A method for turbulence measurement and correction based on bistatic radar

Abstract

The present application relates to the field of radar technology, specifically to a turbulence measurement and correction method based on bistatic radar, comprising: obtaining radial wind speed data of at least two radars, wherein the radial wind speed data includes first radar radial wind speed data and second radar radial wind speed data; reconstructing the second-level wind speed sequence of each height layer according to the first radar radial wind speed data to obtain the turbulence intensity to be corrected; reconstructing the wind speed field of a specific height layer according to the spatial intersection beam data of the first radar radial wind speed data and the second radar radial wind speed data to obtain the reference turbulence intensity; calculating the turbulence error according to the turbulence intensity to be corrected and the reference turbulence intensity, training the data-driven model according to the turbulence error to establish the mapping relationship between the turbulence intensity to be corrected and the turbulence error, and correcting the turbulence intensity to be corrected according to the mapping relationship. Thus, the problems of low wind measurement accuracy of single laser radar in complex terrain, high cost, difficult deployment and low efficiency of multi-radar cooperation in the prior art are solved.

Description

Technical Field

[0001] This invention relates to the field of radar technology, specifically to a method for turbulence measurement and correction based on bistatic radar. Background Technology

[0002] Wind resource assessment is the core foundation for wind farm planning, design, and investment return calculation. The accuracy of the assessment results directly determines the wind farm's turbine location layout, installed capacity configuration, and overall life-cycle investment return level, making it a crucial preliminary step for wind power project development and implementation. Accurate wind resource data, especially core data such as horizontal wind speed, three-dimensional wind vectors, and turbulence parameters, can effectively mitigate wind power project development risks and improve wind farm power generation efficiency and economic benefits. Therefore, high-precision, multi-dimensional, and high-efficiency wind measurement technology is a core support for the high-quality development of the wind power industry.

[0003] In traditional wind resource assessment systems, wind measurement towers are the mainstream wind measurement equipment, enabling long-term wind data monitoring at fixed locations and multiple height levels. They offer stable measurement accuracy and high data reliability, making them core equipment for wind resource surveying in complex terrains. However, traditional wind measurement towers have many shortcomings, such as large footprint, long construction period, high construction costs, limited installation sites, and high maintenance difficulty. Furthermore, for complex terrains such as mountains, hills, and coastal areas, the site selection for wind measurement towers is extremely difficult, making it difficult to achieve flexible and large-scale wind field data surveys. This makes them unsuitable for the current development needs of large-scale, refined, and comprehensive wind power projects.

[0004] Compared to traditional wind measurement towers, laser wind radar boasts significant advantages such as compact size, long detection range, convenient installation and deployment, no terrain limitations, and flexible mobile observation capabilities. In recent years, it has been widely used in early-stage wind energy resource assessment for wind power projects, gradually replacing traditional wind measurement towers as the mainstream wind measurement equipment. The working principle of laser wind radar is as follows: a directional detection beam is emitted into the air through the device's light source. The beam collides with aerosol particles in the atmosphere, generating backscattered signals. After receiving the echo signals, the radar calculates the air velocity along the beam's propagation direction, i.e., the radial wind speed, based on the Doppler effect.

[0005] In practical wind resource assessment projects, the core observation parameters are horizontal wind speed and three-dimensional wind vector. Radial wind speed obtained from a single beam cannot meet the survey requirements. According to the principle of wind vector calculation, a full-rank matrix needs to be constructed based on at least three sets of radial wind speeds from different directions to solve for the three components of wind speed and then synthesize an accurate three-dimensional wind vector. Current mainstream lidar wind measurement solutions all adopt a single-source multi-directional beam detection mode, that is, emitting detection beams from the same device location to multiple different directions, collecting radial wind speed data at different spatial locations, and completing the synthesis of three-dimensional wind vectors to achieve continuous monitoring and assessment of wind field wind resource parameters.

[0006] However, existing single-lidar wind measurement technology has inherent technical flaws. Its core reliance is on the assumption of wind field uniformity, which presupposes that wind vectors at the same height but different spatial locations within the measured wind field are completely identical. This is also the core premise for this scheme to synthesize a single-point wind vector from radial wind speeds at multiple locations. This assumption is largely valid in flat plains with uniform airflow distribution, and the measured data shows a high degree of matching with actual wind field conditions. However, in complex terrains such as mountains, hills, complex mountainous areas, and coastal mudflats, the wind field parameters at different locations at the same height differ significantly due to factors such as topographic relief, surface cover, and vertical airflow, rendering the wind field uniformity assumption completely ineffective.

[0007] Due to the aforementioned issues, the accuracy of traditional single-lidar wind measurement solutions for turbulence intensity measurement in complex terrains drops significantly. The measurement results deviate markedly from the actual data measured by high-precision wind towers, failing to meet the requirements for refined assessment of wind resources in complex terrains. This is the core reason why laser wind measurement radar cannot completely replace traditional wind towers at present, and why wind power projects in complex terrain still need to rely on wind towers for supplementary data measurement. This greatly limits the application scenarios and industry promotion value of laser wind measurement radar. Figure 1 As shown, traditional lidar turbulence measurement methods rely heavily on the assumption of wind field uniformity, making it difficult to guarantee the accuracy of turbulence measurement in complex terrain scenarios, resulting in significant measurement error problems.

