Meteorological state prediction method and system based on big data

Abstract

The embodiment of the invention discloses a meteorological state prediction method and system based on big data, and belongs to the technical field of meteorological data analysis. By acquiring the observation data of the multi-source wind power plant and identifying the state trend description event of the target meteorological observation data, information in the data can be fully mined, and a detailed description distribution label is obtained. The scale coupling evolution weight is determined based on the labels, so that the influence degree of meteorological elements on the overall meteorological state under different scales can be accurately mastered. Global trend evolution parameters and multi-source coupling evolution discrete indexes are obtained, the overall characteristics of various meteorological elements and multi-source data can be comprehensively considered, and the comprehensiveness of meteorological state analysis is improved. And state prediction viewpoints and confidence features are determined according to the parameters, so that a prediction result has a clear direction and credibility evaluation. The finally determined meteorological state prediction result is higher in accuracy, and reliable support can be provided for operation management of the wind power plant.

Description

Technical Field

[0001] This application belongs to the field of meteorological data analysis technology, specifically relating to a meteorological state prediction method and system based on big data. Background Technology

[0002] Accurate weather forecasting is crucial for the operation and management of wind farms. Traditional weather forecasting methods often fall short of the high-precision requirements of wind farms under complex conditions. Meteorological observation data from a single data source has limitations and cannot comprehensively reflect the meteorological conditions of the wind farm. Furthermore, there is a lack of effective means to comprehensively consider the influence of meteorological elements at different scales and the differences between different data sources when analyzing meteorological data. This leads to inaccurate weather forecast results, which cannot provide a reliable basis for the operation and management of wind farms. For example, the wind turbine layout may be unreasonable, the power generation efficiency may not reach the optimal level, and the safety of wind turbines may be difficult to guarantee under complex weather conditions. Summary of the Invention

[0003] This application provides a weather condition prediction method and system based on big data, which can solve or partially solve the technical problems involved in the background art.

[0004] This application provides a meteorological state prediction method based on big data, applied to a meteorological state prediction system. The method includes: acquiring multi-source wind farm observation data, and identifying state trend description events in the target meteorological observation data included in the multi-source wind farm observation data to obtain description distribution labels for several target state trend description events in the target meteorological observation data; determining the scale coupling evolution weight of at least one set target scale coupling evolution vector in the target meteorological observation data based on the description distribution labels of the several target state trend description events; acquiring the global trend evolution parameters and multi-source coupling evolution discrete index of each set target scale coupling evolution vector, wherein the global trend evolution parameters and multi-source coupling evolution discrete index are obtained by processing the scale coupling evolution weight of each set target scale coupling evolution vector of several different target meteorological observation data; determining the state prediction viewpoint and confidence feature of each set target scale coupling evolution vector based on the global trend evolution parameters, multi-source coupling evolution discrete index and scale coupling evolution weight of each set target scale coupling evolution vector; and determining the meteorological state prediction result of the multi-source wind farm observation data based on the state prediction viewpoint of each set target scale coupling evolution vector.

[0005] This application provides a weather condition forecasting system, including at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to perform the above-described method.

[0006] This application provides a readable storage medium on which a program or instruction is stored, and when the program or instruction is executed by a processor, it implements the steps of the above method.

[0007] This application's embodiments, by acquiring multi-source wind farm observation data and identifying state trend description events in the target meteorological observation data, can fully mine the information in the data and obtain detailed descriptive distribution labels. Determining scale-coupled evolution weights based on these labels helps to accurately grasp the degree of influence of meteorological elements on the overall meteorological state at different scales. Obtaining global trend evolution parameters and multi-source coupled evolution discrete indices allows for comprehensive consideration of the overall characteristics of various meteorological elements and multi-source data, improving the comprehensiveness of meteorological state analysis. Determining state prediction viewpoints and confidence features based on these parameters ensures that the prediction results have both a clear direction and a credibility assessment. The final meteorological state prediction results are more accurate and can provide reliable support for wind farm operation and management (optimizing wind turbine layout, improving power generation efficiency, and ensuring safe wind turbine operation). Attached Figure Description

[0008] Figure 1 A flowchart illustrating a weather state prediction method based on big data, provided as an embodiment of this application.

[0009] Figure 2 This is a schematic diagram of the structure of a weather condition prediction system provided in an embodiment of this application. Detailed Implementation

[0010] 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 the embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the embodiments of this application.

[0011] In the embodiments of this application, the terms "first," "second," etc., are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, the first object can be one or more. Furthermore, in the embodiments of this application, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0012] Figure 1 A meteorological state prediction method based on big data is shown and applied to a meteorological state prediction system. The method includes the following steps 101-105.

[0013] Step 101: The meteorological state prediction system acquires multi-source wind farm observation data and identifies state trend description events in the target meteorological observation data included in the multi-source wind farm observation data to obtain description distribution labels for several target state trend description events in the target meteorological observation data.

[0014] Step 102: The meteorological state prediction system determines the scale coupling evolution weight of at least one target scale coupling evolution vector in the target meteorological observation data by using the description distribution labels of the several target state trend description events.

[0015] Step 103: The meteorological state prediction system obtains the global trend evolution parameters and multi-source coupling evolution discrete index of each set target scale coupling evolution vector. The global trend evolution parameters and multi-source coupling evolution discrete index are obtained by processing the scale coupling evolution weights of each set target scale coupling evolution vector of several different target meteorological observation data.

[0016] Step 104: The meteorological state prediction system determines the state prediction viewpoint and the confidence characteristics of each state prediction viewpoint based on the global trend evolution parameters, multi-source coupling evolution discrete index and scale coupling evolution weight of each set target scale coupling evolution vector.

[0017] Step 105: The meteorological state prediction system determines the meteorological state prediction results of the multi-source wind farm observation data by using the state prediction perspective of the coupled evolution vectors of each set target scale.

[0018] The first step in the meteorological condition prediction system is to acquire multi-source wind farm observation data. Wind farm observation data comes from various sources, including direct observation data from weather stations, wind profiler radar data, and satellite remote sensing data. This multi-source data contains a wealth of meteorological information, and the system must perform state trend descriptions and event identification on the target meteorological observation data. For example, if the target meteorological observation data consists of wind speed and wind direction data, the state trend description event identification might classify wind speed as a series of events, such as a continuous increase (e.g., wind speed increases by more than 2 m / s in each of the last 5 observation periods compared to the previous period), stable wind speed (wind speed fluctuates within ±1 m / s in each of the last 10 observation periods), and decreasing wind speed (wind speed decreases by more than 1.5 m / s in each of the last 3 observation periods). Similarly, clockwise changes in wind direction (e.g., wind direction rotates more than 10 degrees clockwise in each of the last 8 observation periods), counterclockwise changes (wind direction rotates more than 15 degrees counterclockwise in each of the last 6 observation periods), or constant changes (wind direction fluctuates within ±5 degrees in each of the last 12 observation periods) are also classified as different events. Through this detailed identification, the system obtains descriptive distribution labels for several target state trend description events in the target meteorological observation data. For example, if the frequency of an event with continuously increasing wind speed is high within a certain period (e.g., a frequency of 0.4, meaning that continuously increasing wind speed occurs in 40% of the observation period), then its descriptive distribution label will reflect this high frequency characteristic. On the other hand, a stable wind direction event may have a unique descriptive distribution label in another period, which can reflect information such as the proportion of stable wind direction events relative to other events in that period (e.g., stable wind direction events account for 0.6 of all events in that period, i.e., 60%).

[0019] After obtaining these descriptive distribution labels, the meteorological state prediction system can determine the scale coupling evolution weight of at least one target-scale coupled evolution vector in the target meteorological observation data. Setting a target-scale coupled evolution vector is a concept established to accurately describe the evolution process of meteorological states. For example, for the meteorological element of wind speed, wind speed evolution vectors over time might be set at different spatial scales (such as small-scale near-surface and large-scale mesoscale). These vectors comprehensively reflect the changing trends of wind speed at different scales. The scale coupling evolution weight reflects the relative importance of each scale coupling evolution vector in the entire meteorological state evolution process. For example, if in a certain wind farm area, the change in wind speed at a small-scale near-surface has a significant impact on the overall meteorological state, then the weight of the scale coupling evolution vector related to the small-scale near-surface will be relatively high (e.g., the weight of the wind speed evolution vector at a small-scale near-surface is 0.7, while the weight of the wind speed evolution vector at a large-scale mesoscale is 0.3). This weight is determined based on the analysis of the descriptive distribution labels of the target state trend descriptive events obtained earlier. If a certain state trend describes an event that occurs frequently at a certain scale and has a significant impact on the overall meteorological state, then the weight of the scale-coupled evolution vector corresponding to that scale will increase accordingly.

[0020] Next, the meteorological state prediction system acquires the global trend evolution parameter and multi-source coupling evolution discrete index for each target-scale coupled evolution vector. The global trend evolution parameter is obtained by comprehensively processing the scale coupling evolution weights of each target-scale coupled evolution vector from several different target meteorological observation data. It reflects the overall evolution trend of the target-scale coupled evolution vector across the entire wind farm area, considering multiple meteorological factors. For example, for wind speed and wind direction, under different target-scale coupled evolution vectors, a global trend evolution parameter reflecting the comprehensive evolution trend of wind speed and wind direction across the entire wind farm is obtained through comprehensive analysis of their scale coupling evolution weights. The multi-source coupling evolution discrete index measures the degree of dispersion of the target-scale coupled evolution vector across different data sources from the perspective of multi-source data. For example, there may be some differences between wind speed data observed at meteorological stations and wind speed data obtained from satellite remote sensing; the multi-source coupling evolution discrete index reflects this difference in the target-scale coupled evolution vector. If the performance of a certain target-scale coupled evolution vector differs significantly between meteorological station data and satellite remote sensing data, its multi-source coupled evolution dispersion index will be high (for example, if the average wind speed observed by the meteorological station at a certain target scale is 8 m / s, and the average wind speed obtained by satellite remote sensing at the same scale is 10 m / s, such a large difference will lead to a high multi-source coupled evolution dispersion index for the target-scale coupled evolution vector, such as 0.6; while if the difference between the two is small, such as the average wind speed observed by the meteorological station being 9 m / s and the average wind speed obtained by satellite remote sensing being 9.5 m / s, the multi-source coupled evolution dispersion index may be 0.2).

