Multilayer laser wind finding radar data processing method and system

Abstract

The invention provides a multilayer laser wind finding radar data processing method and system, and relates to the technical field of radar data processing. The method comprises the following steps: acquiring echo signals of each height layer, synchronously acquiring meteorological parameters of visibility, humidity and rainfall intensity, judging weather types according to meteorological conditions, and setting corresponding SNR reference thresholds; dividing a near-surface layer, a middle layer and a high-altitude layer based on the SNR statistical characteristics of the height layer, and implementing threshold difference adjustment; performing threshold dynamic smoothing according to the meteorological change rate; judging the data validity based on the smoothed threshold sequence, and performing graded repair on discrete missing points, continuous missing segments and large-range missing regions; and finally, calculating the confidence coefficient through the SNR threshold value factor and the repair algorithm factor, and outputting a wind field profile with a reliability identifier. And the data utilization rate, the restoration precision and the result credibility under the complex meteorological condition are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of radar data processing technology, specifically to a multi-layer laser wind radar data processing method and system. Background Technology

[0002] Laser wind radar, an instrument that obtains wind field information by measuring the Doppler shift of laser light scattered by aerosol particles in the atmosphere, has been widely used in meteorological observation, wind energy assessment, aviation safety, and environmental monitoring. Currently, this field generally uses a fixed signal-to-noise ratio threshold to determine the validity of data, and supplements it with simple interpolation or extrapolation strategies to repair missing data in order to generate wind field profiles.

[0003] However, the signal-to-noise ratio (SNR) threshold is unstable under different weather conditions, leading to significant data loss. Furthermore, data restoration methods for complex weather conditions are limited and lack fine-grained confidence estimation. Specifically, these issues significantly reduce the reliability of wind field data, failing to provide reliable references for data users, impacting subsequent meteorological analysis and applications, and severely restricting the effectiveness of laser wind radar under complex weather conditions, particularly in terms of data utilization and accuracy in low SNR environments. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides a multi-layer laser wind radar data processing method and system.

[0005] On one hand, the present invention provides a multi-layer laser wind measurement radar data processing system, comprising the following components connected in sequence via communication: A signal acquisition and preprocessing unit for collecting radar echoes and calculating multi-layer SNR, and a meteorological parameter acquisition module for synchronously acquiring visibility, humidity and precipitation intensity; The meteorological parameter acquisition module is connected to the adaptive threshold regulator, which is used to determine the weather type based on the meteorological parameters and generate the corresponding SNR benchmark threshold. The adaptive threshold adjuster is used to divide the near-ground layer, intermediate layer and high-altitude layer according to the SNR distribution characteristics of each altitude layer, and to make differentiated adjustments to the SNR threshold of each layer in order to achieve multi-layer adaptive threshold optimization. The data validity judge is used to select an exponential weighted moving average or Kalman filter strategy based on the rate of weather change, and to perform time smoothing on the adjusted SNR threshold to form a continuous dynamic threshold. The data validity judge is used to filter invalid or missing data based on the dynamic threshold, and automatically select the repair algorithm according to its distribution pattern. It performs cubic spline interpolation on discrete missing points, polynomial interpolation with wind speed gradient constraints on continuous missing segments, and boundary layer model reconstruction based on external wind tower or radiosonde data on large-scale missing areas. The data repair estimator is used to calculate the confidence factor of the original data based on the difference between SNR and threshold, and to generate a comprehensive confidence value by combining the repair algorithm factor. The output multi-layer wind field profile is then output with a confidence label attached.