[0008] To address the insufficient accuracy of single-lidar wind measurement in complex terrain, existing research has proposed a multi-radar collaborative wind measurement scheme. The mainstream approach involves using three scanning laser wind radars focused on the same measurement point. By collecting three sets of radial wind velocities from different azimuths at the same spatial point, a three-dimensional wind vector is synthesized, completely avoiding the assumption of wind field uniformity relied upon by traditional schemes. Existing experimental data shows that this collaborative wind measurement mode can effectively eliminate measurement errors caused by uneven airflow due to terrain, significantly improve the measurement accuracy of wind vectors and turbulence parameters, and significantly enhance the consistency between the measurement results and the actual data measured by the wind tower.

[0009] However, this three-radar collaborative wind measurement scheme has many unavoidable drawbacks in practical engineering applications, exhibiting extremely poor practicality and adaptability, making it unsuitable for large-scale application in wind power measurement projects. Firstly, multi-device collaborative observation places extremely high demands on radar installation locations, detection angles, and focusing accuracy. In complex terrain conditions, equipment site selection, debugging, and calibration are difficult, and power supply is also challenging, significantly increasing the workload of initial deployment. Secondly, the simultaneous operation of three scanning radars significantly increases equipment procurement, maintenance, and labor costs, substantially increasing the initial survey investment for wind power projects. Thirdly, wind power resource assessment requires continuous wind resource data at multiple height levels, while this multi-radar collaborative scheme can only achieve fixed-point measurement at a single height level. Switching detection heights leads to a significant reduction in the amount of effective observation data, prolonging the wind measurement cycle and severely impacting the efficiency of wind resource assessment, failing to meet the needs of engineering-oriented and routine wind field surveys. Figure 2 As shown, although the method of using three scanning radars to observe turbulence in a coordinated manner can effectively improve the accuracy of turbulence measurement, the overall scheme has prominent problems such as limited effective measurement data and significantly increased engineering implementation costs.

[0010] In summary, current mainstream wind measurement technologies have significant shortcomings: traditional single-lidar wind measurement methods are limited by the assumption of wind field uniformity, resulting in insufficient accuracy in measuring turbulence and wind vectors in complex terrain; multi-lidar collaborative high-precision wind measurement schemes suffer from high costs, difficult deployment, limited measurement dimensions, and long cycle times. Existing technologies cannot balance measurement accuracy, survey efficiency, and engineering practicality, making them unsuitable for the refined, efficient, and low-cost wind energy resource assessment needs of wind farms in complex terrain. Therefore, there is an urgent need to develop a lidar wind measurement method that balances accuracy, efficiency, and economy, and is adaptable to complex terrain, thus overcoming the current technological bottlenecks. Summary of the Invention

[0011] This application provides a method for turbulence measurement and correction based on bistatic radar to solve the problems of low wind measurement accuracy of single lidar in complex terrain, high cost of multi-radar collaboration, difficult deployment, and low wind measurement efficiency in the prior art.

[0012] This application provides a method for turbulence measurement and correction based on bistatic radar, including:

[0013] Acquire radial wind speed data from at least two radars, including radial wind speed data from the first radar and radial wind speed data from the second radar.

[0014] Based on the radial wind speed data from the first radar, the second-level wind speed sequence for each height layer is reconstructed to obtain the turbulence intensity to be corrected.

[0015] Based on the spatially intersecting beam data of the first radar radial wind speed data and the second radar radial wind speed data, the wind speed field at a specific height layer is reconstructed to obtain the reference turbulence intensity.

[0016] The turbulence error is calculated based on the turbulence intensity to be corrected and the reference turbulence intensity. The model is driven by the turbulence error training data to establish a mapping relationship between the turbulence intensity to be corrected and the turbulence error. The turbulence intensity to be corrected is then corrected based on the mapping relationship.

[0017] Prioritized, when reconstructing the second-level wind speed sequence at various altitudes or reconstructing the wind speed field at a specific altitude, the wind speed inversion equations used are as follows: ; ; ; in, Let be the azimuth angle of the i-th beam. Let be the elevation angle of the i-th beam. Let be the radial wind speed of the i-th beam. For horizontal wind speed, , For the horizontal wind speed component, Vertical wind speed, Wind direction; This is the north-alignment error.

[0018] Preferred, when solving the wind speed inversion equations, at least one of the following methods is used: least squares method, gradient descent method, Gauss-Newton method, or singular value decomposition method; wherein, when reconstructing the second-level wind speed sequence of each height layer using radial wind speed data from multiple beams of different azimuths, at least three beams of different azimuths from the first radar are used; when reconstructing the wind speed field of a specific height layer, spatially intersecting beam data from at least one beam from each of the first and second radars are used.

[0019] Preferably, the turbulence intensity to be corrected and the reference turbulence intensity are calculated using the following formulas: ; ; in, This represents the i-th second-level horizontal wind speed reconstructed from the radial wind speed data of the first radar. This corresponds to the average wind speed. This refers to the i-th second-level horizontal wind speed reconstructed from spatially intersecting beam data. denoted as the corresponding average wind speed, and N is the number of sampling points within the statistical period.

[0020] Prior to this, the turbulence error is calculated based on the turbulence intensity to be corrected and the reference turbulence intensity, wherein the formula for calculating the turbulence error is: ; in, For the turbulence intensity to be corrected, For reference turbulence intensity, This is the turbulence error.

[0021] Prior to this, a model is driven by the turbulence error training data to establish a mapping relationship between the turbulence intensity to be corrected and the turbulence error, wherein the mapping relationship is as follows: ; in, This is a mapping function.

