sky conditions. The present invention adopts a double-peak spectrum fitting method of a double Gaussian model and designs the identification of spectral width and skewness parameters for the single-peak rainfall spectrum under preset conditions to achieve the separation of raindrop signals and aerosol signals.

[0045] In addition, to address the non-stationary characteristics during rainfall, a correction method combining Von Karman spectrum fitting (stationary pulsating wind) and discrete wavelet transform (non-stationary pulsating wind) is proposed. The wind speed time series is dynamically decomposed using the db10 wavelet basis, and the optimal decomposition layer is determined through energy mutation. The time-varying average wind speed is extracted, significantly improving the calculation accuracy of the gust factor and turbulence intensity. Therefore, the present invention achieves continuous, high-precision wind field measurement during rainfall, providing key data support for research on extreme weather such as typhoons and rainstorms, and filling the gap in wind field measurement technology under rainfall conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0047] Figure 1 is the raw power spectrum data recorded by the lidar;

[0048] Figure 2 It is the separation process of rainfall power spectrum signal;

[0049] Figure 3 is the rainfall power spectrum with a bimodal distribution;

[0050] Figure 4 is the original echo power spectrum (normalized);

[0051] Figure 5 is the original echo power spectrum (normalized);

[0052] Figure 6 It is the relationship between rainfall intensity and spectrum width / skewness of rainfall power spectrum;

[0053] Figure 7 is the spectral width / skewness threshold of the rainfall power spectrum;

[0054] Figure 8 This is the LiDAR meteorological module data on July 6 (left) and August 6 (right);

[0055] Figure 9 This is a comparison of wind speed and direction from a 3D anemometer (40m) and a lidar (42m) during rainfall;

[0056] Figure 10Comparison of wind speed and direction between the lidar (42m) and the 3D anemometer (40m) before and after correction;

[0057] Figure 11 This is a comparison of wind spectra at 160m and 320m heights using lidar and 3D ultrasonic anemometer;

[0058] Figure 12 This is the decomposition diagram of the downwind DWT of the lidar;

[0059] Figure 13 This is the downwind DWT exploded view of the three-dimensional ultrasonic anemometer;

[0060] Figure 14 This is the decomposition diagram of the lidar's crosswind DWT;

[0061] Figure 15 This is the crosswind DWT exploded view of the three-dimensional ultrasonic anemometer;

[0062] Figure 16 The optimal decomposition layer for the downwind DWT method: (left) LiDAR; (right) 3D anemometer

[0063] Figure 17 This is the fluctuating wind speed extracted based on the DWT method in the downwind direction: (left) lidar; (right) 3D wind instrument

[0064] Figure 18 This is the fluctuating wind speed extracted based on the DWT method in the downwind direction: (left) lidar; (right) 3D wind instrument

[0065] Figure 19 The optimal decomposition layer for the across-wind DWT method: (left) LiDAR; (right) 3D anemometer

[0066] Figure 20 The fluctuating wind speed in the crosswind direction is extracted based on the DWT method: (left) LiDAR; (right) 3D wind instrument

[0067] Figure 21 The fluctuating wind speed in the crosswind direction is extracted based on the DWT method: (left) LiDAR; (right) 3D wind instrument

[0068] Figure 22 It is a comparison of the downwind pulsating wind spectra of the lidar (42m) and the three-dimensional ultrasonic anemometer (40m);

[0069] Table 1 shows the comparison of wind turbulence parameters before and after lidar calibration with those of the three-dimensional ultrasonic anemometer;

[0070] Table 2 compares the wind turbulence parameters before and after Lidar correction with those of CAST3 (non-stationary model). DETAILED DESCRIPTION

[0071] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0072] An embodiment of the present disclosure provides a method for correcting laser radar wind field error measurements during rainfall.

[0073] In the embodiment of the present disclosure, the sampling point (Lidar) is located on an open flat ground 50 m away from the Shenzhen Meteorological Gradient Tower. A WindMast PBL boundary layer type device is set at the sampling location, and the reference sampling point is the Shenzhen Meteorological Gradient Tower.

