A downscaling wind resource assessment method and system

Abstract

The present invention belongs to the technical field of wind resource assessment, and discloses a downscaling wind resource assessment method and system, including constructing a mesoscale forecast model for a preset area, and simulating to obtain a mesoscale meteorological data set for the preset area within a preset period; obtaining a refined meteorological data set for the preset area within a preset period, and obtaining a downscaling model based on the data set training; converting the mesoscale forecast results output by the mesoscale forecast model into refined forecast results according to the downscaling model, and performing wind resource assessment based on the refined forecast results. The present invention optimizes the downscaling process through an artificial intelligence method, reduces the computing resources required for converting mesoscale forecast results to refined forecast results, shortens the time for data processing and result output, and improves overall work efficiency. It overcomes the limitations of traditional wind resource assessment methods in terms of computing cost, prediction accuracy, and real-time performance. It provides precise and reliable data support for wind farm site selection, etc.

Description

Technical Field

[0001] The invention discloses a downscaling wind resource assessment method and system, belonging to the technical field of wind resource assessment. Background Art

[0002] With the transformation of the global energy structure and the in-depth implementation of the sustainable development strategy, wind energy, as a clean and renewable energy form, has been increasingly valued for its development and utilization. As a pre-stage of wind energy project development, wind resource assessment has immeasurable value in maximizing wind farm power generation efficiency, optimizing project layout, reducing investment risks, and improving economic returns. Accurate wind resource assessment can enhance the diversity and stability of energy supply, ensure national energy security, and provide solid data support for the government to formulate long-term and scientific renewable energy policies and plans.

[0003] At present, there are two main methods used in the field of wind resource assessment: computational fluid dynamics (CFD) methods and mesoscale model methods. However, these methods have many shortcomings. The CFD method is computationally complex, has limited accuracy, and has a high demand for high-resolution terrain and meteorological data, resulting in high costs. Mesoscale models focus on simulations over larger spatial ranges and time scales, and can capture the impact of changes in larger-scale meteorological systems on wind energy resources, but the resolution is insufficient and cannot capture local changes. At the same time, they rely on parameterized descriptions and are more likely to introduce systematic errors. In addition, for large-scale, long-term simulation tasks, mesoscale models also require a large amount of computing resources, and the real-time and accuracy of the simulation results are difficult to guarantee. Summary of the invention

[0004] In view of the limitations and shortcomings of existing wind resource assessment methods, the purpose of the present invention is to provide a downscaling wind resource assessment method and system to solve the technical problems in the prior art that the reliance on parameterized description easily leads to the introduction of errors and large amount of calculation. To achieve the above purpose, the present invention proposes a downscaling wind resource assessment method and system, and the specific scheme is as follows:

[0005] First aspect:

[0006] A downscaling wind resource assessment method, comprising:

[0007] Step 1: construct a mesoscale forecast model for a preset area, and simulate to obtain a mesoscale meteorological data set for the preset area within a preset period;

[0008] Step 2: obtaining a refined meteorological data set of the preset area within a preset period, and training a downscaling model based on the mesoscale meteorological data set and the refined meteorological data set;

[0009] Step 3: Convert the mesoscale forecast results output by the mesoscale forecast model into refined forecast results according to the downscaling model, and perform wind resource assessment based on the refined forecast results.

[0010] Preferably, obtaining a refined meteorological data set for the preset area within a preset time period specifically includes:

[0011] Dividing the preset area into multiple areas according to preset requirements, and constructing a regional refined forecast model for each area;

[0012] According to each of the regional refined forecast models, refined meteorological data sets for the multiple regions within a preset time period are obtained by simulation.

[0013] Preferably, after dividing the preset area into a plurality of areas according to preset requirements, the method further includes:

[0014] A target area is selected in the preset area.

[0015] Preferably, the step 3 specifically includes:

[0016] Simulating the mesoscale forecast model to obtain a mesoscale forecast result for the preset area;

[0017] Dividing the preset area into a plurality of first areas, and converting the mesoscale forecast result of each first area into a regional refined forecast result according to the downscaling model;

[0018] The regional refined forecast results of the plurality of first regions are combined to obtain the refined forecast result of the preset region, and the wind resource assessment is performed according to the refined forecast result.

[0019] Preferably, after converting the mesoscale forecast result output by the mesoscale forecast model into a refined forecast result according to the downscaling model, the method further comprises:

[0020] Correcting the refined forecast result.