[0022] Prior to establishing the mapping relationship between the turbulence intensity to be corrected and the turbulence error based on the turbulence error training data driving model, the method further includes constructing a dataset, wherein the turbulence intensity to be corrected is synchronized with the reference turbulence intensity in time, and data that meets the preset data efficiency threshold, preset wind speed threshold range and preset wind direction threshold range are selected to construct a target height layer dataset.

[0023] Prioritizing the correction of the turbulence intensity to be corrected according to the mapping relationship, including: ; in, This is the corrected turbulence intensity; The turbulence intensity to be corrected.

[0024] Prior to, after correcting the turbulence intensity to be corrected according to the mapping relationship, the process includes: when the corrected turbulence intensity error does not meet a preset threshold, recalculating the turbulence error, training the data-driven model, and correcting the error using the currently accumulated maximum available dataset, so as to update the mapping relationship.

[0025] Prior to acquiring radial wind speed data from at least two radars, the following steps are included: Obtain the target height of the virtual wind measurement tower and the elevation angle of the radar beam; The radar deployment spacing is determined based on the target height of the virtual wind measurement tower and the elevation angle of the radar beam. The radars are installed according to the radar deployment spacing, and it is determined whether the installation conditions are met. The installation conditions include that all radars are installed in the same direction and at the same altitude. If the installation conditions are not met, geometric compensation is added. The geometric compensation includes wind direction offset compensation and altitude correction compensation.

[0026] Therefore, this application has the following beneficial effects: (1) This invention constructs a virtual wind measurement tower by using spatially intersecting beam data from dual-station radars to obtain high-precision reference turbulence intensity, replacing the traditional physical wind measurement tower. This eliminates the strong dependence on the assumption of wind field uniformity and uses spatially intersecting beams from dual radars to reconstruct the real wind speed field, effectively improving the accuracy of turbulence measurement under complex terrain. It solves the problem of low wind measurement accuracy of single lidar in complex terrain. At the same time, it optimizes the azimuth distribution scheme of dual-station radar beams, and only two radars are needed to construct a virtual wind measurement tower, which greatly reduces equipment investment and the difficulty of site selection and deployment. It avoids the defects of high cost, difficult site selection, and poor engineering implementation of multi-radar collaborative schemes. After the model training is completed, the second radar can be removed, and a single radar can continuously output high-precision wind measurement results, balancing measurement accuracy and system cost. (2) In the process of establishing the compensation model, this application provides local training data by constructing a virtual wind measurement tower, which can effectively solve the problems of the mobility and adaptability of the compensation model and avoid correction errors caused by differences in terrain, climate and other factors. By processing the measured data through a data-driven method, the measurement error can be accurately estimated and the accuracy of turbulence assessment can be improved. During the operation of the equipment, training data can be continuously generated and used for continuous model iteration. During the measurement process, the accuracy of the model is detected and training data is continuously accumulated, realizing real-time iteration of the correction model, ensuring the continuous operation accuracy of the model, and avoiding the impact of annual and seasonal climate changes on the accuracy of turbulence correction. (3) This invention obtains the turbulence to be corrected by reconstructing the wind speed sequence at the second level of a single radar, constructs the reference turbulence by combining the intersecting beams of two radars, establishes the mapping relationship between turbulence intensity and error by using a data-driven model, realizes automatic correction of turbulence at all height levels, does not require equipment to be debugged layer by layer, ensures continuous data acquisition at multiple height levels, significantly improves wind measurement efficiency and shortens the wind measurement cycle; (4) The present invention introduces radar deployment spacing planning, installation condition verification and geometric compensation mechanism, which can be adapted to different terrain and altitude installation conditions. Combined with data screening, time synchronization and model iteration update strategy, it further ensures the stability and reliability of turbulence correction results, and can meet the engineering application needs of refined assessment of wind energy resources in complex terrain.

[0027] This solves the problems of low wind measurement accuracy of single lidar in complex terrain, high cost of multi-radar collaboration, difficult deployment, and low wind measurement efficiency in existing technologies.

[0028] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0029] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of conventional lidar measurement based on the background technology. Figure 2 This is a schematic diagram of three radars conducting coordinated observation based on the background technology. Figure 3 This is a flowchart of a bistatic radar-based turbulence measurement and correction method provided according to an embodiment of this application; Figure 4 This is a schematic diagram of the beam distribution of a bistatic radar according to an embodiment of this application; Figure 5 This is a logical framework diagram of a bistatic radar-based turbulence measurement and correction method according to an embodiment of this application; Figure 6This is a schematic diagram of the relative positions of a bistatic radar deployment scheme according to an embodiment of this application. Detailed Implementation

[0030] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] The following describes an embodiment of the present application of a turbulence measurement and correction method based on bistatic radar, with reference to the accompanying drawings. Addressing the low turbulence accuracy issue mentioned in the background art, this application provides a turbulence measurement and correction method based on bistatic radar. In this method, a virtual wind measurement tower is constructed using spatially intersecting beams of bistatic radar, overcoming the limitations of the traditional assumption of wind field uniformity, accurately reconstructing the real wind speed field, and significantly improving the accuracy of turbulence measurement in complex terrain. Using only two radars avoids the drawbacks of high cost, difficult deployment, and low efficiency of multi-radar collaboration. After model training, the auxiliary radar can be removed, allowing for long-term high-precision wind measurement with a single radar. Simultaneously, relying on localized data-driven modeling solves the model migration and adaptation problem, enabling real-time iterative updates to adapt to annual and seasonal climate changes. Combined with automatic turbulence correction at all altitude levels, radar deployment planning, and geometric compensation mechanisms, this method balances low cost, high efficiency, and strong engineering adaptability while ensuring measurement accuracy, meeting the needs of refined assessment of wind energy resources in complex terrain. Therefore, it solves the problems of low wind measurement accuracy of single-laser radar in complex terrain, and high cost, difficult deployment, and low wind measurement efficiency of multi-radar collaboration in the prior art.