[0074] Specifically, perform the following steps:

[0075] Using lidar at sampling points, we acquire raw spectral data of a rainfall event in the target area. This raw spectral data is then preprocessed to remove background noise. First, we describe the lidar output data, which includes Level 0, Level 1, and Level 2. Level 0 includes raw spectral data, radial wind speed data, spectral width, spectral intensity, and signal-to-noise ratio. Level 1 includes instantaneous wind speed data, wind direction data, signal-to-noise ratio data, and vertical airflow. Level 2 includes the time-averaged wind profile, vertical airflow, turbulence intensity, and wind shear.

[0076] The raw spectral data records the backscattered signal of laser light from atmospheric aerosol particles and serves as the underlying data for the subsequent identification and separation of rainfall and aerosol signals. The spectral data is derived directly from lidar measurements and is stored in CSV format in the Level 0 directory. The first 300 points represent background noise, with each 100 points representing a background noise bin. Starting at point 601, each 100 points represents a range bin. The range resolution is 15 meters, and points 601 to 700 correspond to a radial distance of 45 meters.

[0077] In the embodiment of the present disclosure, preprocessing is to process the noise in the original spectrum data to Figure 1 As shown, in Figure 1 In the figure, (a), (b) and (c) are the background noises of the three gates respectively, (d) is the average background noise of the three gates, (e) is the spectrum including the background noise, and (f) is the spectrum after removing the average background noise (d) from (e). Figure 1 The power spectra provided in the figures other than are all after background noise is removed.

[0078] Afterwards Figure 2 As shown, corresponding preprocessing is performed according to the type of power spectrum signal, where:

[0079] If the power spectrum signal structure is a "double-peak structure" power spectrum, the aerosol Doppler frequency f is obtained by directly fitting the double Gaussian model. a and the raindrop Doppler frequency f r , the separation results are as follows Figure 3 shown.

[0080] If the power spectrum signal structure is a "single peak structure" power spectrum, the spectrum width and skewness are used to identify whether the power spectrum is a sunny spectrum or a rainy spectrum, such as Figure 4 As shown in the figure, according to the rainfall information recorded in the meteorological sensor carried by the lidar, under rainfall conditions, the spectral width of the original echo power spectrum captured by the lidar is significantly widened compared with that under clear sky conditions.

[0081] Under rainfall conditions, due to the different falling speeds of aerosols and raindrops, the spectrum width of the rainfall spectrum will increase compared to the clear sky spectrum, indicating that there are multiple components in the power spectrum (S a 、S r , I r , I a , I r :I a ). In the embodiment of the present disclosure, the type of power spectrum can be distinguished more accurately by combining the spectrum width and skewness, such as Figure 5 As shown in the figure, under rainfall conditions, the original power spectrum measured by the lidar is compared with the clear sky power spectrum. The skewness range of the rainfall power spectrum is significantly greater than the skewness S of the clear sky power spectrum. k0 .

[0082] The calculation formula for skewness is:

[0083]

[0084] Where S k ——skewness; v i ——Velocity spectrum value at sampling point i; ——average speed; P i ——power spectrum value at sampling point i; S k0 ——Pre-skewness, that is, the skewness value of the power spectrum under clear weather;

[0085] like Figure 6 As shown in the figure, there is a positive correlation between the spectral width and spectral intensity in the power spectrum and the rainfall intensity.

[0086] For the power spectrum with a peak structure type of "double peak structure", it is judged to be a rainfall spectrum through judgment (two obvious signals, the left one is aerosol and the right one is raindrop signal).

[0087] For the power spectrum with a peak structure type of "single peak structure", since the raindrop signal and the aerosol signal are fused in the peak structure, as shown in the embodiment of the present disclosure, Figure 4 and Figure 5 As described above, in the embodiment of the present disclosure, whether it is a rainfall spectrum is determined by skewness and spectral width. Specifically, in the embodiment of the present disclosure, thresholds for skewness and spectral width are set by statistics.