[0006] On the other hand, the present invention provides a multi-layer laser wind radar data processing method, including: S1. Collect echo signals from the laser wind radar at different altitudes and calculate the signal-to-noise ratio (SNR). At the same time, acquire meteorological observation parameters corresponding to the current detection time, including visibility, humidity, and precipitation intensity. S2. Determine the current weather type based on meteorological observation parameters, and select the corresponding SNR benchmark threshold according to the preset weather classification standard to reflect the signal attenuation characteristics under different weather conditions. S3. Based on the statistical characteristics of SNR distribution at each altitude level, the detection space is divided into three altitude levels: near-ground layer, intermediate layer and upper-level layer. The SNR threshold of each level is adjusted accordingly to compensate for the differences between near-ground clutter interference and upper-level signal attenuation, thereby obtaining an adaptive SNR threshold that is suitable for the multi-layer atmospheric environment. S4. Determine meteorological stability based on the rate of weather change. When the weather changes slowly, apply an exponentially weighted moving average to the adaptive SNR threshold for strong smoothing. When the weather changes drastically, apply weak smoothing to obtain a threshold sequence that is time-continuous and dynamically responsive. S5. Based on the smoothed threshold sequence, the original data is evaluated for validity. Data that is deemed invalid or has low signal-to-noise ratio is analyzed for its spatiotemporal characteristics. Missing data is divided into three categories according to its distribution pattern: discrete missing points, continuous missing segments, and large-scale missing areas. The following repair strategies are then implemented for each category: cubic spline interpolation is used for discrete missing points, polynomial interpolation with wind speed gradient constraints is used for continuous missing segments, and external wind tower or radiosonde data is introduced into large-scale missing areas to reconstruct the data based on the boundary layer model, thereby achieving hierarchical repair. S6. Calculate the confidence score for the repaired data. For the original data, calculate the SNR threshold factor based on the difference between SNR and threshold. For the repaired data, calculate the repair algorithm factor based on the goodness of fit and interpolation distance. Combine the two types of factors according to the weighting rules to generate the final confidence score. Add a confidence score label to the output multi-layer wind field profile to complete the wind field data output.

[0007] As a further technical solution, the specific method for determining the weather type and selecting the SNR benchmark threshold based on meteorological observation parameters in step S2 is as follows: Visibility, humidity, and precipitation intensity data acquired in real time are compared with preset weather classification standards, and differentiated SNR benchmark thresholds are set according to different weather conditions. Specifically, under clear weather conditions with visibility greater than 10km, humidity less than 60%, and no precipitation, the SNR benchmark threshold is set to -8dB; under rainy weather conditions with precipitation intensity greater than 0mm / h, the SNR benchmark threshold is set to -12dB; under foggy weather conditions with visibility less than 1km and humidity less than 60%, the SNR benchmark threshold is set to -12dB. When the signal density is above 90%, the SNR baseline threshold is set to -15dB. This setting is based on the physical laws of signal attenuation under different weather conditions. Foggy days have the highest aerosol concentration, resulting in the most severe signal attenuation, so the lowest threshold is set to retain more effective data. Rainy days have strong precipitation particle scattering, so a medium threshold is set to balance data quantity and quality. Sunny days have the best atmospheric transmittance, so a higher threshold is set to suppress noise interference. Through the joint determination of multiple parameters and the setting of differentiated thresholds, accurate adaptation to signal characteristics under different weather conditions is achieved, avoiding data waste or noise residue caused by a single fixed threshold.

[0008] As a further technical solution, the SNR threshold adaptive adjustment process based on height layering in step S3 includes: firstly, analyzing the statistical distribution characteristics of SNR at each height layer, including mean, standard deviation, and quantile indices, and obtaining typical SNR distribution patterns for each height layer through training with historical data; then, dividing the detection space into three typical height layers based on the analysis results, whereby the near-ground layer is defined as the 0-500 meter height range, and due to the influence of ground clutter, building reflections, and multipath interference, the SNR threshold needs to be increased by 2dB to suppress false signals; the intermediate layer is defined as the 500-2000 meter height range, which is a relatively stable signal area, and the baseline threshold remains unchanged; the high-altitude layer is defined as the height range above 2000 meters, where the SNR threshold is reduced by 3dB to avoid the loss of effective data because the signal attenuates significantly with increasing transmission distance; this layering optimization process dynamically adjusts the threshold by statistically analyzing the distribution characteristics of SNR at each layer, and performs targeted processing of the signal characteristics at each height layer, thereby improving the integrity and reliability of the vertical profile data.