[0021] Preferably, correcting the refined forecast result specifically includes:

[0022] Obtaining a refined forecast result of a grid point at a first preset time according to the downscaling model;

[0023] Acquire actual meteorological data of a preset point at the first preset time;

[0024] The refined forecast result is corrected according to the actual meteorological data and the distance between the preset point and the grid point.

[0025] Preferably, after correcting the refined forecast result, the method further includes:

[0026] Constructing a first data set according to the corrected refined forecast result and its corresponding mesoscale forecast result;

[0027] The downscaling model is trained based on the first data set.

[0028] Second aspect:

[0029] A downscaled wind resource assessment system, comprising: a cloud processing center and a user access terminal;

[0030] The cloud processing center stores computer program instructions for executing the downscaling wind resource assessment method described in the first aspect;

[0031] The user access terminal is connected to the cloud processing center to obtain the downscaled wind resource assessment result and perform a visual display.

[0032] Preferably, the system further comprises: a wind tower collection unit;

[0033] The wind tower collection unit is used to obtain wind measurement data and transmit the wind measurement data to the cloud processing center.

[0034] Preferably, the cloud processing center includes: a business server, a digital model server and an artificial intelligence server;

[0035] The business server is used to obtain the wind measurement data and perform visual data interaction with the user access terminal;

[0036] The digital model server is embedded with a mesoscale forecast model and a regional refined forecast model for forecasting meteorological data;

[0037] The artificial intelligence server is connected to the digital model server and is used to train a downscaling model according to the meteorological data, and the downscaling model outputs a refined forecast result;

[0038] The artificial intelligence server is also used to correct the refined forecast result according to the wind measurement data.

[0039] Beneficial effects: The present invention optimizes the downscaling process through artificial intelligence methods, greatly reducing the computing resources required to process mesoscale forecast results to high-resolution refined forecast results, making large-scale, high-resolution wind resource assessment possible, while shortening the time for data processing and result output, and improving overall work efficiency. It also overcomes the limitations of traditional wind resource assessment methods in terms of computing cost, prediction accuracy and real-time performance.

[0040] The present invention utilizes the powerful nonlinear mapping capability of deep learning models and combines artificial intelligence downscaling technology to more accurately refine mesoscale meteorological data to a smaller spatial scale, significantly improving the spatial resolution and accuracy of wind resource prediction, and providing more refined and reliable data support for wind farm site selection, layout optimization, and wind turbine performance evaluation. It realizes the intelligent transformation from data to decision-making.

[0041] The present invention has the ability to access and process the latest meteorological data in real time, can update the wind resource prediction results in real time as the meteorological conditions change, realize dynamic adjustment, provide timely and accurate decision-making basis for wind farm operation and management, effectively cope with the complex and changeable natural environment, and improve the stability and efficiency of wind energy utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a schematic flow chart of a downscaled wind resource assessment method in Embodiment 1 of the present invention;

[0043] Figure 2 This is a schematic diagram of a process for constructing a refined meteorological data set in Embodiment 1 of the present invention;

[0044] Figure 3 This is a schematic diagram of dividing a mesoscale meteorological region into multiple refined regions in Embodiment 1 of the present invention;

[0045] Figure 4 This is a schematic diagram of a downscaled wind resource assessment system in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with specific implementation methods. It should be understood that the specific implementation methods described here are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0047] Embodiment 1:

[0048] like Figure 1 As shown, in one embodiment, a downscaling wind resource assessment method is disclosed to achieve efficient and refined wind resource assessment of an area to be assessed, comprising the following steps:

[0049] S1. Construct a mesoscale forecast model. Specifically, construct a mesoscale forecast model for a preset area, and simulate and obtain a mesoscale meteorological data set for the preset area within a preset period;

[0050] Specifically, in this embodiment, a mesoscale meteorological numerical forecast model is constructed, and the mesoscale meteorological numerical forecast model has a large coverage. For example, it covers the whole of China, or covers a rectangular area of ​​several provinces. The resolution of the mesoscale meteorological numerical forecast model is usually several kilometers, such as 6-10km; at the same time, the model is used to simulate the past weather change process based on historical data. In order to ensure the number of simulated data sets for training, the simulation time span reaches more than 1 year, and the simulation results are made into a mesoscale meteorological data set. Through this process, we not only ensure the temporal continuity of the simulation results, but also enhance the diversity and representativeness of the data set by accumulating data covering a longer period of time, which provides strong data support for subsequent downscaling processing and wind resource assessment.

[0051] S2. Construct multiple regional refined forecast models to obtain refined meteorological data sets for preset areas within preset time periods.