[0088] The statistical method provided by the embodiment of the present disclosure is specifically to continuously sample the relatively abundant rainfall in Guangzhou during the summer at the set sampling location. In the embodiment of the present disclosure, the sampling time is from June to December 2023. The sampling is performed according to the preset time period to obtain all the original power spectrum data within the preset time period, calculate the spectrum width and spectrum intensity, and synchronously record the rainfall information through the meteorological sensor. The original power spectrum data at the time corresponding to the rainfall information is recorded. Figure 7 The thresholds of spectrum width and skewness are shown. According to statistics, when the spectrum width is greater than 8 or the skewness is greater than 0.2 or the skewness is less than -0.1, the power spectrum of the obtained single-peak structure is a rainfall spectrum.

[0089] For the rainfall spectrum determined by the power spectrum of the "single peak structure", it is assumed that the instantaneous speed of raindrops falling at each moment in the same height is consistent, that is, the radial wind speed V of the raindrops is losN 、V losE 、V losS and V losW Considered to be the same value, the f at these four moments r It is considered unchanged, and then the double-Gaussian model is fitted to the "double-peak structure" power spectrum of the four moments to obtain f r , by fixing the value f r Fitting unimodal rainfall spectrum to reduce overfitting.

[0090] In the embodiment of the present disclosure, a double Gaussian model is used for fitting as follows:

[0091]

[0092] Where S a (f), S r (f)——aerosol and raindrop signals respectively; I a , I r ——are the signal strengths of aerosols and raindrops respectively; f a 、f r ——the Doppler frequencies of aerosols and raindrops respectively; σ a , σ r ——spectral width of aerosol and raindrop respectively; f——Doppler frequency.

[0093] Then, the separated aerosol Doppler frequency f aThe radial wind speed is calculated by using the DBS four-beam wind field inversion algorithm to calculate the three-dimensional wind field information. Specifically, the calculation method is as follows:

[0094] The calculation formula of the radial wind speed is as follows:

[0095]

[0096] Among them, λ is the central wavelength of the laser emitted by the lidar, f D is the average Doppler shift of the power spectrum, which is obtained by subtracting the zero frequency point 49.26 from the frequency f at the maximum power spectrum intensity during lidar inversion. los is the radial wind speed. The radial wind speed can be calculated into a three-dimensional wind field using the DBS four-beam wind field inversion algorithm. The lidar mistakenly identifies the stronger raindrop signal in the power spectrum as an aerosol signal, and the inversion result is the raindrop falling speed rather than the wind speed. Therefore, the above f is the separated aerosol frequency f a The specific calculation formula of the inversion algorithm is:

[0097]

[0098] Where V losN 、V losE 、V losS 、V losW ——are radial wind speeds in the north, east, south and west directions respectively; u, v, w——are the horizontal component from south to north, the horizontal component from west to east and the vertical component respectively; θ——elevation angle, that is, the angle between the laser beam and the horizontal plane; λ——wavelength of the emitted laser.

[0099] Use the vector method to synthesize horizontal wind speed and direction:

[0100]

[0101] Where, — horizontal wind speed; ——Horizontal wind direction.

[0102] In order to verify whether the inverted wind field parameters after removing the rainfall signal by this method are more reasonable, the embodiment of the present disclosure conducted a simultaneous and co-located comparative experiment between the lidar and the Shenzhen Meteorological Gradient Tower. During this period, two rainfall data were recorded, at 13:00-15:00 on July 6, 2024 and 11:00-14:00 on August 6. Figure 8 This is the temperature (T), humidity (RH), and rainfall (Rainfall) data recorded by the lidar meteorological module. As can be seen from the figure, the temperature and humidity values ​​experienced a sharp drop and a sharp rise, respectively, at the time of rainfall. It's important to note that current downburst research incorporates parameters such as temperature and humidity to determine whether a downburst event is a true one.

[0103] For the two rainfall events, the horizontal, vertical wind speed and horizontal wind direction recorded by the gradient tower three-dimensional anemometer (CSAT3) and laser radar (Lidar) during the period are plotted as follows: Figure 9 As shown. Combined Figure 8 and Figure 9 Meteorological and wind field data show that the vertical wind speed recorded by the lidar experienced a sudden drop during both rainfall events, while the vertical wind speed recorded by the 3D anemometer showed no sudden drop. Clearly, the velocity measured by the lidar was the speed of falling raindrops, not wind speed. Furthermore, on July 6th, the horizontal wind speed recorded by the lidar and the 3D anemometer did not deviate significantly, with the lidar's horizontal wind speed being slightly higher. On August 6th, the horizontal wind speed recorded by the 3D anemometer was relatively stable, while the lidar's horizontal wind speed experienced several sudden increases. Regarding horizontal wind direction, the two recorded horizontal wind directions were roughly consistent during the two rainfall events.