[0032] Figure 3 This is a flowchart illustrating a method for turbulence measurement and correction based on bistatic radar, provided in an embodiment of this application.

[0033] This application provides a method for turbulence measurement and correction based on bistatic radar, including:

[0034] Step S101: Obtain radial wind speed data from at least two radars, including radial wind speed data from the first radar and radial wind speed data from the second radar.

[0035] Among them, radial wind speed data refers to the wind speed component measured in the direction of the radar beam, that is, the projected velocity of the wind vector in the line of sight of the radar beam, with the direction facing the radar or the direction away from the radar being positive, and the specific positive direction is defined according to the radar configuration; the first radar radial wind speed data refers to the raw radial wind speed data collected by the first radar, which serves as the main observation radar, and is used to reconstruct the wind speed field at each height layer to obtain the turbulence intensity to be corrected; the second radar radial wind speed data refers to the raw radial wind speed data collected by the second radar, which serves as the auxiliary observation radar, and is used in conjunction with the first radar radial wind speed data to reconstruct the wind speed field at the location of the virtual wind measurement tower through spatially intersecting beams.

[0036] It is understood that the number of radars in this embodiment is not limited to two. Multiple auxiliary radars can be deployed to participate in collaborative observation, depending on the actual wind field terrain conditions, the number of measurement height layers, and accuracy requirements. At the same time, when deploying the two radars, it is necessary to ensure that the spatial baseline is reasonable and the detection beams can form a spatial intersection at the target height, so as to ensure that effective intersecting beam data can be extracted, providing a reliable original data foundation for subsequent wind speed field reconstruction, reference turbulence intensity calculation, and error model training.

[0037] It should be noted that the embodiments of this application acquire radial wind speed data from at least two radars. Although two radars (the first radar and the second radar) are used as an example for illustration, the technical solution of this application is not limited to two radars. Those skilled in the art will understand that the principles and methods described in this application can be extended to application scenarios involving three or more radars.

[0038] For example, in complex terrain conditions, three or more radars can be used for collaborative observation. Multiple virtual wind towers can be constructed using multiple sets of spatially intersecting beam data, or redundant observation data can be provided for the same virtual wind tower location, further improving the accuracy and reliability of the reference turbulence intensity. Simultaneously, any one of the multiple radars can serve as the primary observation radar (i.e., the first radar), while the remaining radars act as auxiliary observation radars, jointly participating in the selection of spatially intersecting beam data and the construction of virtual wind towers.

[0039] Understandably, as the number of radars increases, system redundancy and measurement accuracy can be further improved, but equipment costs and deployment complexity will also increase accordingly. In practical applications, those skilled in the art can flexibly select two or more radars for collaborative observation based on engineering needs (such as wind measurement accuracy requirements, terrain complexity, cost budget, etc.). Any technical solution that expands two radars into multiple radars based on the principles and methods described in this application falls within the protection scope of this application.

[0040] In this embodiment of the application, before acquiring radial wind speed data from at least two radars, the following steps are included:

[0041] Obtain the target height of the virtual wind measurement tower and the elevation angle of the radar beam;

[0042] The radar deployment spacing is determined based on the target height of the virtual wind measurement tower and the elevation angle of the radar beam.

[0043] The radars are installed according to the radar deployment spacing, and it is determined whether the installation conditions are met. The installation conditions include that all radars are installed in the same direction and at the same altitude. If the installation conditions are not met, geometric compensation is added. The geometric compensation includes wind direction offset compensation and altitude correction compensation.

[0044] Among them, the target height of the virtual wind measurement tower refers to the altitude of the location where the virtual wind measurement tower is to be constructed or its height relative to the radar mounting base; the radar beam elevation angle refers to the angle between the center line of the radar beam and the horizontal plane, which is usually determined by the radar's scanning strategy or optical design.

[0045] It is understood that the embodiments of this application calculate the optimal deployment spacing of the radar by matching the target height and the beam elevation angle, which can ensure that the detection beams of the two radars effectively intersect at the target height and form a stable spatial intersection observation point. At the same time, by unifying the installation orientation and controlling the installation altitude, the measurement error caused by terrain and installation deviation is reduced. For scenarios where the on-site terrain is limited and the same direction and altitude installation conditions cannot be met, wind direction offset compensation and height correction compensation are introduced to correct geometric errors. This can effectively eliminate the influence of installation attitude and altitude difference on wind speed field reconstruction and turbulence calculation, and improve the adaptability and measurement reliability of the wind measurement scheme under complex field conditions.

[0046] It should be noted that wind direction offset compensation is used to correct wind direction measurement deviations caused by inconsistent radar installation orientations. It unifies the measurement results of each radar to the same reference coordinate system through coordinate transformation. Altitude correction compensation is used to correct altitude layer matching deviations caused by different radar installation altitudes. It normalizes the measurement results of each radar to the same reference altitude through interpolation or geometric projection.

[0047] Step S102: Based on the radial wind speed data from the first radar, reconstruct the second-level wind speed sequence for each height layer to obtain the turbulence intensity to be corrected.