[0009] As a further technical solution, the adaptive smoothing processing based on the meteorological change rate includes: firstly, quantifying the meteorological change rate by calculating the change gradients of key meteorological parameters, including visibility, humidity, and precipitation intensity, within a continuous time window; specifically, calculating the slope of the parameters on the time series through difference operations; when the slope is lower than a preset threshold, it is determined that the weather change is slow; when the slope is higher than the preset threshold, it is determined that the weather change is drastic; when the weather change is slow, a strong smoothing process is performed using an exponentially weighted moving average method, the calculation process of which is expressed as follows: Where α is the smoothing coefficient, ranging from 0.1 to 0.3, the stability of the threshold is ensured by increasing the weight of historical data; when the weather changes drastically, Kalman filtering is used for weak smoothing, and its state equation is expressed as follows: The observation equation is expressed as ,in For process noise, set to 0.5. To measure the noise level, we set it to 1.0. By adjusting the noise parameters, we can enhance the threshold's ability to respond to rapid changes, ensuring that the threshold remains continuous and adaptable over time, and avoiding data gaps caused by sudden weather changes.

[0010] As a further technical solution, the cubic spline interpolation repair process for discrete missing points includes: first, locating the position of the missing point in the time dimension; then, extracting multiple valid data points adjacent to it as interpolation nodes, typically selecting two nodes before and after, for a total of four data points; and using cubic spline basis functions. The coordinates represent the time series, and a, b, c, and d are spline coefficients. The coefficients are solved by boundary continuity conditions and natural spline conditions. Matrix operations are used to ensure that the interpolation function has continuous first and second derivatives at the nodes, thereby ensuring the smoothness and physical rationality of the repaired wind speed data in the time dimension.

[0011] As a further technical solution, the physical constraint polynomial interpolation repair process for continuous missing segments includes: firstly, selecting multiple valid data points from both ends of the missing segment, typically four points from each end, and calculating the mean boundary wind speed. and wind speed slope As boundary conditions; then, the order of the polynomial is dynamically determined according to the length of the missing segment. When there are 5 to 10 missing data points, a third-order polynomial is used, and when there are 11 to 20 missing data points, a fourth-order polynomial is used to ensure that the goodness of fit meets the preset requirements; finally, the wind speed gradient is constrained using the Lagrange multiplier method. The process of solving embedded polynomial coefficients involves constructing a constrained optimization problem to solve the polynomial coefficients, ensuring that the rate of change of wind speed within the missing segment is consistent with the boundary trend, and maintaining the physical consistency of the wind field.

[0012] As a further technical solution, the boundary layer model reconstruction process for the large-scale missing area in step S5 includes: firstly, selecting representative data points from the remaining valid data as anchor points and recording their wind speed values. and corresponding height z i Then, the Power-Law model was fitted using the least squares method. Parameters are adjusted through iterative optimization algorithms. , and To minimize the error at the anchor point, an error threshold of 0.8 m / s was set. Simultaneously, external reference data from the wind turbine was introduced, based on the Log-Law model from boundary layer similarity theory. Inverse friction speed The roughness length z0 is determined by k, a Karman constant with a value of 0.4. Finally, the parameters obtained from the inversion are used as physical constraints. The reconstruction results of the boundary layer model are optimized by weighted fusion to ensure that the reconstructed data and the wind field distribution in the boundary layer are consistent with the external reference data.