[0104] For the above reasons, it is believed that the wind speed measured by the lidar is disturbed by rainfall. Figure 2 The original spectrum data of the lidar during the two rainfall periods were corrected by the process. Figure 10 The comparison of wind speed and direction data before and after the laser radar correction and the three-dimensional anemometer is shown. Figure 10 As can be seen in the figure, the lidar was affected by rainfall during the rainy period. Rain signals dominated parts of the power spectrum, and the lidar recorded the velocity of raindrops rather than aerosol particles. This resulted in a series of anomalous wind speeds, with sudden increases and decreases in the horizontal and vertical wind speeds measured by the lidar. By separating the wind and rain signals from the raw lidar spectrum data and removing the raindrop signals, the inverted wind speed series eliminated the effects of rainfall and showed no "sudden" abnormal wind changes. Furthermore, compared to the raw lidar wind field data, the wind speed and direction data corrected by wind and rain separation were consistent with those recorded by the 3D anemometer. In particular, the "sudden drop" in vertical wind speed, after correction, was closer to the gradient tower data.

[0105] The Von Karman spectrum is used to fit the specific frequency band of the measured wind spectrum, namely the low-frequency (<0.01 Hz) region, such as Figure 11 As shown, to determine the turbulence integral scale (L u ) and the standard deviation of fluctuating wind speed (σ u ) and then through the parameter L u and σ u The Von Karman spectrum is quantified, and the expressions of the Von Karman spectrum in the downwind direction, the crosswind direction and the vertical direction are:

[0106]

[0107] Where, n is the pulsation frequency; Si (n) —power spectrum density of fluctuating wind speed in direction i; ——Variance of the fluctuating wind speed in direction i; f——Mourning coordinate.

[0108] Turbulence integral scale (L i ) and the correction value of the pulsation variance The gust factor is obtained by fitting the Von Karman spectrum to the measured wind spectrum / the measured wind spectrum in a specific frequency band, and then corrected by the following formula. According to the definition of GF:

[0109]

[0110] Where τ is the duration of gust, which is 3 seconds; T is the duration of average speed, which is 10 minutes; g(T,τ,z) is the peak factor, which can be expressed as:

[0111]

[0112] Where v is the horizontal penetration rate of the average displacement per unit time, which can be obtained by the following formula:

[0113]

[0114] Where S x1 — A reasonable representation of the wind spectrum, in contrast to the wind spectrum measured by lidar, which is subject to certain distortion effects.

[0115] The corrected peak factor and gust factor are calculated by the formula.

[0116] Figure 11 The comparison of longitudinal pulsation wind spectra collected synchronously by lidar and 3D ultrasonic anemometer (CSAT3) on Shenzhen Meteorological Gradient Tower is shown. u ) and the standard deviation of fluctuating wind speed (σ u ) is calculated from the measured data of a three-dimensional ultrasonic anemometer (CSAT3).

[0117] The analysis found that the Von Karman spectrum and the CSAT3 measured spectrum are highly consistent in most frequency regions, although in some high-frequency regions (marked in yellow), the measured spectrum is slightly higher than the predicted spectrum. At the same time, the lidar spectrum differs significantly from the CSAT3 spectrum in the frequency range above 0.01 Hz, but the two spectra are basically consistent in the low-frequency region (<0.01 Hz). Similar trends were observed in observations at other times and different altitudes, which further confirms that Figure 11 The reliability of the observation results in the figure. and The four regions are the distortion of lidar wind turbulence, region The reason for this is that the sampling frequency of the lidar is not high enough, and the area It is related to the special structure of typhoons, and the regional and Related to the working principle / working mode of lidar.

[0118] Table 1 shows a comparison of the Lidar parameter values ​​before and after correction with the CSAT3 measurements. The results show that the Lidar parameter values ​​after correction are highly consistent with the CSAT3 measurements, verifying the effectiveness and reliability of this correction method.