[0048] It is understood that the embodiments of this application utilize multi-azimuth radial wind speed data from the main observation radar, and perform wind speed component calculations at each height layer through the wind speed inversion equation set to reconstruct a continuous wind speed time series on a second-level time scale. Based on this second-level wind speed series, the turbulence intensity at each height layer is statistically calculated. This turbulence intensity relies solely on independent observations from a single radar and has not undergone spatial consistency error correction. It is affected by the assumption of wind field homogeneity and airflow distortion in complex terrain, and thus has inherent measurement bias. Therefore, it is defined as the turbulence intensity to be corrected, serving as the basic input for subsequent model error training and accuracy correction.

[0049] Step S103: Based on the spatially intersecting beam data of the first radar radial wind speed data and the second radar radial wind speed data, reconstruct the wind speed field at a specific height layer to obtain the reference turbulence intensity.

[0050] Among them, spatially intersecting beam data refers to the radial wind speed data corresponding to the beams emitted by the first radar and the second radar that intersect at a specific location in space.

[0051] It is understood that this application embodiment utilizes the intersecting beam data of two radars at the same spatial intersection point, abandoning the traditional assumption of wind field homogeneity, and directly reconstructs the real wind speed field at a specific altitude layer where the intersection point is located; it calculates the three-dimensional wind vector based on the multi-angle observation information from two stations, and then statistically obtains high-precision turbulence intensity. This turbulence intensity originates from the inversion of the real wind field at the same point, avoiding the spatial wind field heterogeneity error caused by terrain undulations, and can be used as a true benchmark to form a reference turbulence intensity, providing a reliable reference for subsequent calculation of turbulence error and training data to drive model correction.

[0052] In this embodiment of the application, the wind speed inversion equations used when reconstructing the second-level wind speed sequence at each altitude level or reconstructing the wind speed field at a specific altitude level are as follows: ; ; ; in, Let be the azimuth angle of the i-th beam. Let be the elevation angle of the i-th beam. Let be the radial wind speed of the i-th beam. For horizontal wind speed, , For the horizontal wind speed component, Vertical wind speed, Wind direction; This is the north-alignment error.

[0053] In this embodiment of the application, when solving the wind speed inversion equation set, at least one of the following methods is used: least squares method, gradient descent method, Gauss-Newton method, or singular value decomposition method; wherein, when reconstructing the second-level wind speed sequence of each height layer using radial wind speed data from multiple beams of different azimuths, at least three beams of different azimuths from the first radar are used; and when reconstructing the wind speed field of a specific height layer, spatially intersecting beam data of at least one beam from each of the first and second radars are used.

[0054] It should be noted that the specific number of beams is not limited in the embodiments of this application, as long as the solvability condition of the equation system is met. Specifically, the wind speed inversion equation system involves three unknowns (horizontal wind speed components u and v, and vertical wind speed w), therefore, radial wind speed data from at least three beams in different directions are required for the solution. In practical applications, using four beams can obtain more stable solution results, but three or five beams are also applicable.

[0055] like Figure 4 As shown, radar 1 has five beams a, b, c, d, and e in different azimuths, while radar 2 has beams 1, 2, 3, 4, and 5. In the single-station radar wind speed inversion process, data from four beams (such as a, b, c, and d) are used to reconstruct the second-level wind speed sequence at each altitude level. In practice, this application is not strictly limited to using these four beams; any three to five beams or different combinations thereof can be used.

[0056] When reconstructing the wind speed field at a specific altitude, spatially intersecting beam data is used. Provided the spatial intersection condition is met, each radar can use multiple beams in the calculation. For example... Figure 4 As shown, beams d and e from the first radar and beams 1 and 5 from the second radar, totaling four beams, are used for joint calculation to obtain a more accurate reference turbulence intensity. This application does not limit the specific selection of spatially intersecting beams; any combination of beams capable of forming spatial intersections at the target altitude layer is applicable.

[0057] In addition, the least squares method, gradient descent method, Gauss-Newton method, and singular value decomposition method are all commonly used numerical methods for solving overdetermined systems of equations. Those skilled in the art can choose the appropriate method based on actual computing resources and accuracy requirements. For example, the least squares method has high computational efficiency and is suitable for real-time processing scenarios; the singular value decomposition method has good numerical stability and is suitable for scenarios with poor data quality.

[0058] In this embodiment, the turbulence intensity to be corrected and the reference turbulence intensity are calculated using the following formulas: ; ; in, This represents the i-th second-level horizontal wind speed reconstructed from the radial wind speed data of the first radar. This corresponds to the average wind speed. This refers to the i-th second-level horizontal wind speed reconstructed from spatially intersecting beam data. denoted as the corresponding average wind speed, and N is the number of sampling points within the statistical period.

[0059] Step S104: Calculate the turbulence error based on the turbulence intensity to be corrected and the reference turbulence intensity, drive the model based on the turbulence error training data to establish the mapping relationship between the turbulence intensity to be corrected and the turbulence error, and correct the turbulence intensity to be corrected based on the mapping relationship.

[0060] The mapping relationship refers to the functional relationship between the turbulence intensity to be corrected and the turbulence error established through a data-driven model. The data-driven model includes, but is not limited to, neural networks, support vector machines, random forests, decision trees, etc.

[0061] It is understood that this application uses a high-precision, spatially hypothetical-free reference turbulence intensity as the true benchmark to solve for the inherent turbulence error of the turbulence intensity to be corrected obtained from a single radar measurement. Through training data from massive amounts of real-world samples, the model is driven to autonomously fit the nonlinear mapping law between turbulence measurements and turbulence errors, overcoming the limitations of traditional fixed-formula correction. Based on this mapping relationship, adaptive error compensation can be quickly performed on the turbulence intensity to be corrected output by single radar at various altitude levels, effectively eliminating measurement deviations caused by the assumption of wind field uniformity under complex terrain. Ultimately, high-precision turbulence intensity data that closely matches real wind field conditions is output, achieving low-cost, high-efficiency, and accurate measurement of turbulence across the entire domain.