[0013] As a further technical solution, the confidence assessment system in step S6 includes two independently calculated quality factors: For the original valid data based on adaptive SNR threshold determination, an SNR threshold factor is calculated. This factor is calculated by linearly mapping the difference between SNR and the threshold to a value in the range of 0.3 to 1.0. Specifically, when SNR exceeds the threshold by 10 dB, the confidence reaches its maximum value of 1.0; when SNR equals the threshold, the confidence is a baseline value of 0.6; and when SNR is below the threshold, the confidence decreases proportionally but is never lower than 0.3. For the data after repair... Based on the data, a repair algorithm factor is calculated. This factor is jointly determined by the goodness-of-fit R² and the normalized interpolation distance d. R² reflects the degree of fit between the repair model and the actual data, while d reflects the proximity of the missing location to the valid data. The two factors are combined with equal weights to generate a value in the range of 0.2 to 1.0. Finally, a confidence score in the range of 0 to 1 is generated by weighted combination of these two factors. The weighting rule is to add them with equal weights and take the average. Boundary constraints are applied to the calculation results to ensure the rationality and interpretability of the output confidence score, providing a basis for reliability classification for data users.

[0014] This invention provides a multi-layer laser wind radar data processing method and system, which has the following beneficial effects: 1. This invention solves the core problem that fixed thresholds cannot accurately distinguish between effective signals and noise under complex and variable weather conditions, resulting in a large amount of data loss or noise residue, by introducing an adaptive SNR threshold mechanism based on weather type and altitude. This significantly improves the data acquisition capability and utilization rate of laser wind radar under different weather conditions and altitude levels.

[0015] 2. Based on the above-mentioned adaptive threshold screening, this invention adopts a hierarchical and physically constrained repair strategy for different missing data distribution patterns, which solves the problem of poor physical consistency and low reliability of repair results when the single interpolation method is severely missing data. Thus, while improving data integrity, it ensures the accuracy and physical rationality of the repaired wind field data.

[0016] 3. This invention solves the problem of only providing binary judgments and not being able to quantify data reliability by constructing a confidence assessment system that integrates the quality of the original data and the reliability of the repaired data. Ultimately, it assigns a quantifiable reliability label to each data point in the final output multi-layer wind field profile, which greatly improves the practical value and credibility of the output data in subsequent meteorological analysis and decision-making applications. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 Flowchart for adaptive SNR threshold adjustment; Figure 3 Flowchart of the data repair estimation algorithm; Figure 4 A comparison chart of the system's processing performance under different weather conditions; Figure 5 This is a schematic diagram of the confidence assessment model; Figure 6 The figure shows the experimental verification results. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The following detailed description, with reference to the accompanying drawings, illustrates a specific implementation method and system for multi-layer laser wind radar data processing according to the present invention.

[0020] refer to Figure 1 This demonstrates the overall architecture of the system of the present invention; the system includes a signal acquisition and preprocessing unit, a meteorological parameter acquisition module, an adaptive threshold regulator, an adaptive threshold regulator, a data validity judge, a data validity judge, and a data repair estimator, which are connected in sequence; each module works together to form a complete data processing link; the execution of this embodiment will revolve around this architecture.

[0021] Suppose that at a meteorological observation station, a multi-layer laser wind radar is conducting routine wind field detection; the radar's detection altitude range is from 0 to 3000 meters, divided into 30 altitude layers; after the system starts operating, the signal acquisition and preprocessing unit first acquires the radar echo signals at each altitude layer and calculates the raw signal-to-noise ratio (SNR) data corresponding to each altitude layer; at the same time, the meteorological parameter acquisition module synchronously acquires the meteorological observation parameters at the current detection time from the supporting automatic weather station, including but not limited to: visibility of 8 kilometers, relative humidity of 75%, and precipitation intensity of 2 mm / hour.

[0022] refer to Figure 2 The system details the calculation and adjustment process of the adaptive SNR threshold. After acquiring the initial data, the adaptive threshold adjuster starts working. The system determines the current weather type based on the preset weather classification standards. According to these standards, a clear day is defined as visibility > 10km, humidity < 60%, and no precipitation; a rainy day is defined as precipitation intensity > 0mm / h; and a foggy day is defined as visibility < 1km and humidity > 90%. The current visibility is 8km, which is between 1km and 10km, humidity is 75% to less than 90%, and there is precipitation of 2mm / h. The system comprehensively determines the current weather type as "rainy day". Based on this, the system calls the SNR baseline threshold corresponding to rainy days, which is -12dB.