[0119]

[0120] Table 1

[0121] The time-varying mean wind speed is extracted using the discrete wavelet transform (DWT) method, using the highly accurate db10 wavelet basis. The wind speed time histories in the along- and across-wind directions are decomposed into n layers, with the maximum number of layers n being log2(N), where N is the length of the wind speed time histories. Figure 12 and Figure 13 They are the wavelet decomposition diagrams of the crosswind wind speed time history of the lidar and the three-dimensional anemometer at the maximum decomposition layer n; Figure 14 and Figure 15 These are the wavelet decomposition diagrams of the downwind wind speed time history at the maximum decomposition layer n for the lidar and three-dimensional anemometer, respectively.

[0122] After decomposing the wind speed signal to the maximum number of layers, the wavelet energy is calculated for each decomposition layer. The layer at which the wavelet energy undergoes a sudden change is considered the most suitable decomposition layer and is used as the time-varying trend term. The wavelet energy of each layer is calculated for the along- and across-wind wind speed time history, resulting in the energy sudden change layer. The fluctuating along- and across-wind wind speeds are then calculated. The wind speed calculation process obtained by wavelet decomposition of CSAT3 in this disclosure is a relatively mature technical approach in the field and will not be further elaborated here.

[0123] Figure 16 The energy distribution of each layer of the maximum decomposition layer and the energy mutation layer (optimal decomposition layer) based on the DWT method are shown. Figure 16 It can be seen that for the fluctuating wind speed time history in the downwind direction, the maximum decomposition layer of the lidar wavelet decomposition is 12 layers, and the energy mutation layer is at the 9th layer. Therefore, Figure 16 Level 9 is the time-varying average wind, and level 1 to level 9 are the high-frequency parts. The maximum decomposition layer of the three-dimensional anemometer wavelet decomposition is 16 layers, and the energy mutation layer is at the 13th layer. Therefore, Figure 16Level 13 in the figure is the time-varying average wind, and levels 1 to 13 are the high-frequency parts. Based on the above analysis, the optimal time-varying average wind extracted by the lidar and 3D anemometer for the crosswind direction is as follows: Figure 17 As shown, the fluctuating wind speed in the downwind direction can be calculated according to the above formula, as shown in Figure 18 shown.

[0124] Figure 19 The energy distribution of each layer of the maximum decomposition layer and the energy mutation layer (optimal decomposition layer) based on the DWT method are shown for the crosswind wind speed time history of the lidar and the three-dimensional anemometer. Figure 19 It can be seen that for the fluctuating wind speed time history in the crosswind direction, the maximum decomposition layer of the lidar wavelet decomposition is 12 layers, and the energy mutation layer is at the 9th layer. Therefore, Figure 19 Level 9 is the time-varying average wind, and level 1 to level 9 are the high-frequency parts. The maximum decomposition layer of the three-dimensional anemometer wavelet decomposition is 16 layers, and the energy mutation layer is at the 12th layer. Therefore, Figure 19 Level 12 is the time-varying average wind, and levels 1 to 12 are the high-frequency parts. Based on the above analysis, the optimal time-varying average wind extracted by the lidar and 3D anemometer for the crosswind direction is as follows: Figure 20 As shown, the fluctuating wind speed in the crosswind direction can be calculated according to the above formula, as shown in Figure 21 shown.

[0125] Correct the downwind pulsating wind characteristic parameters of the Lidar as follows: Figure 22 The "Yulear-Lidar" and "Yulear-SZGT" curves in the figure represent the power spectra estimated using the Yule-Walker method for the downwind fluctuating wind speeds of the lidar and three-dimensional anemometer, respectively; the "Von Karman" curve in the figure represents the fitting curve obtained by Von Karman spectrum fitting of the wind spectrum measured by the three-dimensional anemometer "Yulear-SZGT"; and the "Fit Curve" curve in the figure represents the fitting curve obtained by Von Karman spectrum fitting of the low-frequency band (<0.01Hz) of the wind spectrum measured by the lidar "Yulear-Lidar".