[0021] Subsequently, the meteorological state prediction system determines the state prediction viewpoint and confidence characteristics of each target-scale coupled evolution vector based on the global trend evolution parameters, multi-source coupled evolution discrete index, and scale coupled evolution weight. The state prediction viewpoint is a predictive judgment of the future state of the target-scale coupled evolution vector based on the previously obtained parameters and indices. For example, for the target-scale coupled evolution vector of wind speed, if the global trend evolution parameters show an increasing trend in wind speed (e.g., the global trend evolution parameters indicate a high probability of wind speed increase in the next few observation periods), the multi-source coupled evolution discrete index indicates small differences in wind speed observations between different data sources (e.g., the multi-source coupled evolution discrete index is 0.1), and the scale coupled evolution weight also indicates that wind speed evolution at this scale has a significant impact on the overall meteorological state (e.g., the scale coupled evolution weight is 0.8), then the state prediction viewpoint might be that wind speed will continue to increase in the future. The confidence characteristics describe the credibility of this state prediction viewpoint. If the global trend evolution parameters are highly stable (e.g., the global trend evolution parameters fluctuate very little over multiple historical periods), the multi-source coupling evolution discrete index is low (indicating good consistency between data sources), and the scale coupling evolution weights are also reasonable, then the confidence feature of this state prediction view will be high (e.g., a confidence feature of 0.9 indicates high reliability of the prediction). Conversely, if these parameters and indicators have some uncertainties or contradictions, then the confidence feature will be low (e.g., if the global trend evolution parameters fluctuate greatly, the multi-source coupling evolution discrete index is 0.5, and the scale coupling evolution weights are not reasonable, the confidence feature may be 0.4).

[0022] Finally, the meteorological state prediction system determines the meteorological state prediction results of multi-source wind farm observation data by using the state prediction perspective of coupled evolution vectors at various set target scales. For example, considering the state prediction perspective of coupled evolution vectors at set target scales corresponding to multiple meteorological elements such as wind speed and wind direction, if the wind speed is predicted to increase (e.g., the wind speed is predicted to increase from the current 8 m / s to 12 m / s in the next 3 hours) and the wind direction is predicted to be stable (e.g., the wind direction will fluctuate within ±3 degrees in the next 3 hours), then the meteorological state prediction results will reflect that the wind farm will be in a meteorological state of increasing wind speed and stable wind direction in the future. This prediction result can provide important reference for the operation and management of wind farms. In the operation of wind farms, accurate prediction of wind speed and wind direction has a crucial impact on the power generation efficiency and safety of wind turbines. If an increase in wind speed is predicted, the wind farm can make adjustments and maintenance work on the wind turbines in advance to ensure safe and efficient operation under high wind speed conditions; if the wind direction is stable, it also helps to optimize management decisions such as the layout and orientation of wind turbines.

[0023] Furthermore, throughout the process, the meteorological state prediction system fully leverages the advantages of different data sources when assimilating multi-source wind field observation data. For example, meteorological station data features high precision and high spatiotemporal resolution, accurately reflecting local meteorological conditions; wind profiler radar data provides vertical wind speed and direction information, crucial for understanding the vertical structure of the wind field; and satellite remote sensing data covers a wide area, providing macroscopic meteorological information. Through the comprehensive processing of these multi-source data, the meteorological state prediction system can more comprehensively reflect wind field information.

[0024] Regarding coupled large eddy models, the meteorological state prediction system has undergone specific optimizations. For the mesoscale-to-microscale transition scheme, for example, by adjusting some parameter settings during the transition from mesoscale to microscale models, the model can better adapt to the complex terrain and meteorological conditions of wind farms. Improvements have been made to the configuration of physical process parameterization schemes, such as optimizing the parameterization of physical processes like turbulent exchange and radiative transfer. Taking turbulent exchange as an example, more accurate parameterization of the turbulent exchange process can improve the model's ability to simulate energy and mass exchange within the wind farm, thereby generating meteorological information at resolutions of hundreds of meters or even higher. This high-resolution meteorological information is invaluable for the refined management of wind farms. For instance, within a wind farm, wind speed and direction may vary significantly in different areas due to the influence of terrain and turbine layout. Meteorological information at resolutions of hundreds of meters can accurately reflect these differences, helping wind farms optimize turbine site selection, layout, and operation strategies.

[0025] The meteorological state prediction system mines and analyzes processed multi-source data and large eddy model outputs to construct accurate meteorological state prediction models. In this process, the system performs in-depth analysis of various meteorological elements in the multi-source data and large eddy model outputs. For example, it analyzes the correlations between different meteorological elements, such as the potential correlation between wind speed and temperature (e.g., when the temperature rises by 5 degrees Celsius, wind speed may increase by 2-3 m / s), and in some cases, temperature changes can affect wind speed; it also analyzes the relationship between humidity and precipitation. By mining these relationships, the system can construct more accurate meteorological state prediction models. This model not only considers the changing trends of the meteorological elements themselves but also their interactions and influences, thereby improving the accuracy of meteorological state predictions.

[0026] Therefore, the meteorological condition prediction system, through a series of complex and orderly steps, from the acquisition of multi-source wind farm observation data to the determination of the final meteorological condition prediction results, and by combining multi-source data assimilation, large eddy model optimization and other technical means, effectively improves the accuracy of meteorological condition prediction under complex wind farm conditions, and provides more reliable meteorological support for the operation and management of wind farms.

[0027] In the above technical solution, the following is a detailed introduction to setting the target scale coupling evolution vector, scale coupling evolution weight, global trend evolution parameter, and multi-source coupling evolution discrete index based on the above content.

[0028] I. Define the target scale coupling evolution vector The concept of defining a target-scale coupled evolution vector is used to accurately describe the evolution of meteorological conditions. Taking wind speed as an example, we consider the evolution of wind speed over time at different spatial scales.

[0029] For example, when studying a wind farm area, the spatial scale can be divided into a near-surface small scale and a larger mesoscale. For the near-surface small scale, a wind speed evolution vector over time can be defined. For instance, over 10 consecutive time intervals (e.g., every 10 minutes), the observed wind speed values ​​are [3 m / s, 3.5 m / s, 4 m / s, 4.2 m / s, 4.5 m / s, 4.3 m / s, 4.8 m / s, 5 m / s, 5.2 m / s, 5.5 m / s]. This set of data constitutes a wind speed evolution vector over time at the near-surface small scale, representing the trend of wind speed change at this small scale.

[0030] Similarly, for the mesoscale range, within the same 10 time intervals, wind speed observations might be [2.5 m / s, 2.8 m / s, 3 m / s, 3.2 m / s, 3.5 m / s, 3.3 m / s, 3.6 m / s, 3.8 m / s, 4 m / s, 4.2 m / s], which represents the mesoscale wind speed evolution vector. These wind speed evolution vectors at different scales are examples of setting target-scale coupled evolution vectors. They comprehensively reflect the changing trends of wind speed over time at different scales, and target-scale coupled evolution vectors for other meteorological elements such as wind direction can be constructed in a similar manner.

[0031] II. Scale Coupling Evolution Weights The scale-coupled evolution weights reflect the relative importance of each scale-coupled evolution vector in the entire meteorological state evolution process.

[0032] Continuing with the wind speed example above, analysis of wind farm meteorological conditions reveals that near-surface small-scale wind speed changes have a significant impact on the overall meteorological conditions. For instance, based on previous identification and correlation analysis of state trend description events in target meteorological observation data, near-surface small-scale wind speed changes are more strongly correlated with key indicators such as power generation efficiency and stability of wind turbines within the wind farm. In this case, the weight of the scale coupling evolution vector related to the near-surface small scale will be relatively high. For example, the weight of the near-surface small-scale wind speed evolution vector can be set at 0.7, while the weight of the larger-scale (medium-scale) wind speed evolution vector is 0.3. This means that when comprehensively considering meteorological state evolution, near-surface small-scale wind speed evolution plays a more important role in the overall assessment, and the weight is determined based on the descriptive distribution label analysis of the target state trend description events. For example, if the frequency of an event that causes a continuous increase in wind speed at a small near-surface scale (such as the continuous increase in wind speed over 10 consecutive time intervals mentioned above) is relatively high (e.g., a frequency of 0.6), and this increase in wind speed is critical to the overall meteorological impact of the wind farm (such as on wind turbine power and stability), then the weight of the scale coupling evolution vector corresponding to this scale will increase accordingly.

[0033] III. Global Trend Evolution Parameters The global trend evolution parameter is a parameter obtained by comprehensively processing the scale coupling evolution weights of each set target scale coupling evolution vector of several different target meteorological observation data. It reflects the overall evolution trend of the set target scale coupling evolution vector in the entire wind farm area and under the consideration of multiple meteorological factors.

[0034] For example, consider two target meteorological observation data: wind speed and wind direction. For wind speed, there are near-surface small-scale and meso-scale coupled evolution vectors of the target, with scale coupling evolution weights of 0.7 and 0.3 respectively (as in the example above). For wind direction, there are also target scale coupled evolution vectors at different scales, for example, a weight of 0.6 at the small scale and 0.4 at the meso-scale. For example, the evolution vectors of wind speed and wind direction at different scales and their weights can be combined to consider the meteorological state evolution trend of the entire wind farm area.

[0035] If wind speed shows a rapid increasing trend at a small scale (as seen in the wind speed evolution vector at a small scale above), and wind direction shows a slight clockwise rotation trend at a small scale (for example, the clockwise rotation angles of wind direction over 10 time intervals are [2 degrees, 3 degrees, 4 degrees, 3 degrees, 5 degrees, 4 degrees, 6 degrees, 5 degrees, 7 degrees, 6 degrees]), a global trend evolution parameter can be obtained by comprehensively considering the weights of wind speed and wind direction and their evolution at their respective scales. This parameter can be understood as a comprehensive description that considers different meteorological elements, different scales, and their respective importance, representing the trend of the entire wind farm's meteorological state developing in a certain direction with an overall value or characteristic. For example, this global trend evolution parameter might indicate that the overall meteorological state of the wind farm is developing in a direction where wind speed increases and wind direction has a clockwise rotation trend.

[0036] IV. Discrete Indices of Multi-Source Coupling Evolution The multi-source coupling evolution discrete index measures the degree of dispersion of the coupling evolution vector at a set target scale across different data sources from the perspective of multi-source data.

[0037] Taking wind speed observations from two data sources, weather stations and satellite remote sensing, as an example. For instance, the wind speed values ​​observed by the weather station at a certain target scale (such as a small near-surface scale) at 10 observation times are [8 m / s, 8.5 m / s, 9 m / s, 9.2 m / s, 9.5 m / s, 9.3 m / s, 9.8 m / s, 10 m / s, 10.2 m / s, 10.5 m / s], while the wind speed values ​​obtained by satellite remote sensing at the same scale at the same 10 observation times are [9 m / s, 9.2 m / s, 9.5 m / s, 9.8 m / s, 10 m / s, 10.2 m / s, 10.5 m / s, 10.8 m / s, 11 m / s, 11.2 m / s].