[0062] In this embodiment of the application, the turbulence error is calculated based on the turbulence intensity to be corrected and the reference turbulence intensity, wherein the formula for calculating the turbulence error is: ; in, For the turbulence intensity to be corrected, For reference turbulence intensity, This is the turbulence error.

[0063] In this embodiment of the application, a model is driven by turbulence error training data to establish a mapping relationship between the turbulence intensity to be corrected and the turbulence error, wherein the mapping relationship is as follows: ; in, This is a mapping function.

[0064] In this embodiment of the application, before establishing the mapping relationship between the turbulence intensity to be corrected and the turbulence error based on the turbulence error training data-driven model, the method further includes constructing a dataset, wherein the turbulence intensity to be corrected is synchronized with the reference turbulence intensity in time, and data that meets the preset data efficiency threshold, preset wind speed threshold range and preset wind direction threshold range are selected to construct a target height layer dataset.

[0065] Among them, time synchronization refers to aligning the turbulence intensity to be corrected with the reference turbulence intensity in the time dimension to ensure that they correspond to the same observation time; the preset data validity threshold refers to the minimum pass rate requirement for screening radar data quality, such as ≥80%; the preset wind speed threshold range refers to the upper and lower boundaries for screening valid wind speed data. When the wind speed is below the lower limit, it may be in the radar detection blind zone or in a state of low signal-to-noise ratio. When the wind speed is above the upper limit, it may exceed the radar's range or linear response range, such as 0 m / s to 60 m / s; the preset wind direction threshold range refers to the angle range for screening valid wind direction data, such as 0° to 360°.

[0066] It is understood that the embodiments of this application first ensure the temporal matching of the two sets of turbulence data through time synchronization to avoid model training deviation caused by time misalignment; then, through multiple conditions such as data effectiveness, wind speed range, and wind direction range, invalid samples under conditions of low radar signal-to-noise ratio, detection blind zone, over-range, and abnormal operating conditions are eliminated, and high-quality, highly consistent effective data are retained to construct a target height layer-specific dataset, thereby reducing the interference of abnormal data on the fitting accuracy of the data-driven model and ensuring the generalization ability and correction effect of the established mapping relationship.

[0067] In this embodiment of the application, the correction of the turbulence intensity to be corrected according to the mapping relationship includes: ; in, This is the corrected turbulence intensity; The turbulence intensity to be corrected.

[0068] In this embodiment of the application, after correcting the turbulence intensity to be corrected according to the mapping relationship, the process includes: when the corrected turbulence intensity error does not meet the preset threshold, recalculating the turbulence error, training the data-driven model, and correcting the error using the currently accumulated maximum available dataset, so as to update the mapping relationship.

[0069] The preset threshold can be specifically calibrated, for example, setting the relative error of the corrected turbulence intensity to be no more than 3% to 5%.

[0070] It is understood that, after the initial correction of turbulence intensity, this embodiment introduces an accuracy self-verification mechanism. A preset error threshold is used as the evaluation standard to determine in real time whether the correction result meets the standard. When the accuracy does not meet the requirements, all valid observation samples accumulated on-site are automatically retrieved, the turbulence error is recalculated, the training data drives the model iteratively, and the mapping relationship between the turbulence intensity to be corrected and the turbulence error is updated. Through the closed-loop self-verification and dynamic update mechanism, it can adapt to changes in wind field characteristics caused by seasonal changes, climate change, and wind field environment evolution, continuously optimize the model correction capability, and ensure the stability and accuracy of turbulence measurement and correction results under complex terrain in the long term.

[0071] The present application proposes a method for turbulence measurement and correction based on bistatic radar. This method utilizes spatially intersecting beams from bistatic radars to construct a virtual wind measurement tower, overcoming the limitations of traditional wind field uniformity assumptions. It accurately reconstructs the real wind speed field and significantly improves the accuracy of turbulence measurement in complex terrain. Using only two radars avoids the drawbacks of high cost, difficult site selection, and low efficiency associated with multi-radar collaboration. After model training, the auxiliary radar can be removed, allowing for long-term, high-precision wind measurement with a single radar. Furthermore, relying on localized data-driven modeling solves the model migration and adaptation problem, enabling real-time iterative updates to adapt to annual and seasonal climate changes. Combined with automatic turbulence correction across all altitude layers, radar deployment planning, and geometric compensation mechanisms, this method balances low cost, high efficiency, and strong engineering adaptability while ensuring measurement accuracy, meeting the needs of refined wind energy resource assessment in complex terrain. Therefore, it solves the problems of low wind measurement accuracy with single-laser radar in complex terrain and high cost, difficult deployment, and low wind measurement efficiency associated with multi-radar collaboration in existing technologies.

[0072] The following will illustrate the proposed bistatic radar-based turbulence measurement and correction method through a specific embodiment, such as... Figure 5 As shown, by accumulating measured data in situ at the test site to complete model training, it can effectively adapt to various complex terrain turbulence measurement and error correction scenarios. The specific implementation details are as follows:

[0073] Step 1: Determine the radar deployment plan based on the target height of the virtual wind measurement tower, build the observation network environment, and achieve data exchange;

[0074] Step 11: The radar deployment spacing can be determined through simple geometric calculations based on the target height of the virtual wind measurement tower; in addition, the target height can be adjusted according to the actual situation.