[0052] Furthermore, obtaining a refined meteorological data set for a preset area within a preset time period specifically includes: dividing the preset area into multiple areas according to preset requirements, and constructing a regional refined forecast model for each area; simulating and obtaining a refined meteorological data set for multiple areas within a preset time period according to the refined forecast model for each area.

[0053] Specifically, Figure 2 As shown, in this embodiment, multiple refined meteorological numerical prediction models are constructed, that is, multiple regional refined prediction models are constructed, and the coverage of the mesoscale meteorological numerical prediction model involved in S1 is divided into multiple regions according to the topography. In this embodiment, the multiple regions include major typical topographic regions (such as mountains, hills, plateaus, plains, basins, deserts, coasts, oceans, rivers, lakes and glaciers), and a high-resolution regional refined meteorological numerical prediction model is constructed in each region.

[0054] Specifically, the regional refined meteorological numerical prediction model for each region has appropriate parameters, and the regional refined meteorological numerical prediction models for multiple regions perform numerical simulations within the same time range as the mesoscale region involved in S1, and output simulation results of the same time, and produce a high-resolution refined meteorological data set based on the simulation results, and classify the refined meteorological data set as the basic data set.

[0055] Furthermore, after dividing the preset area into a plurality of areas according to the preset requirements, the method further includes: selecting a target area in the preset area.

[0056] In order to ensure that the meteorological forecast results obtained for specific areas (such as the area where a wind farm is planned to be built) are more detailed and accurate, it also includes selecting a target area within the coverage area involved in S1, building a regional refined meteorological numerical prediction model for the target area, and performing numerical simulations within the same time range as the mesoscale area involved in S1 based on the regional refined meteorological numerical prediction model for the target area, and outputting simulation results of the same time. Based on the simulation results, a high-resolution refined meteorological dataset is produced, and the refined meteorological dataset produced for the target area is classified as a supplementary dataset.

[0057] The above-mentioned high-resolution refined meteorological datasets and mesoscale meteorological datasets match each other, that is, they correspond to the same time range and geographical coverage.

[0058] S3. Training a downscaling model. Specifically, the downscaling model is obtained by training the mesoscale meteorological dataset and the refined meteorological dataset.

[0059] Specifically, in this embodiment, an artificial intelligence downscaling model, i.e., a downscaling model, is trained, with the data in the mesoscale meteorological dataset obtained in S1 as input, and the refined meteorological dataset obtained in S2 as output, and a deep learning framework is built to complete the artificial intelligence downscaling model training to obtain an artificial intelligence downscaling model; the training usually requires the use of all the data in the basic datasets in the mesoscale meteorological dataset and the refined meteorological dataset; in a further embodiment, it also includes using part or all of the data in the supplementary dataset in the refined meteorological dataset according to the regional characteristics that need to be evaluated.

[0060] After the above steps are completed, the model can be constructed and trained at one time, and then it can be put into the downscaling wind resource assessment stage, i.e., the operational operation stage. In this stage, the model will predict and evaluate the wind resources in the specified area at a preset frequency, daily, hourly, or at any set frequency. This provides a scientific basis for wind farm planning and operation management. In the operational operation process, the specific working process of the downscaling model includes S4 to S5. The details are as follows:

[0061] The mesoscale forecast results output by the mesoscale forecast model are converted into refined forecast results based on the downscaling model.

[0062] Furthermore, the mesoscale forecast results output by the mesoscale forecast model are converted into refined forecast results according to the downscaling model, including:

[0063] S4. Forecasting the mesoscale meteorological area according to the mesoscale forecast model, that is, obtaining the mesoscale forecast results of the preset area according to the mesoscale forecast model simulation.

[0064] Specifically, in this embodiment, a mesoscale meteorological numerical prediction model is operated in an operational manner, and regular operational mesoscale meteorological data forecasts are carried out to obtain the latest mesoscale meteorological forecast data, that is, the mesoscale forecast results.

[0065] S5. Converting the mesoscale forecast results of the mesoscale meteorological area into refined forecast results according to the downscaling model; specifically, dividing the preset area into multiple first areas, converting the mesoscale forecast results of each first area into regional refined forecast results according to the downscaling model; combining the regional refined forecast results of the multiple first areas to obtain the refined forecast results of the preset area, and conducting wind resource assessment based on the refined forecast results.