[0023] Next, the adaptive threshold adjuster is activated. It first analyzes data from a recent period, such as the statistical distribution characteristics of SNR at various altitudes over the past hour, including the mean, standard deviation, and quantiles. Based on these statistical characteristics, the adaptive threshold adjuster divides the detection space from 0 to 3000 meters into three typical altitude levels: the near-surface level (0-500 meters), the intermediate level (500-2000 meters), and the upper-air level (above 2000 meters). Subsequently, the SNR baseline threshold is adjusted accordingly: for the near-surface level, due to ground clutter and building interference, the threshold is increased by 2dB to -10dB to more strictly suppress false signals; for the intermediate level, where the signal is relatively stable, the baseline threshold of -12dB remains unchanged; for the upper-air level, because the signal attenuates significantly with increasing distance, to avoid valid data being misjudged as invalid, the threshold is decreased by 3dB to -15dB. Thus, the system obtains an adaptive SNR threshold adapted to the current multi-level atmospheric environment.

[0024] Then, the data validity checker begins its work; this checker quantifies the rate of weather change by calculating the gradients of key meteorological parameters within a continuous time window, such as visibility and precipitation intensity over five consecutive observation periods; assuming the currently calculated gradient is small, it is classified as a "slow weather change" scenario; a strong smoothing process is performed using an exponentially weighted moving average method; its calculation process is as follows: ;in, The unsmoothed adaptive threshold (i.e., the layered threshold) is calculated at time t based on the aforementioned steps. The threshold is the smoothed threshold at time t-1. The smoothing coefficient α is set to a small value in this scenario, such as 0.1, to give higher weight to historical data and ensure the stability of the threshold over time. After this smoothing process, a threshold sequence that is continuous in time and has a dynamic response is finally output.

[0025] refer to Figure 3 It explains the selection and execution process of the data repair algorithm; the system uses the smoothed dynamic threshold sequence as a basis to determine the validity of the original SNR data; any original data with an SNR value lower than the threshold of its corresponding height layer is judged as invalid or low signal-to-noise ratio data; assuming that in a single detection, the system identifies some invalid data and performs spatiotemporal feature analysis on it; based on the distribution pattern of invalid data in time and space dimensions, the system automatically classifies them and executes the corresponding hierarchical repair strategy.

[0026] For example, in the 1500-meter altitude layer, which belongs to the middle layer, the system identified several discrete missing points that were isolated in the time series, with valid data points in both preceding and following positions. For such discrete missing points, the data validity judge initiated a cubic spline interpolation repair strategy. Specifically, the location of the missing point was located in the time dimension, and then four valid data points in both preceding and following positions were extracted as interpolation nodes. Cubic spline basis functions were used. Curve fitting is performed, where x represents the time series coordinates, and a, b, c, and d are spline coefficients obtained through boundary continuity conditions and natural spline conditions. This process ensures that the interpolation function has continuous first and second derivatives at the nodes, thereby ensuring the smoothness and physical rationality of the repaired wind speed data in the time dimension.

[0027] Meanwhile, within the altitude range of 1800-2200 meters spanning the mesosphere and upper atmosphere, the system identified a continuous missing segment where multiple consecutive detection cycles were invalid. This segment is applicable when there are ≥8 valid data points and ≥3 consecutive invalid data points within the measurement cycle. For such continuous missing segments, physical constraint polynomial interpolation is used for repair. First, four valid data points are selected from each end of the time dimension of the missing segment (the beginning and the end), and the average boundary wind speed is calculated for each. and wind speed slope As boundary conditions; next, the order of the polynomial is dynamically determined based on the number of data points contained in the continuous missing segment, for example, if 8 points are missing; a third-order polynomial is used for missing 5-10 points; finally, the wind speed gradient is constrained using the Lagrange multiplier method. The process of solving embedded polynomial coefficients ensures that the rate of change of wind speed in the repaired missing segment is consistent with the boundary trend, thus avoiding drastic non-physical jumps.