[0075] Combination such as Figure 6 As shown, when the target height of the virtual wind measuring tower is H and the beam elevation angle is known, the radar spacing D can be calculated using trigonometric functions.

[0076] Step 12: Install and deploy the radars according to the radar spacing determined in Step 11. At the same time, it is necessary to ensure that the radars are installed in the same direction and at the same altitude. If this cannot be guaranteed, geometric compensation, such as wind direction shift and altitude correction, needs to be added in subsequent steps.

[0077] Step 13: Establish the observation network environment to achieve data exchange between the two ends;

[0078] Step 2: Based on single-end data, the wind field inversion algorithm is used to reconstruct the wind speed at multiple altitude levels of the radar in sequence, and turbulence information is statistically analyzed;

[0079] Step 21: Collect and analyze radial wind speed data of a single-station radar target height layer within a statistical period;

[0080] Step 22: Based on the radial wind speed obtained in Step 21, reconstruct the second-level wind speed using a wind speed inversion algorithm. The equations are as follows. ; ; ; in, Horizontal wind speed; Wind direction; This is the north-alignment error; Radial wind speed; The radial beam azimuth angle; The radial beam elevation angle is denoted by 1, 2, and N, which are the radial beam numbers. The above equations can be solved using the least squares method.

[0081] Step 23: Repeat steps 21 and 22 at each altitude level to obtain the second-level wind speed results for each altitude level under one statistical period.

[0082] Step 24: Repeat step 23 within a statistical period to obtain the second-level wind speed sequence for each height layer, and statistically obtain the mean and standard deviation of the wind speed. Calculate the turbulence intensity using the following formula. ; ; ; in, Turbulence intensity; Let i be the horizontal wind speed in the i-th second. The statistical period is determined based on the statistical cycle; in the wind power sector, the statistical cycle is typically 10 minutes. Take 600; Average wind speed; The standard deviation of wind speed;

[0083] according to Figure 4 As can be seen, radar 1 has 5 beams ae in different azimuths, and radar 2 has beams 1-5. The wind speed inversion equations involve three unknowns (u, v, w), and solving these equations requires at least 3 beams of data in different azimuths. Therefore, 4 beams of data (a, b, c, d) are used in the single-station radar wind speed inversion process. In fact, this invention is not strictly limited to using 4 beams of data, and can use 3-5 beams of data or different combinations thereof.

[0084] The wind speed inversion equations can be solved using methods other than the least squares method, including but not limited to gradient descent, Gauss-Newton method, and singular value decomposition.

[0085] Step 3: Reconstruct the wind speed at the target height level of the virtual wind measurement tower based on the dual-end data, and collect turbulence information;

[0086] Step 31: Collect and analyze radial wind speed data at the target height level of the bistatic radar within one statistical period;

[0087] Step 32: Based on the intersecting azimuth of the bistatic radar, the radial wind speed obtained in step 31 is screened, and the wind speed inversion algorithm is used to reconstruct the second-level wind speed.

[0088] Step 33: Repeat steps 31 and 32 within a statistical period to obtain the second-level wind speed sequence at the target height level, and statistically obtain the mean and standard deviation of the wind speed.

[0089] To construct a virtual wind measurement tower, it is necessary to use spatially intersecting beam data, combined with... Figure 4 As can be seen, the four beams—beams d and e from radar 1 and beams 1 and 5 from radar 2—can be used to solve the wind speed inversion equations.

[0090] Step 4: Repeat steps 2 and 3 to construct the wind measurement dataset, train and establish the turbulence compensation model;

[0091] Step 41: Construct a target height layer dataset using the radar data obtained in step 3 and the virtual wind tower data obtained in step 4, and process it, mainly including: (1) time matching to align radar and virtual wind tower data; (2) radar data effectiveness rate not less than 80%; (3) effective wind speed range: 0-60m / s; (4) effective wind direction range: 0-360°; In addition, data cleaning rules can be added or adjusted according to the actual situation.

[0092] Step 42: Calculate the radar turbulence and virtual wind tower turbulence error based on the dataset obtained in Step 41. The calculation formula is as follows: ; in, The turbulence intensity was measured using a virtual wind measurement tower; To measure the turbulence intensity using radar;

[0093] Step 43: Use the turbulence error training data obtained in step 42 to drive the model and establish the mapping relationship between radar measurement results and turbulence error; ; in, To establish the mapping relationship between radar measurement results and turbulence errors, a neural network model can be used.

[0094] Mapping relationships can be constructed using methods other than neural networks, including but not limited to support vector machines, random forests, and decision trees.

[0095] Step 5: Using the turbulence compensation model obtained in Step 4, correct the measurement results of each height layer of the single-station radar in sequence.

[0096] Step 51: Using the turbulence compensation model trained in Step 4, turbulence errors are deduced based on radar measurement results at a specified altitude, and the radar results are corrected. The calculation formula is as follows: ; in, This is the corrected turbulence intensity; The original turbulence intensity; To estimate the deviation; Step 52: Repeat step 51 at each altitude level to compensate for the turbulence at each altitude level of the radar. Step 53: Evaluate the inference error of the target height layer. If the error fails to meet the threshold, repeat step 4 using the currently accumulated maximum available dataset to retrain the turbulence compensation model.