[0038] It can be seen that there are certain differences between the two. The multi-source coupling evolution discrete index can be obtained by calculating how this difference is reflected in the coupled evolution vector at the set target scale. For example, the variance statistic of the difference between the observations from the two data sources can be calculated. If this difference is relatively large, it will result in a higher multi-source coupling evolution discrete index for the coupled evolution vector at the set target scale, such as 0.6. If the wind speed observed by the weather station is [9 m / s, 9.1 m / s, 9.2 m / s, 9.3 m / s, 9.4 m / s, 9.5 m / s, 9.6 m / s, 9.7 m / s, 9.8 m / s, 9.9 m / s], and the value obtained by satellite remote sensing is [9.2 m / s, 9.3 m / s, 9.4 m / s, 9.5 m / s, 9.6 m / s, 9.7 m / s, 9.8 m / s, 9.9 m / s, 10 m / s, 10.1 m / s], then the difference between the two is small, and the multi-source coupling evolution discrete index may be 0.2. This index reflects the degree of consistency of the coupling evolution vector of different data sources for the same set target scale, and is of great significance for evaluating the accuracy and reliability of meteorological state prediction.

[0039] In summary, this application's embodiments, by acquiring multi-source wind farm observation data and identifying state trend description events in the target meteorological observation data, can fully mine the information in the data and obtain detailed descriptive distribution labels, laying the foundation for subsequent analysis. Determining scale-coupled evolution weights based on these labels helps to accurately grasp the degree of influence of meteorological elements on the overall meteorological state at different scales. Obtaining global trend evolution parameters and multi-source coupled evolution discrete indices allows for comprehensive consideration of the overall characteristics of various meteorological elements and multi-source data, improving the comprehensiveness of meteorological state analysis. Determining state prediction viewpoints and confidence features based on these parameters ensures that the prediction results have both a clear direction and a credibility assessment. The final meteorological state prediction results are more accurate and can provide reliable support for wind farm operation and management, such as optimizing wind turbine layout, improving power generation efficiency, and ensuring the safe operation of wind turbines.

[0040] In some preferred embodiments, obtaining the global trend evolution parameters and multi-source coupling evolution discrete indices of each set target scale coupling evolution vector includes: performing physical process quantization mapping on the target meteorological observation data in the multi-source wind farm observation data to obtain the target physical process quantization features corresponding to the multi-source wind farm observation data; and obtaining the global trend evolution parameters and multi-source coupling evolution discrete indices corresponding to each set target scale coupling evolution vector through the target physical process quantization features.

[0041] Based on this preferred embodiment, the method further includes: acquiring several multi-source historical observation data, and identifying state trend description events in the historical meteorological observation data included in each multi-source historical observation data to obtain historical description distribution labels for several historical state trend description events in each historical meteorological observation data; determining the historical scale coupling evolution weight of at least one set target scale coupling evolution vector in each historical meteorological observation data based on the historical description distribution labels of the several historical state trend description events in each historical meteorological observation data; performing physical process quantization mapping on each historical meteorological observation data to obtain the historical physical process quantization features of each multi-source historical observation data; and determining the state evolution thermodynamic relationship network of each set target scale coupling evolution vector corresponding to the same historical physical process quantization feature by using the historical scale coupling evolution weights of each set target scale coupling evolution vector in historical meteorological observation data with the same historical physical process quantization features.

[0042] In the process of predicting the meteorological state of wind farms, this embodiment first focuses on obtaining the global trend evolution parameters and multi-source coupled evolution discrete indices of the coupled evolution vectors at various set target scales. The crucial first step is to perform physical process quantification mapping on the target meteorological observation data from multi-source wind farm observation data. For example, if the target meteorological observation data includes wind speed and wind direction data, for wind speed, the physical process quantification mapping may involve transforming wind speed values ​​at different times and spatial locations into a quantifiable feature according to certain physical rules. For instance, at different observation points in a wind farm, wind speed has different values ​​at different times. By considering the relationship between wind speed and physical factors such as air density and the influence of topography on airflow, the wind speed observation value is transformed into a quantitative feature that reflects the essence of its physical process. The same applies to wind direction; the influence of factors such as Earth's rotation and pressure gradient force on wind direction must be considered to obtain the target physical process quantification features corresponding to the multi-source wind farm observation data. These target physical process quantification features are an important basis for subsequently obtaining the global trend evolution parameters and multi-source coupled evolution discrete indices.

[0043] When obtaining global trend evolution parameters and multi-source coupled evolution discrete indices corresponding to the coupled evolution vectors at various target scales through the quantitative characteristics of target physical processes, taking wind speed as an example, the wind speed evolution vectors at different scales are comprehensively analyzed by integrating multi-source data within the framework of the quantitative characteristics of target physical processes. For instance, the quantitative characteristics of near-surface small-scale wind speeds over a specific time period are obtained from meteorological station data, while the quantitative characteristics of mesoscale wind speeds over a larger area over the same time period are obtained from satellite remote sensing data. By comprehensively considering these wind speed quantitative characteristics from different scales and data sources, the global trend evolution parameters are determined. This parameter reflects the comprehensive evolution trend of wind speed at different scales throughout the wind farm area. Similarly, for the multi-source coupled evolution discrete indices, based on the quantitative characteristics of target physical processes, the degree of dispersion of wind speed in the coupled evolution vectors at the target scales from different data sources is analyzed. For example, the quantitative characteristics of wind speed at a certain scale observed by a meteorological station may exhibit a relatively stable but slightly fluctuating state over a period of time, while the quantitative characteristics of wind speed at the same scale obtained from satellite remote sensing show larger fluctuations and some differences from the meteorological station data. This difference will be reflected in the multi-source coupled evolution discrete indices.

[0044] Based on this preferred embodiment, the technical solution also includes a series of related operations. Acquiring several multi-source historical observation data is an important step for further in-depth analysis. These multi-source historical observation data contain rich historical meteorological observation data. Identifying state trend description events for the historical meteorological observation data included in each multi-source historical observation data is similar to the operation on the current target meteorological observation data. For example, for historical wind speed data, situations such as continuous wind speed increase (e.g., significant increase in wind speed over multiple consecutive historical observation periods), stable wind speed (minimal fluctuation in wind speed over a long period), and decreasing wind speed are identified as different events; for historical wind direction data, large clockwise rotation, large counterclockwise rotation, and constant wind direction are identified as different events, thereby obtaining historical description distribution labels for several historical state trend description events in each historical meteorological observation data.

[0045] Based on the historical description distribution labels of several historical state trend description events in various historical meteorological observation data, the historical scale coupling evolution weight of at least one target-scale coupling evolution vector in each historical meteorological observation data is determined. Taking historical wind speed data as an example, if a certain state trend description event of wind speed at a near-surface small scale (such as a continuous increase in wind speed) occurs frequently during a certain historical period, and data such as the wind farm operation records at that time indicate that this small-scale wind speed change has a significant impact on the meteorological state of the entire wind farm, then the historical scale coupling evolution weight of the target-scale coupling evolution vector related to the near-surface small scale will be relatively high during this historical period. This weight determination process fully considers the frequency of occurrence of various state trend description events in historical data and their degree of impact on the overall meteorological state.

[0046] The physical process quantification mapping of various historical meteorological observation data is similar to the physical process quantification mapping of target meteorological observation data in current multi-source wind farm observation data, but it targets historical data. By considering the influence of various physical factors in historical periods, such as the topography, geomorphology, and atmospheric circulation characteristics on meteorological elements, historical meteorological observation data is transformed into historical physical process quantification characteristics. For example, in a specific climatic period in the past, the area where a wind farm is located was affected by special atmospheric circulation. When performing physical process quantification mapping of wind speed, it is necessary to consider the influence of this special circulation on wind speed, thereby obtaining accurate historical physical process quantification characteristics.

[0047] Finally, by using the historical scale coupling evolution weights of the coupling evolution vectors at each set target scale in historical meteorological observation data with the same historical physical process quantification characteristics, the state evolution thermodynamic relationship network corresponding to the coupling evolution vectors at each set target scale under the same historical physical process quantification characteristics is determined. Taking wind speed and wind direction as examples, for instance, under a specific historical physical process quantification characteristic (such as a specific climate model), wind speed has a higher historical scale coupling evolution weight at a near-surface small scale, and wind direction has a higher historical scale coupling evolution weight at a mesoscale. Therefore, when constructing the state evolution thermodynamic relationship network, the wind speed and wind direction evolution vectors at these two scales will occupy important positions in the network. This state evolution thermodynamic relationship network reflects the interrelationships between coupling evolution vectors at different set target scales under specific historical physical process quantification characteristics. This relationship is determined based on the scale coupling evolution weights in historical data and can provide important reference for current meteorological state prediction. For example, if historical data shows that under a specific state evolution thermodynamic relationship, subsequent meteorological conditions tend to develop in the direction of increasing wind speed and clockwise wind direction, then in current meteorological condition forecasting, if the current state evolution thermodynamic relationship is found to be similar to this historical situation, it can serve as an important reference factor to improve the accuracy of the forecast.

[0048] It is evident that by quantifying and mapping the target meteorological observation data from multi-source wind farm observations to obtain global trend evolution parameters and multi-source coupled evolution discrete indices, the physical information within the data can be extracted more deeply, improving the accuracy of meteorological state analysis. Utilizing multi-source historical observation data for multifaceted analysis, including identifying state trend description events, determining historical scale coupled evolution weights, and constructing a state evolution thermodynamic relationship network, fully leverages historical data to provide more reference for current meteorological state prediction. These operations help to more comprehensively and accurately grasp the evolutionary relationships of meteorological elements at different scales, improve the accuracy of meteorological state prediction, and thus provide more reliable meteorological support for wind farm operation and management.

[0049] In an alternative embodiment, determining the state prediction viewpoint and confidence features of each set target scale-coupled evolution vector using the global trend evolution parameters, multi-source coupled evolution discrete index, and scale-coupled evolution weights of each set target scale-coupled evolution vector includes: embedding features into the scale-coupled evolution weights of each set target scale-coupled evolution vector using the global trend evolution parameters and multi-source coupled evolution discrete indexes to obtain the scale-coupled evolution embedding features of each set target scale-coupled evolution vector; obtaining the vortex mode representation vectors corresponding to each set target scale-coupled evolution vector; and utilizing the global trend evolution parameters, multi-source coupled evolution discrete indexes, and scale-coupled evolution weights of each set target scale-coupled evolution vector. The state prediction viewpoints of each set target scale coupled evolution vector are determined by using the vortex mode characterization vector corresponding to the target scale coupled evolution vector and the scale coupled evolution embedding features of each set target scale coupled evolution vector; the confidence features of each state prediction viewpoint are determined by using the vortex mode characterization vector corresponding to the target scale coupled evolution vector and the scale coupled evolution embedding features of each set target scale coupled evolution vector; or, the confidence features of each state prediction viewpoint are determined by using the vortex mode characterization vector corresponding to each set target scale coupled evolution vector, the global trend evolution parameters of each set target scale coupled evolution vector, the multi-source coupled evolution discrete index, and the scale coupled evolution weight.