[0126] Table 2 shows the correction results for the LiDAR downwind fluctuating wind characteristic parameters. The results show that the corrected LiDAR parameter values ​​are highly consistent with the CSAT3 measurement results, verifying the effectiveness and reliability of the correction method provided by the embodiments of the present disclosure.

[0127]

[0128] Table 2

[0129] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these effects are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods for each specific application to implement the described effects, but such implementations should not be considered to exceed the scope of the embodiments of the present disclosure. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0130] The flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical functions. In some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by dedicated hardware-based devices that perform the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A method for correcting errors in lidar wind field measurements during rainfall, characterized in that: The steps include: S1. Obtain the original spectrum data of the lidar during rainfall and remove its background noise; S2. determining the structure type of the power spectrum in the original spectrum data; If it is a double-peak structure, the aerosol Doppler frequency and raindrop Doppler frequency are obtained by fitting the double-Gaussian model; If it is a single-peak structure, determine whether it is a rainfall spectrum based on the preset spectral width and skewness parameters. If it is a rainfall spectrum, separate the aerosol Doppler frequency and raindrop Doppler frequency using a double Gaussian model. S3, calculate the radial wind speed by the aerosol Püller frequency, and obtain the three-dimensional wind field information by inversion using the DBS four-beam wind field inversion algorithm; S4, executing S2-S4 on the raw spectrum data of the same laser radar in multiple different rainfall periods; S5. Determine the wind field type during different rainfall periods; If the wind is steady and pulsating, the turbulence parameters are corrected by fitting the low-frequency band of the measured wind spectrum through the Von Karman spectrum; If the wind is non-stationary and fluctuating, the time-varying average wind speed is extracted by discrete wavelet transform, and the turbulence parameters are corrected based on the non-stationary model.

2. The method for correcting errors in laser radar wind field measurements during rainfall according to claim 1, characterized in that: In S3, the calculation formula of the radial wind speed is: Where λ is the laser wavelength, f D It is the difference between the frequency at the maximum power spectrum intensity and the zero frequency point 49.26; The DBS four-beam wind field inversion algorithm calculates the three-dimensional wind speed components using the following formula: Among them, V losN 、V losE 、V losS 、V losW are the radial wind speeds in four directions respectively; θ is the elevation angle of the laser beam.

3. The method for correcting laser radar wind field measurement errors during rainfall according to claim 1, characterized in that: The expression of the double Gaussian model is: Among them, I a , I r are the signal strengths of aerosols and raindrops, respectively; σ a , σ r are the spectral widths of the two respectively.

4. The method for correcting laser radar wind field measurement errors during rainfall according to claim 1, characterized in that: In the non-stationary wind speed model, the turbulence intensity is defined as: The gust factor is calculated using the following formula: Among them, t g is the duration of the gust.

5. The method for correcting errors in laser radar wind field measurements during rainfall according to claim 1, characterized in that: The preset conditions for identifying the unimodal structured rainfall spectrum are: Spectral width greater than 8 or skewness greater than 0.2 or less than -0.1; When fitting the unimodal rainfall spectrum, the Doppler frequency fr of raindrops at adjacent moments at the same height is fixed to reduce overfitting.

6. The method for correcting errors in laser radar wind field measurements during rainfall according to claim 1, characterized in that: The discrete wavelet transform adopts the db10 wavelet basis, and the maximum decomposition layer number n is log2(N), where N is the length of the wind speed time history. The optimal decomposition layer is determined by wavelet energy mutation to extract the time-varying average wind speed.

7. The method for correcting errors in laser radar wind field measurements during rainfall according to claim 1, characterized in that: The S5 also includes the following steps: Obtain the preset number of rainfall times recorded by the lidar and determine the abnormal sudden changes in horizontal and vertical wind speeds; The data of the three-dimensional ultrasonic anemometer is obtained, and the wind speed, wind direction and turbulence parameters before and after correction are obtained after removing the interference of the wake effect.

8. The method for correcting errors in laser radar wind field measurements during rainfall according to claim 1, characterized in that The output data of the lidar includes Level 0 raw spectrum data, Level 1 instantaneous wind speed data and Level 2 average wind profile data; wherein, the raw spectrum data is stored in CSV format, the first 300 points are background noise, each distance library corresponds to 100 points, and the distance resolution is 15m.

Images