[0066] Specifically, Figure 3 As shown, in this embodiment, the mesoscale meteorological area corresponding to the mesoscale meteorological numerical forecast model, i.e., the preset area, is divided into several refined areas of equal area and with a certain intersection, i.e., the first area. Usually, the area of ​​intersection between two adjacent areas is about 10%. Subsequently, the mesoscale meteorological forecast data obtained in S4 is used to call the artificial intelligence downscaling model obtained in S3 multiple times to obtain high-resolution regional refined meteorological forecast data corresponding to each area, i.e., the regional refined forecast results.

[0067] The artificial intelligence downscaling model combines the high-resolution regional refined weather forecast data of each region to obtain the refined weather forecast data of the entire mesoscale meteorological region, that is, the refined forecast results. The specific combination method is as follows:

[0068]

[0069]

[0070] Where:

[0071] Indicates the number of the grid point;

[0072] Indicates that in any area The high-resolution results (either lateral wind speed or longitudinal wind speed) are obtained by calling the artificial intelligence downscaling model. is the regional identifier, Indicates the specific grid point coordinates representing the high-resolution results for this area;

[0073] Represents a weight factor, which is used to process grid points located at the junction of multiple regions at the same time; specifically, when the grid point When located at the intersection of multiple regions, it will have a specific weight for each region that contains it, which is assigned based on the distance of the grid point from the center of each region.

[0074] Represents a grid point The distance from the center of different areas, in meters (m);

[0075] Represents a constraint factor to avoid the denominator being 0. Usually a small enough positive value is selected to ensure that When the distance is too close, there will be no division by zero error. In this embodiment, 100 meters is used;

[0076] Dirac Function, used to identify grid points Is it a specific area? , specifically when the grid point is indeed within the region, is 1 if the region is smaller than 0; otherwise, it is 0. This function ensures that only grid points that actually belong to a region are included in the calculation for that region.

[0077] It is the final refined weather forecast data for the mesoscale meteorological region, which is the data of multiple refined regions. The weighted sum of .

[0078] In the process of business operation of the model, in order to ensure the accuracy and timeliness of the model, it is necessary to continuously calibrate the model based on field measured data. This calibration process continues with the business operation, providing direction for model optimization.

[0079] Furthermore, in order to ensure the precision and accuracy of the refined forecast results, after the mesoscale forecast results output by the mesoscale forecast model are converted into refined forecast results according to the downscaling model, it also includes: S6, correcting the refined forecast results.

[0080] Furthermore, correcting the refined forecast results specifically includes: obtaining the refined forecast results of the grid points at a first preset time according to the downscaling model; obtaining the actual meteorological data of the preset points at the first preset time; and correcting the refined forecast results according to the actual meteorological data and the distance between the preset points and the grid points.

[0081] Specifically, in this embodiment, the wind measurement data obtained by the wind tower is used to correct the refined forecast result, and the specific method is as follows:

[0082] Since the frequencies of wind tower data and forecast data are usually different, it is necessary to unify the time and record the wind speed results of a grid point at a preset time in the high-resolution regional refined meteorological forecast data of a certain area obtained by S5 simulation as The wind direction result is recorded as ;

[0083] The wind speed at the preset point where the wind tower is located at a preset time in the area simulated by the model is recorded as , wind direction is recorded as ;

[0084] The time interval between the forecast data output of two forecast simulations of the model (the time interval between the forecast data output is usually 1 hour) is used as the reference time interval. The average values ​​of the actual wind speed and wind direction measured by the wind tower at half the reference time interval around the forecast data output time are counted, and the average values ​​of the actual wind speed and wind direction are recorded as , .

[0085] The distance between the grid point and the wind tower in the forecast data is recorded as , the unit is , the correction formula of wind speed and wind direction of the wind tower is as follows:

[0086]

[0087]

[0088] Where:

[0089] The corrected grid point wind speed for a certain area;

[0090] The wind direction of the grid points after correction for a certain area.

[0091] S7. Outputting a refined wind resource assessment result according to the refined forecast result, that is, performing a wind resource assessment according to the refined forecast result.

[0092] Specifically, the refined meteorological forecast data of the entire mesoscale meteorological area with high resolution after correction, that is, the refined forecast result, is used for wind resource assessment. In this embodiment, the refined wind speed and wind direction results with high resolution and wide range after correction are output as the refined wind resource assessment result.

[0093] In order to avoid the model from being heavily dependent on the correction of measured data for a long time, it is also necessary to optimize the model through the corrected data to reduce the model prediction error and improve the adaptability and accuracy of the model under complex terrain and extreme weather conditions. In this implementation, the model optimization training specifically includes S8. That is, further, after correcting the refined forecast results, it also includes:

[0094] S8, downscaling model optimization training, specifically, constructing a first data set according to the corrected refined forecast results and their corresponding mesoscale forecast results; and training the downscaling model according to the first data set.