[0050] In the following steps, when the target meteorological observation data is wind field environmental sensing monitoring data, the step of determining the state prediction viewpoint of each set target scale coupling evolution vector by utilizing the vortex model characterization vector corresponding to the scale coupling evolution weight of each set target scale coupling evolution vector and the scale coupling evolution embedding feature of each set target scale coupling evolution vector includes: when the set target scale coupling evolution vector is a ground-scale environmental vector or a cumulus-scale boundary vector, determining the state prediction viewpoint of the scale coupling evolution embedding feature of the set target scale coupling evolution vector based on the first vortex model characterization vector and the scale coupling evolution embedding feature of the set target scale coupling evolution vector, wherein the first vortex model characterization vector is used to indicate that the state prediction viewpoint and the scale coupling evolution embedding feature have a first quantification relationship; when the scale coupling evolution weight of the set target scale coupling evolution vector is a cloud microphysical environment vector, determining the state prediction viewpoint of the scale coupling evolution embedding feature based on the second vortex model characterization vector and the set target scale coupling evolution weight of the set target scale coupling evolution vector. The scale coupling evolution embedding feature of the target scale coupling evolution vector is used to determine the state prediction viewpoint of the scale coupling evolution embedding feature of the set target scale coupling evolution vector. The second vortex mode characterization vector is used to indicate a second quantification relationship between the state prediction viewpoint and the scale coupling evolution embedding feature. When the scale coupling evolution weight of the set target scale coupling evolution vector is one of the following: road surface mode weight, gravity wave fracturing mode weight, the ratio of the first radiative transfer mode weight to the second radiative transfer mode weight, the third radiative transfer mode weight, and the heat diffusion mode weight, the state prediction viewpoint of the scale coupling evolution embedding feature of the set target scale coupling evolution vector is determined based on the third vortex mode characterization vector and the scale coupling evolution embedding feature of the set target scale coupling evolution vector. The third vortex mode characterization vector is used to indicate the correlation between the state prediction viewpoint and the prediction offset, where the prediction offset is the comparison result between the scale coupling evolution embedding feature and the state probability statistics map.

[0051] In this alternative embodiment, the key part first involves determining the state prediction viewpoint and the confidence features of each state prediction viewpoint by using the global trend evolution parameters, multi-source coupling evolution discrete index and scale coupling evolution weight of each set target scale coupling evolution vector.

[0052] By using the global trend evolution parameters and multi-source coupling evolution discrete index of the coupling evolution vectors at various target scales, feature embedding is performed on the scale coupling evolution weights of the coupling evolution vectors at various target scales, thus obtaining the scale coupling evolution embedding features of the coupling evolution vectors at various target scales. For example, taking wind speed as a meteorological element, in the meteorological observation of a wind farm, there are two target scale coupling evolution vectors for wind speed at near-surface small scale and meso scale. The global trend evolution parameters show that the wind speed has a certain trend of change overall (e.g., the wind speed generally shows an upward trend over a period of time, and the increase has a certain quantitative value, such as an average increase of 2 meters / second per hour). The multi-source coupling evolution discrete index reflects the degree of dispersion of wind speed observations between different data sources (e.g., meteorological station and satellite remote sensing data) (e.g., the difference between the wind speed value observed by the meteorological station and the wind speed value obtained by satellite remote sensing is between 1 and 3 meters / second at certain times). The scale coupling evolution weights indicate that the near-surface small-scale wind speed has a greater impact on the overall meteorological state (e.g., weight of 0.7), while the meso-scale wind speed has a relatively smaller impact (weight of 0.3). Based on these numerical values, feature embedding operations are performed to obtain their respective scale-coupled evolution embedded features. These embedded features are a more in-depth feature representation that integrates global trends, multi-source dispersion, and their own weights.

[0053] Obtain the vortex model representation vectors corresponding to the coupled evolution vectors at various target scales. For different target scale coupled evolution vectors, the vortex model representation vectors have different meanings and numerical characteristics. For example, for certain scale coupled evolution vectors reflecting the rotational characteristics of airflow in a wind farm, the vortex model representation vector may include quantitative characteristics such as the speed and direction of airflow rotation, as well as its interaction with the surrounding environment. For instance, in a specific area of ​​a wind farm, the vortex model representation vector shows that the airflow rotates at a certain height at a certain speed (e.g., 3 revolutions per minute), in a clockwise direction, and has a specific interaction with the surrounding environment (e.g., terrain, interference from other airflows).

[0054] By utilizing the corresponding vortex model characterization vectors and the scale coupling evolution embedding features of each target-scale coupled evolution vector, the state prediction perspective for each target-scale coupled evolution vector is determined. This is more specific when the target meteorological observation data is wind field environmental sensing monitoring data. When the target-scale coupled evolution vector is a surface-scale environmental vector or a cumulus-scale boundary vector, the state prediction perspective for the scale coupling evolution embedding features of the target-scale coupled evolution vector is determined based on the first vortex model characterization vector and the scale coupling evolution embedding features of the target-scale coupled evolution vector. Taking the surface-scale environmental vector as an example, the surface-scale environmental vector may include the influence of factors such as surface roughness and surface temperature on the wind field. The first vortex model representation vector indicates a first quantification relationship between the state prediction viewpoint and the scale-coupled evolution embedding features. For example, the first vortex model representation vector shows a specific quantification relationship related to ground roughness and surface temperature (e.g., for every 5-degree increase in surface temperature, the trend of wind speed change at the ground scale is related to a certain quantification of the vortex model). Combined with the scale-coupled evolution embedding features (which include comprehensive information such as the previously mentioned global trend, multi-source discretization, and weights), the state prediction viewpoint is determined. For example, the state prediction viewpoint might be that, in the future, due to the increase in surface temperature and the unchanged surface roughness, the wind speed at the ground scale will increase by a certain value (e.g., an increase of 3 m / s).

[0055] When the scale coupling evolution weight of the target scale coupling evolution vector is set to the cloud microphysical environment vector, the state prediction viewpoint of the scale coupling evolution embedding feature of the target scale coupling evolution vector is determined based on the second vortex model characterization vector and the scale coupling evolution embedding feature of the target scale coupling evolution vector. The cloud microphysical environment vector may involve the influence of cloud type, cloud droplet size distribution, etc., on the wind field. For example, the second vortex model characterization vector reflects the quantitative relationship between cloud droplet size distribution and vertical airflow motion in the wind field (e.g., for every 1 micrometer increase in the average cloud droplet radius, the change in vertical airflow velocity is quantitatively correlated with the vortex model). Combined with the scale coupling evolution embedding feature (which includes comprehensive information on the cloud microphysical environment vector in terms of global trend, multi-source discretization, and weights), the state prediction viewpoint is determined. For example, the state prediction viewpoint might be that due to changes in cloud droplet size, vertical airflow will strengthen, thus affecting the overall stability of the wind field.

[0056] When the scale coupling evolution weights of the target scale coupling evolution vector are set as one of the following: road surface mode weight, gravity wave fracturing mode weight, the ratio of the first radiative transfer mode weight to the second radiative transfer mode weight, the third radiative transfer mode weight, and the heat diffusion mode weight, the state prediction viewpoint of the scale coupling evolution embedding feature of the target scale coupling evolution vector is determined based on the third vortex mode characterization vector and the scale coupling evolution embedding feature of the target scale coupling evolution vector. Taking the road surface mode weight as an example, factors such as road surface roughness and temperature affect the wind field. The third vortex mode characterization vector is used to indicate the correlation between the state prediction viewpoint and the prediction offset, which is the comparison result between the scale coupling evolution embedding feature and the state probability statistics map. For example, if the scale coupling evolution embedding feature shows that the road surface temperature is high and the roughness is large, the state probability statistics map indicates that the wind speed is likely to decrease under these conditions. The third vortex mode characterization vector represents a correlation (such as the quantitative correlation between the probability of wind speed reduction and the vortex mode), thereby determining the state prediction viewpoint, such as predicting that the wind speed will decrease by a certain value (e.g., a decrease of 2 m / s).

[0057] In determining the confidence features of each state prediction viewpoint, the same vectors and features mentioned above are used. The confidence features of each state prediction viewpoint are determined by the corresponding vortex model representation vectors of each target scale coupled evolution vector and the scale coupling evolution embedding features of each target scale coupled evolution vector. Alternatively, the confidence features of each state prediction viewpoint can be determined based on the corresponding vortex model representation vectors of each target scale coupled evolution vector, the global trend evolution parameters of each target scale coupled evolution vector, the multi-source coupled evolution discrete index, and the scale coupling evolution weights. For example, for the previously mentioned state prediction viewpoint of wind speed at the ground scale (wind speed will increase by 3 m / s), if the relationship between the airflow rotation shown by the vortex model representation vector and the ground environment is very stable (e.g., the rotation speed fluctuation is minimal), the multi-source coupled evolution discrete index in the scale coupling evolution embedding features is low (indicating good consistency between data sources), the global trend evolution parameters also show a clear trend of increasing wind speed, and the scale coupling evolution weights are reasonable (the weights of the ground scale environment vector are reasonable), then the confidence feature of this state prediction viewpoint will be very high (e.g., a confidence level of 0.9). Conversely, if there are contradictions or uncertainties among these factors (such as the vortex model characterization vector showing unstable airflow rotation, high discrete index of multi-source coupling evolution, large fluctuation of global trend evolution parameters, etc.), the confidence characteristics will be low (e.g., confidence level of 0.3).

[0058] Thus, state prediction viewpoints and confidence features are determined through complex and orderly operations. Feature embedding operations fuse multiple key parameters to obtain a deeper feature representation. The introduction of vortex model representation vectors provides a basis closely related to meteorological and physical processes for determining state prediction viewpoints. In particular, the special processing of evolution vectors coupled to different types of target scales makes the state prediction viewpoints more closely aligned with actual meteorological conditions. Furthermore, the determination of confidence features integrates multiple factors, ensuring that the prediction results not only have clear viewpoints but also accurate credibility assessments. Overall, this significantly improves the accuracy and reliability of meteorological state predictions, providing strong support for meteorological analysis and decision-making related to wind farms.

[0059] In one technical approach, determining the confidence features of each state prediction viewpoint through the vortex mode representation vector corresponding to each set target scale coupled evolution vector and the scale coupled evolution embedding feature of each set target scale coupled evolution vector includes: obtaining a state probability statistical graph vector; performing a fully connected processing on the state probability statistical graph vector through the vortex mode representation vector corresponding to each set target scale coupled evolution vector using the scale coupled evolution embedding feature of each set target scale coupled evolution vector to obtain fully connected feature values ​​corresponding to each set target scale coupled evolution vector; and determining the confidence features corresponding to each set target scale coupled evolution vector through the fully connected feature values ​​and the vortex mode representation vector corresponding to each set target scale coupled evolution vector.