[0097] In summary, this embodiment, based on a dual-station radar collaborative observation architecture, constructs an in-situ virtual wind measurement tower using spatially intersecting beams from two radars. It utilizes on-site measured data to train a data-driven turbulence correction model, effectively solving the problems of poor accuracy in turbulence measurement under complex terrain and high cost of multi-radar collaborative deployment in traditional single-radar systems. This solution significantly improves the terrain adaptability and generalization ability of the correction model through localized measured sample modeling, overcoming the shortcomings of traditional models such as poor transferability and susceptibility to terrain and climate differences. Simultaneously, this solution can continuously accumulate measured data and iteratively update model parameters during long-term wind measurement, dynamically adapting to seasonal and annual climate changes, and ensuring long-term turbulence correction accuracy. Compared to traditional wind measurement schemes, this embodiment only requires two radars to complete high-precision turbulence calibration. After training, it can operate independently using a single radar, achieving accurate turbulence correction across all altitude levels while significantly reducing equipment investment and engineering deployment difficulty. It balances measurement accuracy, operational stability, and engineering economy, making it highly suitable for refined wind energy resource survey scenarios in various complex terrains.

[0098] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for turbulence measurement and correction based on bistatic radar, characterized in that, include: Acquire radial wind speed data from at least two radars, including radial wind speed data from the first radar and radial wind speed data from the second radar. Based on the radial wind speed data from the first radar, the second-level wind speed sequence for each height layer is reconstructed to obtain the turbulence intensity to be corrected. Based on the spatially intersecting beam data of the first radar radial wind speed data and the second radar radial wind speed data, the wind speed field at a specific height layer is reconstructed to obtain the reference turbulence intensity. The turbulence error is calculated based on the turbulence intensity to be corrected and the reference turbulence intensity. The model is driven by the turbulence error training data to establish a mapping relationship between the turbulence intensity to be corrected and the turbulence error. The turbulence intensity to be corrected is then corrected based on the mapping relationship.

2. The turbulence measurement and correction method based on bistatic radar according to claim 1, characterized in that, When reconstructing the second-level wind speed sequence at various altitudes or reconstructing the wind speed field at a specific altitude, the wind speed inversion equations used are as follows: ； ； ； in, Let be the azimuth angle of the i-th beam. Let be the elevation angle of the i-th beam. Let be the radial wind speed of the i-th beam. For horizontal wind speed, , For the horizontal wind speed component, Vertical wind speed, Wind direction; This is the north-alignment error.

3. The turbulence measurement and correction method based on bistatic radar according to claim 2, characterized in that, When solving the wind speed inversion equations, at least one of the following methods is used: least squares method, gradient descent method, Gauss-Newton method, or singular value decomposition method. Among them, radial wind speed data from multiple beams of different azimuths are used. When reconstructing the second-level wind speed sequence at each height layer, at least three beams of different azimuths from the first radar are used. When reconstructing the wind speed field at a specific height layer, spatially intersecting beam data from at least one beam from each of the first and second radars are used.

4. The turbulence measurement and correction method based on bistatic radar according to claim 1, characterized in that, The turbulence intensity to be corrected and the reference turbulence intensity are calculated using the following formulas: ； ； in, This represents the i-th second-level horizontal wind speed reconstructed from the radial wind speed data of the first radar. This corresponds to the average wind speed. This refers to the i-th second-level horizontal wind speed reconstructed from spatially intersecting beam data. denoted as the corresponding average wind speed, and N is the number of sampling points within the statistical period.

5. The turbulence measurement and correction method based on bistatic radar according to claim 1, characterized in that, The turbulence error is calculated based on the turbulence intensity to be corrected and the reference turbulence intensity, wherein the formula for calculating the turbulence error is: ； in, For the turbulence intensity to be corrected, For reference turbulence intensity, This is the turbulence error.

6. The turbulence measurement and correction method based on bistatic radar according to claim 1, characterized in that, The turbulence error training data is used to drive a model to establish a mapping relationship between the turbulence intensity to be corrected and the turbulence error, wherein the mapping relationship is as follows: ； in, This is a mapping function.

7. The method for turbulence measurement and correction based on bistatic radar according to claim 1, characterized in that, Before establishing the mapping relationship between the turbulence intensity to be corrected and the turbulence error based on the turbulence error training data-driven model, the method further includes constructing a dataset, wherein the turbulence intensity to be corrected is synchronized with the reference turbulence intensity in time, and data that meets the preset data efficiency threshold, preset wind speed threshold range and preset wind direction threshold range are selected to construct a target height layer dataset.

8. The turbulence measurement and correction method based on bistatic radar according to claim 1, characterized in that, The correction of the turbulence intensity to be corrected according to the mapping relationship includes: ; in, This is the corrected turbulence intensity; The turbulence intensity to be corrected.

9. A method for turbulence measurement and correction based on bistatic radar according to claim 8, characterized in that, After correcting the turbulence intensity to be corrected according to the mapping relationship, the process includes: when the corrected turbulence intensity error does not meet the preset threshold, recalculating the turbulence error, training the data-driven model, and correcting the error using the currently accumulated maximum available dataset, so as to update the mapping relationship.

10. A method for turbulence measurement and correction based on bistatic radar according to claim 1, characterized in that, Before acquiring radial wind speed data from at least two radars, the following should be included: Obtain the target height of the virtual wind measurement tower and the elevation angle of the radar beam; The radar deployment spacing is determined based on the target height of the virtual wind measurement tower and the elevation angle of the radar beam. The radars are installed according to the radar deployment spacing, and it is determined whether the installation conditions are met. The installation conditions include that all radars are installed in the same direction and at the same altitude. If the installation conditions are not met, geometric compensation is added. The geometric compensation includes wind direction offset compensation and altitude correction compensation.

Images