[0095] Specifically, in this embodiment, with the wind tower as the core, the corrected high-resolution forecast results obtained by S7 and the mesoscale forecast data results within the same regional range obtained by S4 are cut, matched and stored in the refined meteorological data set (supplementary data set) and the mesoscale meteorological data set, respectively, and the artificial intelligence downscaling model is retrained regularly. At the same time, the data with poor quality in the original high-resolution refined meteorological data set is eliminated to complete the upgrade training of the artificial intelligence downscaling model and improve the accuracy of the model.

[0096] Through the above-mentioned correction and re-optimization cycle, the model used for downscaling wind resource assessment can continuously improve itself in practical applications, providing more reliable technical support for the sustainable development of the wind power industry.

[0097] It should be understood that, in another embodiment, the mesoscale prediction model and the downscaling model in this embodiment may be coupled.

[0098] Embodiment 2:

[0099] A downscaled wind resource assessment system, comprising: a cloud processing center and a user access terminal;

[0100] The cloud processing center stores computer program instructions for executing the downscaling wind resource assessment method as in the first embodiment;

[0101] The user access terminal is connected to the cloud processing center to obtain the downscaled wind resource assessment results and perform visual display.

[0102] Furthermore, the system also includes a wind tower collection unit;

[0103] The wind tower acquisition unit is used to obtain wind measurement data and transmit the wind measurement data to the cloud processing center.

[0104] Furthermore, the cloud processing center includes: business servers, digital and analog servers, and artificial intelligence servers;

[0105] The business server is used to obtain wind measurement data and interact with the user access terminal for visual data;

[0106] The digital model server is embedded with mesoscale forecast model and regional refined forecast model to forecast meteorological data;

[0107] The artificial intelligence server is connected to the digital model server to train the downscaling model based on the meteorological data, and the downscaling model outputs refined forecast results;

[0108] The artificial intelligence server is also used to correct and refine forecast results based on wind measurement data.

[0109] Specifically, Figure 4 As shown, a downscaled wind resource assessment system provided in this embodiment includes a cloud core as a cloud processing center, a wind tower transmitter as a wind tower collection unit, and a user access terminal.

[0110] The cloud core consists of business servers, digital model servers, and artificial intelligence servers, which are used to complete the processing of data and business at the front and back ends.

[0111] Among them, the wind tower transmitter is deployed on the wind tower to collect wind measurement data in the wind tower and send it to the business server to improve the accuracy of wind resource assessment, especially real-time dynamic update;

[0112] User access terminals include PC and mobile phones.

[0113] The business server is mainly responsible for the interaction with the user access terminal and the wind tower, including real-time reading of the latest wind measurement data in the wind tower transmitter, eliminating invalid wind measurement data, storing the processed results in the database, and reading high-resolution wind resource assessment data results from the database to provide data visualization for access users;

[0114] The digital model server is mainly composed of a large number of CPUs and is used for numerical meteorological forecasting, including mesoscale numerical meteorological forecasting and high-resolution regional refined numerical meteorological forecasting.

[0115] AI servers are mainly composed of a large number of GPUs, which are used for AI downscaling model training;

[0116] The database is mainly used to store data generated by each server.

[0117] The present invention selects mountains, hills, plateaus, plains, basins, deserts, coasts, oceans, rivers, lakes and glaciers, and other typical landform area cases for simulation to establish a basic data set, ensuring that the data set fully captures the characteristic elements of the wind field in the natural environment, providing solid data support for the subsequent use of artificial intelligence methods to optimize the training of the downscaling model.

[0118] The present invention designs an artificial intelligence method to optimize the downscaling process, greatly reducing the computing resources required to process the mesoscale forecast results to high-resolution refined forecast results, making large-scale, high-resolution wind resource assessment possible, while shortening the time for data processing and result output, and improving overall work efficiency. It also overcomes the limitations of traditional wind resource assessment methods in terms of computing cost, prediction accuracy, and real-time performance.

[0119] The present invention utilizes the powerful nonlinear mapping capability of deep learning models and combines artificial intelligence downscaling technology to more accurately refine mesoscale meteorological data to a smaller spatial scale, significantly improving the spatial resolution and accuracy of wind resource prediction, and providing more refined and reliable data support for wind farm site selection, layout optimization, and wind turbine performance evaluation. It realizes the intelligent transformation from data to decision-making.