[0060] In another technical approach, determining the confidence features of each state prediction viewpoint based on the corresponding vortex mode characterization vector of each set target scale coupled evolution vector, the global trend evolution parameters of each set target scale coupled evolution vector, the multi-source coupled evolution discrete index, and the scale coupled evolution weight includes: determining the state heatmap corresponding to each set target scale coupled evolution vector through the global trend evolution parameters of each set target scale coupled evolution vector and the multi-source coupled evolution discrete index; performing fully connected processing on the state heatmap based on the scale coupled evolution weight of each set target scale coupled evolution vector and the vortex mode characterization vector corresponding to each set target scale coupled evolution vector to obtain the fully connected feature value corresponding to each set target scale coupled evolution vector; and determining the confidence features corresponding to each set target scale coupled evolution vector through the fully connected feature value corresponding to each set target scale coupled evolution vector and the vortex mode characterization vector corresponding to each set target scale coupled evolution vector.

[0061] In this technical solution, there are two technical approaches to determine the confidence features corresponding to the coupling evolution vectors of each set target scale.

[0062] First, the first technical approach is explained. Under this approach, the confidence features of each state prediction viewpoint are determined by the corresponding vortex mode characterization vector of each set target scale coupled evolution vector and the scale coupled evolution embedding feature of each set target scale coupled evolution vector.

[0063] This process begins by obtaining the state probability statistical vector. The state probability statistical vector is a quantitative representation of the distribution of various possibilities for meteorological states, encompassing the probability information of meteorological elements within different value ranges or states. For example, for the meteorological element of wind speed, the state probability statistical vector might include the probability of wind speed occurring in different speed ranges (e.g., 0-5 m / s, 5-10 m / s, etc.). In meteorological observations at a wind farm, through comprehensive analysis of historical data and current multi-source observation data, it might be concluded that the probability of wind speed in the 0-5 m / s range is 0.3, in the 5-10 m / s range is 0.5, and in the range above 10 m / s is 0.2, and so on. This probability information constitutes the state probability statistical vector of wind speed.

[0064] Next, the state probability statistical graph vector is fully connected using the scale coupling evolution embedding features of each target scale coupling evolution vector and the vortex mode representation vector corresponding to each target scale coupling evolution vector. This yields the fully connected feature values ​​corresponding to each target scale coupling evolution vector. Fully connected processing is a computational process that comprehensively considers multiple factors. For example, taking the target scale coupling evolution vector corresponding to wind speed as an example, the scale coupling evolution embedding features may contain information such as the correlation between wind speed and other meteorological elements (such as wind direction and temperature) at different spatial scales (e.g., near-surface small and medium scales). The vortex mode representation vector may reflect information such as wind speed-related airflow rotation patterns. When fully connected processing is performed on the state probability statistical graph vector of wind speed, the scale correlation information in the scale coupling evolution embedding features, the airflow rotation pattern information in the vortex mode representation vector, and the wind speed probability distribution information in the state probability statistical graph vector are comprehensively considered. For example, in some application scenarios, the fully connected feature value of the target scale coupling evolution vector corresponding to wind speed is 0.7.

[0065] Finally, the confidence features corresponding to the coupled evolution vectors at each target scale are determined using the fully connected eigenvalues ​​and vortex pattern representation vectors. Continuing with wind speed as an example, if the vortex pattern representation vector shows a relatively stable airflow rotation pattern for wind speed, and the fully connected eigenvalue is 0.7, this likely indicates a high level of confidence in the wind speed prediction, thus determining a high confidence feature (e.g., 0.8) for the coupled evolution vector at the target scale corresponding to wind speed. This is because the fully connected eigenvalue integrates various wind speed-related information, and the stability of the vortex pattern representation vector further enhances this confidence.

[0066] The second technical approach will be described next. Under this approach, the confidence features of each state prediction viewpoint are determined based on the corresponding vortex mode characterization vector of each set target scale coupling evolution vector, the global trend evolution parameters of each set target scale coupling evolution vector, the multi-source coupling evolution discrete index, and the scale coupling evolution weight.

[0067] First, the state heatmap corresponding to each target-scale coupled evolution vector is determined by using the global trend evolution parameters and multi-source coupled evolution discrete indices. The global trend evolution parameters reflect the overall evolution trend of the target-scale coupled evolution vectors, while the multi-source coupled evolution discrete indices reflect the degree of dispersion among multi-source data. Taking wind direction as an example, the global trend evolution parameters might indicate that the wind direction tends to rotate clockwise over a period of time, and the magnitude of the rotation has a certain quantifiable value (e.g., an average clockwise rotation of 5 degrees per hour). The multi-source coupled evolution discrete indices might show the differences in wind direction observations from different data sources (e.g., weather station and satellite remote sensing data). For example, the average deviation angle between the wind direction observed by the weather station and the wind direction obtained by satellite remote sensing during certain periods might be 10 degrees; this difference is reflected in the multi-source coupled evolution discrete indices. Based on this information, a state heatmap is constructed. The state heatmap is a graphical way to intuitively represent the distribution and trend of wind direction. Different areas in the map may represent different wind direction ranges, and color intensity or other markers can indicate the probability or trend strength of the wind direction in that area.

[0068] Then, based on the scale coupling evolution weights of each target scale coupling evolution vector and the corresponding vortex model representation vector, the state heatmap is fully connected to obtain the fully connected eigenvalues ​​corresponding to each target scale coupling evolution vector. Taking the target scale coupling evolution vector corresponding to wind direction as an example, the scale coupling evolution weights may represent the degree of influence of wind direction on the overall meteorological state at different spatial scales. For example, the scale coupling evolution weight of wind direction at a small near-surface scale is 0.6, which means that the influence of wind direction at a small near-surface scale on the overall meteorological state is relatively large. The vortex model representation vector may contain information about the relationship between wind direction and factors such as airflow rotation and topography. When performing fully connected processing on the state heatmap of wind direction, the scale influence information in the scale coupling evolution weights, the wind direction correlation information in the vortex model representation vectors, and the wind direction state distribution and trend information in the state heatmap are comprehensively considered. For example, after calculation, the fully connected eigenvalue of the target scale coupling evolution vector corresponding to wind direction is 0.6.

[0069] Finally, the confidence features corresponding to each target-scale coupled evolution vector are determined by using the fully connected eigenvalues ​​and vortex model representation vectors corresponding to each target-scale coupled evolution vector. For example, if the vortex model representation vector shows a complex relationship between wind direction and surrounding environmental factors, and the fully connected eigenvalue is 0.6, this may indicate that the prediction confidence of wind direction is at a moderate level, thus determining the confidence feature of the target-scale coupled evolution vector corresponding to wind direction to be around 0.6. This is because the fully connected eigenvalue integrates multiple wind-direction-related information, while the complex relationship reflected by the vortex model representation vector affects the prediction confidence to some extent.

[0070] This approach, employing two distinct technical approaches to determine confidence features, provides a more comprehensive and flexible method for evaluating the reliability of weather state forecasts. Both the first approach, based on factors such as scale-coupled evolution embedding features, and the second approach, incorporating global trend evolution parameters, fully consider the diverse characteristics of meteorological elements through multi-step, complex calculations and information fusion. This contributes to improving the accuracy and reliability of weather state forecasts, providing a more scientific basis for decision-making in areas such as wind farm operation and management.

[0071] Based on either of the above two technical approaches, determining the confidence feature of each set target scale coupled evolution vector through the fully connected eigenvalues ​​corresponding to each set target scale coupled evolution vector and the vortex mode representation vectors corresponding to each set target scale coupled evolution vector includes: when the vortex mode representation vector corresponding to the set target scale coupled evolution vector is a first vortex mode representation vector, determining the fully connected eigenvalue as the confidence feature corresponding to the set target scale coupled evolution vector; when the vortex mode representation vector corresponding to the set target scale coupled evolution vector is a second vortex mode representation vector, determining the difference between the first set value and the fully connected eigenvalue as the confidence feature corresponding to the set target scale coupled evolution vector; when the vortex mode representation vector corresponding to the set target scale coupled evolution vector is a third vortex mode representation vector, determining the weighted result of the second set value and the fully connected eigenvalue as the confidence feature corresponding to the set target scale coupled evolution vector.

[0072] In either of the two technical approaches mentioned above, the core part of determining the confidence features of the coupling evolution vectors at each set target scale by using the fully connected eigenvalues ​​corresponding to the coupling evolution vectors at each set target scale and the vortex mode characterization vectors corresponding to the coupling evolution vectors at each set target scale needs to be elaborated in detail from multiple perspectives.

[0073] First, let's examine the case where the vortex model representation vector corresponding to the target-scale coupled evolution vector is the first vortex model representation vector. In the meteorological state prediction system, under this setting, the fully connected eigenvalues ​​are directly determined as the confidence features corresponding to the target-scale coupled evolution vector. To better understand this process, we will use the wind speed-related target-scale coupled evolution vector as an example for detailed explanation.

[0074] For example, in the meteorological observation data of a wind farm, after a series of preliminary processing and analysis, wind speed-related fully connected eigenvalues ​​were obtained. A fully connected eigenvalue is a numerical value that comprehensively reflects the characteristics of wind speed in multiple aspects; its formation involves the fusion of various information from numerous previous steps. For example, the evolution of wind speed at different spatial scales, such as the specific values ​​of wind speed at the near-surface small scale at different times, for example, the wind speeds in 10 consecutive observation periods are [3 m / s, 3.2 m / s, 3.5 m / s, 3.3 m / s, 3.6 m / s, 3.8 m / s, 4 m / s, 4.2 m / s, 4.5 m / s, 4.8 m / s], these values ​​reflect the fluctuation of wind speed at the near-surface small scale; at the same time, there are similar observed values ​​of wind speed at the mesoscale, such as [2.5 m / s, 2.8 m / s, 3 m / s, 3.2 m / s, 3.5 m / s, 3.3 m / s, 3.6 m / s, 3.8 m / s, 4 m / s, 4.2 m / s]. In addition to wind speed values ​​at different scales, the fully connected eigenvalue also incorporates information from multiple data sources, such as wind speed data directly observed by weather stations and wind speed data acquired by satellite remote sensing. Weather station data, due to its high precision and high spatiotemporal resolution, may provide more localized but more accurate wind speed information, while satellite remote sensing data covers a wider area. There are certain differences and connections between the two. After comprehensive analysis of these data, a comprehensive fully connected eigenvalue is obtained, for example, this value is 0.8.

[0075] When the vortex model representation vector of the coupled evolution vector at the target scale corresponding to wind speed is the first vortex model representation vector, then the confidence feature corresponding to this wind speed vector is directly determined to be 0.8. This result has profound meteorological significance and data logic. Under this specific vortex model representation, the fully connected eigenvalue can fully represent the confidence feature. This means that, considering various influencing factors of wind speed, including wind speed evolution trends at different scales and the consistency of multi-source data, these factors highly match the comprehensive situation represented by the fully connected eigenvalue. From a meteorological perspective, this may indicate that under the current meteorological environment, the wind speed change trend is relatively stable, with fewer interfering factors, thus its prediction reliability is high. For example, if the wind farm is located in a relatively flat and open area, without significant topographical undulations or large obstacles interfering with airflow, then wind speed changes are relatively easier to predict. The comprehensive information represented by the fully connected eigenvalue can well reflect this relatively simple wind speed change pattern, resulting in a confidence feature of 0.8, indicating that wind speed-related predictions have high reliability.