[0028] Furthermore, in the upper atmosphere above 2500 meters, the system identified large-scale missing areas with very few valid data points. In these areas, a 70-meter-high wind tower is required near the observation station to provide concurrent external reference data. For such large-scale missing areas, a boundary layer model reconstruction strategy is implemented. First, representative data points are selected from the remaining valid data as anchor points, recording their wind speed value vi and corresponding height zi. Then, the least squares method is used to fit the Power-Law wind profile model. By optimizing the algorithm, the model parameters v0, z0, and To minimize the wind speed prediction error at these anchor points, for example, with an error ≤ 0.8 m / s; simultaneously, wind speed data v measured by a 70-meter wind tower during the same period is incorporated. raw Log-based wind profile model in boundary layer similarity theory Inverse friction speed And the roughness length z0, where k is the Karman constant, whose value is 0.4 as verified by experiments; finally, the friction velocity obtained by inversion is... Using the roughness length z0 as a physical constraint, the reconstruction results of the Power-Law model are optimized and corrected, thereby obtaining a more physically reliable and complete upper-level wind field profile.

[0029] refer to Figure 5 It explains the multi-factor comprehensive calculation model for confidence assessment; after all data, including the original valid data and the repaired data, are ready, the data repair estimator starts working to calculate a comprehensive confidence score in the range of 0 to 1 for each data point in the final output; the system contains two independently calculated quality factors.

[0030] For wind field data deemed valid in its original form, calculate its SNR threshold factor using the following formula: The 10 here is used as an empirical coefficient in the denominator, which is derived from historical data statistics and is implied by 0.04. This SNR threshold factor reflects the quality level of the original data relative to the threshold of its height layer. The more the SNR exceeds the threshold, the higher the SNR threshold factor value, with a maximum of 1. According to the boundary constraints, if the calculation result is greater than 1, then take 1; if it is less than 0.3, then take 0.3.

[0031] For the repaired wind field data, the repair algorithm factor is calculated using the following formula: Where R2 is the goodness of fit of the model used in the repair algorithm, which can be regarded as the degree of fit with the neighboring points for interpolation, and d is the normalized interpolation distance, which is used to reflect the spatiotemporal distance between the missing point and the nearest valid data point. This repair algorithm factor comprehensively evaluates the accuracy and reliability of the repair algorithm. Similarly, according to the boundary constraints, if the final result is less than 0.2, then 0.2 is taken.

[0032] Finally, the final confidence score for each data point is generated by weighted combination of these two factors. The combination rules are as follows: For data determined to be original and valid, the confidence score is directly composed of the SNR threshold factor, and the repair algorithm factor does not participate in the calculation and can be regarded as having a weight of 0; For data that has undergone repair processing, the confidence score is dominated by the repair algorithm factor, or the SNR threshold factor and the repair algorithm factor are weighted and fused according to preset rules; this ensures that the final output confidence score can accurately reflect the reliability differences of data from different sources.

[0033] refer to Figure 4 and Figure 6 These examples demonstrate the comparison of processing effects under different weather conditions and the experimental verification results with reference equipment. After the complete processing described above, the system outputs a final product containing multi-layered wind field information (wind speed and direction) and a confidence level label for each data point. For example, under rainy conditions in this embodiment, the data coverage of the final output is increased from approximately 40% in the traditional method to over 85%. Compared with reference equipment such as wind towers, the average error is reduced from 2.1 m / s in the traditional method to approximately 1.3 m / s. For the repaired data, the confidence level is mainly concentrated in the range of 0.5 to 0.8, while the confidence level of high-quality original data is above 0.8, providing data users with a clear and quantifiable reliability reference, thus completing the entire data processing process.

[0034] In summary, this embodiment details the complete application process of the method and system of the present invention in a typical weather scenario, including data acquisition, adaptive threshold setting and optimization, time smoothing, hierarchical data repair, and confidence assessment. It fully discloses the technical solution and ensures that those skilled in the art can reproduce the present invention through specific data and process descriptions.