[0120] The present invention has the ability to access and process the latest meteorological data in real time, can update the wind resource prediction results in real time as the meteorological conditions change, realize dynamic adjustment, provide timely and accurate decision-making basis for wind farm operation and management, effectively cope with the complex and changeable natural environment, and improve the stability and efficiency of wind energy utilization.

[0121] The above are only several embodiments of the present invention and are not intended to limit the present invention in any form. Although the present invention is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any technician familiar with the profession, without departing from the scope of the technical solution of the present invention, using the above disclosed technical content to make slight changes or modifications are equivalent to equivalent implementation cases and fall within the scope of the technical solution.

Claims

1. A downscaling wind resource assessment method, characterized in that: include: Step 1: construct a mesoscale forecast model for a preset area, and simulate to obtain a mesoscale meteorological data set for the preset area within a preset period; Step 2: obtaining a refined meteorological data set of the preset area within a preset period, and training a downscaling model based on the mesoscale meteorological data set and the refined meteorological data set; Step 3: converting the mesoscale forecast results output by the mesoscale forecast model into refined forecast results according to the downscaling model, and performing wind resource assessment according to the refined forecast results; The step 3 specifically includes: Simulating the mesoscale forecast model to obtain a mesoscale forecast result for the preset area; Dividing the preset area into a plurality of first areas, and converting the mesoscale forecast result of each first area into a regional refined forecast result according to the downscaling model; Combining the regional refined forecast results of the plurality of first regions to obtain a refined forecast result of the preset region, and performing wind resource assessment according to the refined forecast result; The combination is as follows: , Where: Indicates the number of the grid point; In any first region In the figure, high-resolution results are obtained by calling the downscaling model; represents the weight factor; Represents a grid point distances to the centers of different first regions; Represents a constraint factor to avoid the denominator being 0; Dirac Function, used to identify grid points Whether it belongs to the specific first area , specifically when the grid point is indeed within the first region, is 1; otherwise, it is 0; It is the refined forecast result for the preset area.

2. The downscaling wind resource assessment method according to claim 1, characterized in that: Obtaining a refined meteorological data set for the preset area within a preset period of time specifically includes: Dividing the preset area into multiple areas according to preset requirements, and constructing a regional refined forecast model for each area; According to each of the regional refined forecast models, refined meteorological data sets for the multiple regions within a preset time period are obtained by simulation.

3. The downscaling wind resource assessment method according to claim 2, characterized in that: After dividing the preset area into a plurality of areas according to preset requirements, the method further includes: A target area is selected in the preset area.

4. The downscaling wind resource assessment method according to claim 1, characterized in that: After converting the mesoscale forecast result output by the mesoscale forecast model into a refined forecast result according to the downscaling model, the method further includes: Correcting the refined forecast result.

5. The downscaling wind resource assessment method according to claim 4, characterized in that: Correcting the refined forecast result specifically includes: Obtaining a refined forecast result of a grid point at a first preset time according to the downscaling model; Acquire actual meteorological data of a preset point at the first preset time; The refined forecast result is corrected according to the actual meteorological data and the distance between the preset point and the grid point.

6. The downscaling wind resource assessment method according to claim 4, characterized in that: After correcting the refined forecast result, the method further includes: Constructing a first data set according to the corrected refined forecast result and its corresponding mesoscale forecast result; The downscaling model is trained based on the first data set.

7. A downscaling wind resource assessment system, characterized in that: include: Cloud processing center and user access terminal; The cloud processing center stores computer program instructions for executing the downscaling wind resource assessment method according to any one of claims 1 to 6; The user access terminal is connected to the cloud processing center to obtain the downscaled wind resource assessment result and perform a visual display.

8. The downscaling wind resource assessment system according to claim 7, characterized in that: The system also includes: a wind tower collection unit; The wind tower collection unit is used to obtain wind measurement data and transmit the wind measurement data to the cloud processing center.

9. The downscaling wind resource assessment system according to claim 8, characterized in that: The cloud processing center includes: business servers, digital model servers and artificial intelligence servers; The business server is used to obtain the wind measurement data and perform visual data interaction with the user access terminal; The digital model server is embedded with a mesoscale forecast model and a regional refined forecast model for forecasting meteorological data; The artificial intelligence server is connected to the digital model server and is used to train a downscaling model according to the meteorological data, and the downscaling model outputs a refined forecast result; The artificial intelligence server is also used to correct the refined forecast result according to the wind measurement data.

Images