[0076] Next, the situation becomes more complex when the vortex model representation vector corresponding to the target-scale coupling evolution vector is set as the second vortex model representation vector. In this case, the difference between the first setpoint and the fully connected eigenvalue is determined as the confidence feature corresponding to the target-scale coupling evolution vector. To better understand this mechanism, we will use wind direction in a wind field as an example to illustrate the target-scale coupling evolution vector.

[0077] For example, in wind direction correlation analysis, the first setpoint is set to 1. This first setpoint is not arbitrarily set; it is likely a standard reference value related to wind direction derived from long-term meteorological research, theoretical models, or statistical analysis of a large amount of historical data. This standard reference value represents the comprehensive numerical value of wind direction-related factors under certain ideal or standard conditions. For example, under an ideal atmospheric circulation model, without interfering factors such as local topography or thermal differences, wind direction may change according to a relatively stable pattern. This 1 might be a quantitative representation of the comprehensive wind direction-related factors under this ideal condition.

[0078] Continuing, for example, the fully connected eigenvalue related to wind direction is 0.6. The process of obtaining the fully connected eigenvalue also involves the comprehensive analysis of wind direction observation data at different scales and multi-source data. For instance, wind direction observations at different altitudes, from near-surface to higher atmospheres, may exhibit different patterns of change. Simultaneously, wind direction observation data from different data sources (such as weather station anemometers, Doppler radar, etc.) are also comprehensively considered. In this case, the confidence feature is 1 - 0.6 = 0.4.

[0079] This calculation method shows that the confidence level of wind direction predictions is determined by the difference between the first setpoint and the fully connected feature value. The greater the difference, the lower the confidence level. This is reasonable in meteorology because when the fully connected feature value differs significantly from the first setpoint representing the ideal state, it indicates that the wind direction prediction deviates considerably from the ideal or standard situation, possibly due to the influence of complex local factors such as terrain undulations or uneven thermal distribution. For example, if a wind farm is located in a mountainous area, the obstruction of mountains and the airflow channel effect of valleys can make the wind direction complex and variable, differing greatly from the ideal wind direction pattern. Therefore, the gap between the fully connected feature value and the first setpoint will be large, resulting in a low confidence level and indicating low reliability of the wind direction prediction.

[0080] Finally, when the vortex model characterization vector corresponding to the target scale coupling evolution vector is set as the third vortex model characterization vector, the weighted result of the second set value and the fully connected eigenvalue is determined as the confidence feature corresponding to the target scale coupling evolution vector. This will be explained in detail using the temperature in a wind field as an example to illustrate this point.

[0081] For example, the second setpoint is 0.5, which is also set based on specific meteorological significance and data. It may be related to certain baseline or expected characteristics of temperature under specific environmental conditions. For example, under specific seasons, geographical locations, and climatic backgrounds, there exists an expected baseline value related to temperature, which takes into account the basic influence of macroscopic factors such as solar radiation and atmospheric circulation on temperature.

[0082] For example, the temperature-related fully connected eigenvalue is 0.7. This fully connected eigenvalue is derived from multiple temperature observation data and analyses. These include temperature observations on different land surface types (such as grassland, sand, and water surfaces) and temperature variations at different times of day and night. This data comes from multiple data sources, including thermometers from ground weather stations and thermal infrared data from satellite remote sensing. Using a weighted calculation (e.g., a weighting of 1:1), the confidence eigenvalue is (0.5 + 0.7) / 2 = 0.6.

[0083] This weighted result reflects a comprehensive consideration of the specific factors represented by the second setpoint and the actual overall situation represented by the fully connected eigenvalue. The confidence metric of 0.6 indicates that the reliability of the temperature prediction is at an intermediate level. This means that in this vortex model, temperature prediction is influenced by multiple factors (represented by the fully connected eigenvalue) and a specific baseline factor (represented by the second setpoint). For example, in a wind farm near a large body of water, the water has a moderating effect on the surrounding temperature, but it is also affected by atmospheric circulation and local meteorological conditions. This means that the temperature prediction does not perfectly match the ideal baseline situation (represented by the second setpoint), nor is it entirely dominated by actual observation data (represented by the fully connected eigenvalue), but rather it is a combined result of both. Therefore, the confidence metric is 0.6, indicating that the reliability of the temperature prediction is at an intermediate level.

[0084] This method of determining confidence features based on the characterization vectors of different vortex modes represents a more detailed quantification of the reliability of meteorological state prediction results, building upon previous technical solutions. Previous solutions already involved multifaceted processing of meteorological observation data, such as identifying state trend description events in target meteorological observation data from multi-source wind farm observations, determining scale coupling evolution weights, and obtaining global trend evolution parameters and multi-source coupling evolution discrete indices.

[0085] In the process of identifying events in the state trend description, various meteorological elements (such as wind speed, wind direction, and temperature) are classified into different events based on their specific performance in the observation data. For example, for wind speed, situations such as a continuous increase in wind speed (e.g., wind speed increasing by more than 1.5 m / s in each of the previous 5 observation periods), stable wind speed (wind speed fluctuating within ±1 m / s in 10 observation periods), and decreased wind speed (wind speed decreasing by more than 1 m / s in each of the previous 3 observation periods) are identified as different events; for wind direction, situations such as a large clockwise rotation (wind direction rotating by more than 15 degrees in each of the previous 6 observation periods) and a large counterclockwise rotation (wind direction rotating by more than 15 degrees in each of the previous 8 observation periods) are identified as different events. During the observation period, events such as rotation angles exceeding 10 degrees each time and stable wind direction (wind direction fluctuations within ±5 degrees over 12 observation periods) were classified as different events. For temperature, rapid temperature rises (temperature increases of more than 3 degrees each time over 4 consecutive observation periods), slow temperature rises (temperature increases of 1-2 degrees each time over 8 consecutive observation periods), stable temperature (temperature fluctuations within ±1 degree over 10 observation periods), and temperature drops (temperature drops of more than 2 degrees each time over 3 consecutive observation periods) were identified as different events. This detailed event identification allows for a better grasp of the changing trends and characteristics of meteorological elements.

[0086] In determining the scale coupling evolution weights, the influence of meteorological elements on the overall meteorological state at different spatial scales is considered. For example, for wind speed, changes in wind speed at a small near-surface scale may have a significant impact on the meteorological state of a local area within the wind farm, while changes in wind speed over a larger area at a mesoscale have a significant impact on the macro-meteorological environment of the entire wind farm. Based on the analysis of these influence levels, the scale coupling evolution weights for wind speed at different scales are determined. For example, the scale coupling evolution weight for near-surface small-scale wind speed is 0.7, and the scale coupling evolution weight for mesoscale wind speed is 0.3, indicating that near-surface small-scale wind speed has a higher relative importance in the evolution of the overall meteorological state. Similarly, similar scale coupling evolution weights are determined for meteorological elements such as wind direction and temperature.

[0087] Obtaining global trend evolution parameters and multi-source coupled evolution discrete indices were also crucial aspects of previous technical solutions. Global trend evolution parameters are obtained by comprehensively processing the scale coupling evolution weights of each target-scale coupled evolution vector from several different meteorological observation data. They reflect the overall evolution trend of the target-scale coupled evolution vector across the entire wind farm area, considering multiple meteorological factors. For example, by comprehensively considering the scale coupling evolution weights of wind speed and wind direction, and their evolution trends at different scales, a global trend evolution parameter is obtained that reflects the comprehensive evolution trend of wind speed and wind direction throughout the entire wind farm. Multi-source coupled evolution discrete indices, on the other hand, measure the degree of dispersion of the target-scale coupled evolution vector across different data sources from the perspective of multi-source data. For example, there may be some differences between wind speed data observed at meteorological stations and wind speed data obtained from satellite remote sensing; multi-source coupled evolution discrete indices can reflect the manifestation of this difference in the target-scale coupled evolution vector. If the average wind speed observed by a weather station at a certain target scale is 8 m / s, and the average wind speed obtained by satellite remote sensing at the same scale is 10 m / s, this large difference will result in a high multi-source coupling evolution dispersion index of the coupling evolution vector at that target scale, such as 0.6. However, if the difference between the two is small, such as the average wind speed observed by the weather station being 9 m / s and the average wind speed obtained by satellite remote sensing being 9.5 m / s, the multi-source coupling evolution dispersion index may be 0.2.

[0088] Building upon this foundation, the confidence profile is determined by the relationship between different vortex model representation vectors and fully connected eigenvalues, further refining the accuracy assessment of meteorological condition predictions. Different vortex model representation vectors correspond to different meteorological elements or specific relationships under different meteorological conditions. This targeted calculation method can more accurately reflect the reliability of meteorological predictions under different circumstances, thus providing a more comprehensive and accurate assessment basis for wind farm meteorological condition prediction results. For example, accurate assessment of the reliability of meteorological predictions is crucial in the operation and management of wind farms. If the confidence profile of wind speed prediction is high, the wind farm can more confidently adjust the operating parameters of the wind turbines, such as the blade pitch angle, to improve power generation efficiency. If the confidence profile of wind direction prediction is low, the wind farm needs to treat the wind direction prediction results more cautiously, possibly requiring additional observation methods or adopting conservative operating strategies to avoid damage to the wind turbines due to sudden changes in wind direction. For temperature prediction confidence profiles, it can help wind farms make more rational decisions regarding equipment maintenance and operational safety, such as taking preventative measures against freezing in advance when temperatures are low.

[0089] This design, by employing different methods to determine confidence features for different vortex model characterization vectors, makes the reliability assessment of meteorological state predictions more accurate. Methods such as directly using fully connected eigenvalues, calculating difference values, or weighted results are applicable to different meteorological elements and states, better reflecting the reliability of prediction results under various conditions. This helps wind farm operators and other relevant personnel to more accurately judge the reliability of meteorological prediction results, thereby making more reasonable decisions and improving the efficiency and safety of wind farm operation and management. Furthermore, this accurate reliability assessment is based on the in-depth processing of meteorological observation data from multiple aspects. The entire technical solution forms a complete system, from the comprehensive utilization of multi-source data to the detailed analysis of meteorological elements, and finally to the reliability assessment, comprehensively improving the accuracy and practicality of meteorological state predictions.

[0090] In some alternative embodiments, the step of identifying state trend description events in the target meteorological observation data included in the multi-source wind farm observation data to obtain description distribution labels for several target state trend description events in the target meteorological observation data includes: extracting data from the multi-source wind farm observation data to obtain initial target meteorological observation data in the multi-source wind farm observation data; determining optimization indication features of the initial target meteorological observation data when the initial target meteorological observation data meets optimization conditions; optimizing the initial target meteorological observation data using the optimization indication features to obtain optimized target meteorological observation data; and identifying state trend description events in the optimized target meteorological observation data to obtain description distribution labels for several target state trend description events.