[0035] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-layer laser wind radar data processing method, characterized in that, Includes the following steps: S1. Collect echo signals from the laser wind radar at different altitudes and calculate the signal-to-noise ratio (SNR). At the same time, acquire meteorological observation parameters corresponding to the current detection time, including visibility, humidity, and precipitation intensity. S2. Determine the current weather type based on meteorological observation parameters, and select the corresponding SNR benchmark threshold according to the preset weather classification standard to reflect the signal attenuation characteristics under different weather conditions. S3. Based on the statistical characteristics of SNR distribution at each altitude level, the detection space is divided into three altitude levels: near-ground layer, intermediate layer and upper-level layer. The SNR threshold of each level is adjusted accordingly to compensate for the differences between near-ground clutter interference and upper-level signal attenuation, thereby obtaining an adaptive SNR threshold that is suitable for the multi-layer atmospheric environment. S4. Determine meteorological stability based on the rate of weather change. When the weather changes slowly, apply an exponentially weighted moving average to the adaptive SNR threshold for strong smoothing. When the weather changes drastically, apply weak smoothing to obtain a threshold sequence that is time-continuous and dynamically responsive. S5. Based on the smoothed threshold sequence, the original data is evaluated for validity. Data that is deemed invalid or has low signal-to-noise ratio is analyzed for its spatiotemporal characteristics. Missing data is divided into three categories according to its distribution pattern: discrete missing points, continuous missing segments, and large-scale missing areas. The following repair strategies are then implemented for each category: cubic spline interpolation is used for discrete missing points, polynomial interpolation with wind speed gradient constraints is used for continuous missing segments, and external wind tower or radiosonde data is introduced into large-scale missing areas to reconstruct the data based on the boundary layer model, thereby achieving hierarchical repair. S6. Calculate the confidence score for the repaired data. For the original data, calculate the SNR threshold factor based on the difference between SNR and threshold. For the repaired data, calculate the repair algorithm factor based on the goodness of fit and interpolation distance. Combine the two types of factors according to the weighting rules to generate the final confidence score. Add a confidence score label to the output multi-layer wind field profile to complete the wind field data output.

2. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: In step S2, the visibility, humidity and precipitation intensity data acquired in real time are compared with the preset weather classification standards, and different SNR benchmark thresholds are set according to different weather conditions. Among them, the lowest SNR benchmark threshold is set because the aerosol concentration is the highest under foggy conditions, which leads to the most severe signal attenuation. Rainy weather conditions result in strong scattering of laser signals by falling water particles, so a medium-level SNR benchmark threshold is set. Under clear weather conditions, atmospheric transmittance is optimal, so a relatively high SNR benchmark threshold is set. This differentiated threshold setting method achieves precise adaptation of signal characteristics under different weather conditions.

3. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: In step S3, the statistical distribution characteristics of SNR at each altitude layer are first analyzed, including the mean, standard deviation, and quantile index. Then, based on the analysis results, the detection space is divided into three typical altitude layers. In the near-ground layer, due to interference from ground clutter and buildings, the SNR threshold is increased to suppress false signals. In the intermediate layer, as a relatively stable signal region, the baseline threshold is kept unchanged. In the high-altitude layer, since the signal attenuates significantly with increasing distance, the SNR threshold is reduced to avoid the loss of effective data. This achieves targeted processing of the signal characteristics at each altitude layer.

4. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: In step S4, the rate of meteorological change is quantified by calculating the gradient of key meteorological parameters within a continuous time window. When meteorological changes are slow, a strong smoothing process is performed using an exponentially weighted moving average method. The calculation process is expressed as follows: Where α is the smoothing coefficient, The original threshold value calculated at time t. The threshold value at time t-1 is smoothed, and its stability is ensured by increasing the weight of historical data. When weather changes drastically, weak smoothing is performed, and the state equation for weak smoothing is as follows: The observation equation is ,in For process noise, To observe noise, the threshold's responsiveness to rapid changes is enhanced by adjusting noise parameters, thus maintaining the threshold's continuity and adaptability over time.

5. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: In step S5, the cubic spline interpolation repair process for discrete missing points includes: first, locating the position of the missing point in the time dimension; then, extracting multiple adjacent valid data points as interpolation nodes; and finally, using cubic spline basis functions. Curve fitting is performed, where x represents the time series coordinates, and a, b, c, and d are spline coefficients obtained by solving the boundary continuity condition and the natural spline condition. By ensuring that the interpolation function has continuous first and second derivatives at the nodes, the smoothness and physical rationality of the repaired wind speed data in the time dimension are ensured.

6. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: In step S5, the physical constraint polynomial interpolation repair process for continuous missing segments includes: firstly, selecting multiple valid data points from both ends of the missing segment, and calculating the mean boundary wind speed. and wind speed slope As boundary conditions; then, the order of the polynomial is dynamically determined based on the length of the missing segment, using a lower-order polynomial when there are few missing data points and a higher-order polynomial when there are many missing data points; finally, the wind speed gradient is constrained using the Lagrange multiplier method. The process of solving the coefficients of the embedded polynomial.

7. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: The process of reconstructing the boundary layer model for a large area of ​​missing data includes: first, selecting representative data points from the remaining valid data as anchor points and recording their wind speed values. and corresponding height Then, the Power-Law model was fitted using the least squares method. The error at the anchor point is minimized by optimizing the parameters of the Power-Law model; simultaneously, external reference data from the wind turbine are introduced, based on the boundary layer similarity theory. Inverse friction speed The roughness length z0 is determined, where k is the KAMAN constant and z is the height coordinate. Finally, the parameters obtained from the inversion are used as physical constraints to optimize the reconstruction results of the boundary layer model.

8. The multi-layer laser wind radar data processing method according to claim 1, characterized in that: The confidence assessment system in step S6 includes two independently calculated quality factors: for the original valid data based on adaptive SNR threshold determination, the SNR threshold factor is calculated. The SNR threshold factor reflects the quality level of the original data relative to the threshold; for the repaired data, the repair algorithm factor is calculated. R² represents the goodness of fit, and d represents the normalized interpolation distance. This repair algorithm factor comprehensively evaluates the accuracy and reliability of the repair algorithm. Finally, a confidence score in the range of 0 to 1 is generated by weighted combination of these two factors.

9. A system for processing multi-layer laser wind radar data according to any one of claims 1-8, characterized in that, Including sequential communication connections: A signal acquisition and preprocessing unit for collecting radar echoes and calculating multi-layer SNR, and a meteorological parameter acquisition module for synchronously acquiring visibility, humidity and precipitation intensity; The meteorological parameter acquisition module is connected to the adaptive threshold regulator, which is used to determine the weather type based on the meteorological parameters and generate the corresponding SNR benchmark threshold. The adaptive threshold adjuster is used to divide the near-ground layer, intermediate layer and high-altitude layer according to the SNR distribution characteristics of each altitude layer, and to make differentiated adjustments to the SNR threshold of each layer in order to achieve multi-layer adaptive threshold optimization. The data validity checker is used to select either the exponentially weighted moving average or the Kalman filter strategy based on the rate of weather change, and to perform time smoothing on the adjusted SNR threshold to form a continuous dynamic threshold. The data validity judge is used to filter invalid or missing data based on the dynamic threshold, and automatically select the repair algorithm according to its distribution pattern. It performs cubic spline interpolation on discrete missing points, polynomial interpolation with wind speed gradient constraints on continuous missing segments, and boundary layer model reconstruction based on external wind tower or radiosonde data on large-scale missing areas. The data repair estimator is used to calculate the confidence factor of the original data based on the difference between SNR and threshold, and to generate a comprehensive confidence value by combining the repair algorithm factor. The output multi-layer wind field profile is then output with a confidence label attached.

Images