[0091] In this alternative embodiment, an exemplary description of the process of identifying state trend description events in target meteorological observation data included in multi-source wind farm observation data to obtain description distribution labels for several target state trend description events in the target meteorological observation data is provided below.

[0092] The first step involves data extraction from multi-source wind farm observation data to obtain the initial target meteorological observation data. Multi-source wind farm observation data comes from a wide range of sources, including weather stations, wind profiler radar, and satellite remote sensing. From these rich data sources, the extracted initial target meteorological observation data covers various meteorological elements, such as wind speed, wind direction, temperature, and humidity. Taking wind speed data as an example, hourly wind speed observations may be obtained from weather stations, while satellite remote sensing data may reveal a large-scale wind speed distribution. These data from different sources collectively constitute the wind speed component of the initial target meteorological observation data.

[0093] After obtaining the initial target meteorological observation data, it is necessary to determine when the initial target meteorological observation data meets the optimization conditions. These optimization conditions are set based on the characteristics of the meteorological data and the needs of subsequent processing. For example, for wind speed data, the optimization conditions may be related to the accuracy, completeness, and correlation with other meteorological elements. If the wind speed data has a large number of missing values ​​within a certain time period, or if its variation range exceeds the normal meteorological fluctuation range (for example, in a stable meteorological environment, the wind speed suddenly reaches a maximum or minimum value and differs significantly from data from surrounding time periods), then the optimization conditions may not be met. However, if the time series of wind speed data is complete, its values ​​are within a reasonable meteorological fluctuation range, and there is a reasonable correlation with meteorological elements such as wind direction and temperature during the same period (e.g., wind speed increases or decreases regularly as temperature rises), then the optimization conditions can be determined to be met.

[0094] Once the initial target meteorological observation data is determined to meet the optimization conditions, the optimized indicative characteristics of the initial target meteorological observation data need to be identified. For wind speed data, optimized indicative characteristics may include statistical features such as the mean, standard deviation, and extreme values ​​of wind speed, as well as the correlation coefficient between wind speed and other meteorological elements. For example, if the mean wind speed is 5 m / s, the standard deviation is 1 m / s, and the correlation coefficient with temperature is 0.6 (indicating a positive correlation between wind speed and temperature), these values ​​constitute the optimized indicative characteristics of wind speed data for that time period. For wind direction data, optimized indicative characteristics may include the dominant wind direction and the frequency of wind direction changes. For example, if the dominant wind direction in a certain area is southeast, and the frequency of wind direction changes per hour is low during the observation period, these are the optimized indicative characteristics of wind direction data.

[0095] The initial target meteorological observation data is optimized by refining the indicator features to obtain optimized target meteorological observation data. Taking wind speed data as an example, if the standard deviation in the optimized indicator features is too large, it indicates that the wind speed data fluctuates drastically, possibly due to outliers or local interference factors. In this case, the data can be smoothed based on the mean and standard deviation to remove outliers that deviate significantly from the mean, thus obtaining optimized wind speed data. For wind direction data, if the frequency of wind direction changes is found to be too high and inconsistent with normal meteorological patterns, it may be due to data acquisition errors or factors such as complex local terrain. In this case, the wind direction data can be corrected based on the dominant direction to make the wind direction data more consistent with actual meteorological patterns, resulting in optimized wind direction data.

[0096] Finally, the optimized target meteorological observation data was subjected to state trend description event identification, resulting in description distribution labels for several target state trend description events. For wind speed data, wind speed continuously increasing, wind speed stabilizing, and wind speed decreasing were identified as different events. For example, in the optimized wind speed data, if the wind speed increases by more than 1.5 m / s in each of the five consecutive observation periods compared to the previous period, it can be identified as a wind speed continuously increasing event; if the wind speed fluctuates within ±1 m / s in 10 observation periods, it is identified as a wind speed stabilizing event; and if the wind speed decreases by more than 1 m / s in each of the three consecutive observation periods compared to the previous period, it is identified as a wind speed decreasing event. For wind direction data, significant clockwise rotation (e.g., more than 15 degrees per rotation in each of the six consecutive observation periods), significant counterclockwise rotation (more than 10 degrees per rotation in each of the eight consecutive observation periods), and stable wind direction (within ±5 degrees of fluctuation in the wind direction in the twelve observation periods) were identified as different events. For temperature data, rapid temperature increases (more than 3 degrees Celsius each time over 4 consecutive observation periods), slow temperature increases (1-2 degrees Celsius each time over 8 consecutive observation periods), stable temperature (temperature fluctuations within ±1 degree Celsius over 10 observation periods), and temperature decreases (more than 2 degrees Celsius each time over 3 consecutive observation periods) are identified as different events. Through this identification process, the distribution of each event in the optimized target meteorological observation data is obtained. These distributions constitute descriptive distribution labels for several target state trend events. For example, within a certain time period, the descriptive distribution label for a sustained increase in wind speed might show a frequency of 0.2 (i.e., sustained wind speed increases occur in 20% of the observation periods), the descriptive distribution label for a stable wind speed event might show a frequency of 0.6, and the descriptive distribution label for a decrease in wind speed might show a frequency of 0.2.

[0097] This optimization of initial target meteorological observation data improves data quality, leading to more accurate identification of subsequent state trend description events. The process of determining optimization conditions, optimizing indicator features, and performing optimization effectively removes outliers and data that do not conform to meteorological patterns, thereby more accurately identifying state trend description events for meteorological elements. This contributes to providing a more reliable data foundation for the entire meteorological state prediction system, improving the accuracy of meteorological state predictions, and ultimately providing more accurate meteorological information support for wind farm operation and management.

[0098] In an extended technical solution, when the target meteorological observation data is wind field environmental sensing and monitoring data, the set target scale coupling evolution vector includes at least one of the following: road surface pattern weight, gravity wave fracturing pattern weight, third radiative transfer pattern weight, wind speed attention weight, heat diffusion pattern weight, ground-scale environmental vector, cloud microphysical environment vector, the ratio of the first radiative transfer pattern weight to the second radiative transfer pattern weight, three-dimensional wind field pattern weight, wind field noise weight, wind field remote sensing pattern weight, and multi-mode wind field monitoring change weight. Correspondingly, the determination of at least one scale coupling evolution weight of the set target scale coupling evolution vector through the description distribution labels of the plurality of target state trend description events includes at least one of the following: (1) Based on the description distribution labels of several first-state trend description events, determine the scale coupling evolution weight of at least one of the road surface mode weight and gravity wave fracturing mode weight; (2) Based on the description distribution labels of several fourth-state trend description events, determine the scale coupling evolution weight of at least one of the following: the weight of the third radiative transfer mode, the weight of wind speed attention, the weight of heat diffusion mode, and the weight of at least one of the ground-scale environmental vectors. (3) Based on the description distribution labels of several second-state trend description events, several third-state trend description events, and several fourth-state trend description events, determine at least one of the scale coupling evolution weights in the ratio of cloud microphysical environment vector, first radiative transfer mode weight, and second radiative transfer mode weight. (4) Determine the scale coupling evolution weight of the three-dimensional wind field model weight based on the description distribution labels of several third-state trend description events; (5) Based on the description distribution label of the fifth state trend description event, determine the scale coupling evolution weight of at least one of the wind field noise weight, wind field remote sensing model weight and multi-mode wind field monitoring change weight.

[0099] In this extended technical solution, when the target meteorological observation data is wind field environmental sensing and monitoring data, the target scale coupling evolution vector is set to include a variety of different weights and vector types, which reflects the characteristics and influencing factors of the wind field meteorological state in many aspects.

[0100] For the road surface pattern weights and gravity wave fracturing pattern weights in the target scale-coupled evolution vector, their scale-coupled evolution weights are determined by the description distribution labels of several first-state trend description events. Taking the road surface pattern weight as an example, certain specific state trend description events in the wind field environmental sensing and monitoring data have a crucial impact on its weight. For instance, in the wind field environmental sensing and monitoring data, if the first-state trend description events include descriptions of the impact of road surface roughness changes on near-surface wind speed and direction, and the description distribution labels show that the frequency of road surface roughness increase events is high within a certain time period, and the associated near-surface wind speed changes are large (for example, in 10 consecutive observation periods, the near-surface wind speed decreases by an average of 1.5 m / s after each increase in road surface roughness), and the wind direction also changes significantly (e.g., the wind direction deviation angle reaches an average of 10 degrees), this indicates that the road surface condition has a significant impact on the wind field, and therefore the scale-coupled evolution weight of the road surface pattern will increase accordingly. The same applies to the weighting of gravity wave breaking modes. If the description of the impact of gravity wave breaking on the vertical structure of the wind field in the first state trend description event shows that the frequency of gravity wave breaking events increases within a specific time period, and there are significant changes in the wind speed and temperature gradients in the vertical direction of the wind field (such as the changes in wind speed and temperature exceeding specific values ​​at certain heights in the vertical direction), this means that the importance of gravity wave breaking modes in the evolution of wind field meteorological conditions has increased, and their scale coupling evolution weight will also increase.

[0101] The weights for the third radiative transfer mode, wind speed attention weight, heat diffusion mode weight, and ground-scale environmental vector are determined based on the descriptive distribution labels of several fourth-state trend description events. Regarding the weight of the third radiative transfer mode, the fourth-state trend description events may involve the relationship between radiative transfer and wind field energy exchange. For example, in wind field environmental sensing data, if the fourth-state trend description events include information on solar radiation transmission in the wind field and its energy conversion with the wind field, and the descriptive distribution labels indicate that when a solar radiation enhancement event occurs within a certain time period (e.g., solar radiation intensity increases by 50 W / m² each time over five consecutive observation periods), the energy distribution in the wind field changes significantly (e.g., in a specific area, the increase in air temperature is proportional to the increase in radiation intensity; for example, for every 50 W / m² increase in radiation intensity, the air temperature increases by an average of 2 degrees Celsius). This shows that radiative transfer has a significant impact on the wind field energy state, so the scale-coupling evolution weight of the third radiative transfer mode will be adjusted accordingly. For wind speed attention weights, if the fourth-state trend description event includes the impact of wind speed changes on wind field energy and material diffusion, and the description distribution label shows that a sudden increase in wind speed (e.g., an increase of more than 2 m / s each time for three consecutive observation periods) causes a faster rate of material diffusion in the wind field (e.g., the diffusion range of a specific pollutant in the wind field expands proportionally with increasing wind speed), this indicates that wind speed is becoming more important in the evolution of wind field conditions, and the scale-coupled evolution weight of the wind speed attention weight will change accordingly. The same applies to heat diffusion pattern weights. If the fourth-state trend description event shows that heat diffuses from high-temperature areas to low-temperature areas within a certain region, and the description distribution label indicates that this heat diffusion rate accelerates within a specific time period (e.g., the heat diffusion range increases by 10% each time for four consecutive observation periods) and is significantly correlated with factors such as wind speed and wind direction (e.g., the higher the wind speed, the more consistent the heat diffusion direction with the wind direction, and the faster the diffusion rate), then the scale-coupled evolution weight of the heat diffusion pattern weight will be determined based on this situation. For the ground-scale environment vector, the fourth-state trend description event includes the impact of factors such as ground temperature, humidity, and roughness on the wind field. When the description distribution label shows that within a certain time period, the ground temperature increases (e.g., the ground temperature increases by 1 degree each time in 6 consecutive observation periods), the ground humidity decreases (e.g., the humidity decreases by 5% each time), and the ground roughness remains unchanged, the wind field near the ground will show corresponding changes in wind speed, wind direction, and energy distribution (e.g., near-surface wind speed decreases, and wind direction deflects slightly). This indicates the degree of influence of the ground-scale environment on the meteorological state of the wind field, thereby determining the scale coupling evolution weights of the ground-scale environment vector.

[0102] The ratio of the cloud microphysical environment vector, the weight of the first radiative transfer mode, and the weight of the second radiative transfer mode is determined based on the descriptive distribution labels of several second-state trend description events, several third-state trend description events, and several fourth-state trend description events to determine its scale coupling evolution weight. The cloud microphysical environment vector is closely related to cloud formation, development, and its impact on the wind field. For example, second-state trend description events may involve changes in cloud type. When the descriptive distribution labels show that the frequency of cumulus cloud occurrence increases over a period of time (e.g., the frequency of cumulus cloud occurrence increases from 10% to 30% within 10 observation periods), and the height and thickness of cumulus clouds also change (e.g., the average height of cumulus clouds increases from 1000 meters to 1500 meters, and the thickness increases from 500 meters to 800 meters), this will affect the vertical structure of the wind field and water vapor distribution. The third-state trend description event may be related to cloud microphysical processes (such as the formation, growth, and descent of water droplets). When the description distribution label shows that the average radius of water droplets in the cloud increases over certain periods (e.g., the average radius of water droplets increases from 10 micrometers to 15 micrometers over five consecutive observation periods), this affects the optical properties and precipitation potential of the cloud, thus having a feedback effect on the wind field. The fourth-state trend description event, as mentioned above, involves the relationship between radiative transfer and the wind field. Changes in the cloud can alter radiative transfer characteristics; for example, when cloud cover increases, the absorption, scattering, and reflection of solar radiation change, thereby affecting the energy balance of the wind field. By combining the description distribution labels of these different state trend description events, the scale coupling evolution weights of the cloud microphysical environment vector can be determined. The ratio of the weights of the first and second radiative transfer modes is also based on the description distribution labels of these state trend description events. For example, the first radiative transfer mode may be associated with shortwave radiation, and the second radiative transfer mode may be associated with longwave radiation. When changes in clouds (described by the second and third state trends) and the interaction between radiation and wind fields (described by the fourth state trend) affect the transmission and conversion of shortwave and longwave radiation, the scale-coupled evolution weights of the ratio of the weights of the first radiative transfer mode to the weights of the second radiative transfer mode are determined based on relevant information in the description distribution labels, such as the changes in the energy ratios of shortwave and longwave radiation under different cloud conditions and wind field conditions.

[0103] The weights of the 3D wind field model are determined based on the descriptive distribution labels of several third-state trend description events. The 3D wind field model weights reflect the structure and evolution characteristics of the wind field in three-dimensional space. Third-state trend description events may include changes in wind speed and direction in both the horizontal and vertical directions. For example, if the descriptive distribution labels show that the vertical wind speed shear increases within a certain time period (e.g., from 2 m / s / 100 m to 3 m / s / 100 m in the 100-500 m height range), and the horizontal wind direction shear also changes (e.g., from 10 degrees to 15 degrees within a 10 km range), this indicates a significant change in the 3D structure of the wind field. The importance of the 3D wind field model in the evolution of wind field meteorological conditions becomes prominent, and its scale-coupled evolution weights will be adjusted accordingly.

[0104] For the weights of wind field noise, wind field remote sensing models, and multi-mode wind field monitoring changes, their scale-coupled evolution weights are determined based on the descriptive distribution labels of the fifth-state trend description events. The wind field noise weight is related to various interference factors in wind field observations. The fifth-state trend description events may include the impact of wind field sensor malfunctions, external environmental interference, etc., on the accuracy of wind field observation data. For example, if the descriptive distribution labels show an increase in the frequency of wind field sensor malfunctions within a certain time period (e.g., the number of sensor malfunctions increases from 1 to 3 within 10 observation periods), this will lead to increased noise in the wind field observation data. In this case, the scale-coupled evolution weight of the wind field noise weight will increase accordingly to reflect the impact of this interference factor on the analysis of wind field meteorological conditions. For the weights of wind field remote sensing models, the fifth-state trend description events may involve the accuracy and coverage of remote sensing data in wind field monitoring. When the description distribution label shows that, within a certain time period, the resolution of remote sensing data improves (e.g., spatial resolution increases from 1 km to 500 m), the coverage expands (e.g., the coverage area increases from 100 km² to 150 km²), and the consistency with other observation data (e.g., ground observation data) increases (e.g., the average wind speed difference between the two decreases from 2 m / s to 1 m / s), this indicates that the role of wind field remote sensing models in monitoring wind field meteorological conditions is becoming more important, and the scale coupling evolution weight of the wind field remote sensing model weight will be adjusted accordingly. For the weight of changes in multi-mode wind field monitoring, the fifth state trend description event includes the comprehensive situation of wind field change monitoring by multiple monitoring modes (e.g., ground observation, remote sensing observation, aerial sounding, etc.). When the description distribution label is displayed within a certain time period, the synergy of multiple monitoring modes is enhanced (such as the deviation of wind speed and wind direction observation results of different monitoring modes in the same area of ​​the wind field is reduced), and the ability to capture wind field changes is improved (such as being able to detect sudden changes in wind speed and wind direction in the wind field more timely). This indicates that the importance of multi-mode wind field monitoring in the evolution of wind field meteorological conditions has increased, and the scale coupling evolution weight of the multi-mode wind field monitoring change weight will be determined according to this situation.

[0105] This design, by associating different target scale coupling evolution vectors with descriptive distribution labels of specific state trend descriptive events to determine scale coupling evolution weights, can more accurately reflect the complex characteristics of wind farm meteorological conditions. This targeted association considers the influence of various physical processes, environmental factors, and observation modes in the wind farm, ensuring that the determination of each weight is based on solid data and events. This helps improve the accuracy of wind farm meteorological condition prediction, provides a more comprehensive understanding of the dynamic changes in the wind farm, and thus offers a more reliable basis for wind farm operation management and meteorological research.

[0106] Furthermore, Figure 2 This is a schematic diagram of the structure of a weather condition prediction system 200 provided in an embodiment of this application. Figure 2 The weather condition forecasting system 200 shown includes a processor 210, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0107] Optionally, such as Figure 2 As shown, the weather forecasting system 200 may further include a memory 230. The processor 210 can retrieve and run computer programs from the memory 230 to implement the methods described in this embodiment.

[0108] The memory 230 can be a separate device independent of the processor 210, or it can be integrated into the processor 210.

[0109] Optionally, such as Figure 2 As shown, the weather forecasting system 200 may also include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices. Specifically, it may send information or data to other devices or receive information or data sent by other devices.

[0110] Optionally, the weather condition prediction system 200 can implement the corresponding processes of the storage engine or components (such as processing modules) in the storage engine or the device with the storage engine deployed in the various methods of the embodiments of this application. For the sake of brevity, these will not be described in detail here.

[0111] It should be understood that the processor in this application embodiment may be an integrated circuit chip with signal processing capabilities.

[0112] It is understood that the memory in the embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. It should be noted that the memory in the systems and methods described herein is intended to include, but is not limited to, suitable types of memory.

[0113] Based on the above, a readable storage medium is provided, on which a program or instructions are stored, and when the program or instructions are executed by a processor, the steps of the above method are implemented.

[0114] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0115] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to the prior art, can be embodied in the form of computer software products. The computer software products are stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and include several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the embodiments of this application.

[0116] The embodiments of this application have been described above with reference to the accompanying drawings. However, the embodiments of this application are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the embodiments of this application without departing from the spirit and protection scope of the embodiments of this application, and all of these forms are within the protection scope of the embodiments of this application.

Claims

1. A meteorological state prediction method based on big data, characterized in that, The method is implemented through a meteorological state prediction system, and the method includes: Acquire multi-source wind farm observation data, and identify state trend description events in the target meteorological observation data included in the multi-source wind farm observation data to obtain description distribution labels of several target state trend description events in the target meteorological observation data; By using the description distribution labels of the several target state trend description events, determine the scale coupling evolution weight of at least one target scale coupling evolution vector in the target meteorological observation data; The global trend evolution parameters and multi-source coupling evolution discrete index of each set target scale coupling evolution vector are obtained. The global trend evolution parameters and multi-source coupling evolution discrete index are obtained by processing the scale coupling evolution weights of each set target scale coupling evolution vector of several different target meteorological observation data. Based on the global trend evolution parameters, multi-source coupling evolution discrete index, and scale coupling evolution weight of each set target scale coupling evolution vector, the state prediction viewpoint and the confidence feature of each state prediction viewpoint are determined. The meteorological state prediction results of the multi-source wind farm observation data are determined by using the state prediction perspective of the coupled evolution vectors of each set target scale.

2. The method as described in claim 1, characterized in that, The process of obtaining the global trend evolution parameters and multi-source coupled evolution discrete indices of the coupled evolution vectors at each set target scale includes: Physical process quantification mapping is performed on the target meteorological observation data in the multi-source wind farm observation data to obtain the target physical process quantification characteristics corresponding to the multi-source wind farm observation data. The global trend evolution parameters and multi-source coupling evolution discrete indices corresponding to the coupling evolution vectors of each set target scale are obtained by quantifying the characteristics of the target physical process.

3. The method as described in claim 2, characterized in that, The method further includes: Acquire several multi-source historical observation data, and identify state trend description events in the historical meteorological observation data included in each multi-source historical observation data to obtain historical description distribution labels for several historical state trend description events in each historical meteorological observation data. Based on the historical description distribution labels of several historical state trend description events in each historical meteorological observation data, determine the historical scale coupling evolution weight of at least one target scale coupling evolution vector in each historical meteorological observation data. Physical process quantification mapping is performed on each historical meteorological observation data to obtain the historical physical process quantification characteristics of each multi-source historical observation data. By determining the historical scale coupling evolution weights of the respective target scale coupling evolution vectors in historical meteorological observation data with the same historical physical process quantification characteristics, the state evolution thermodynamic relationship network of the respective target scale coupling evolution vectors corresponding to the same historical physical process quantification characteristics is determined.

4. The method as described in claim 1, characterized in that, The process of determining the state prediction viewpoints and confidence features of each set target scale coupled evolution vector by using the global trend evolution parameters, multi-source coupled evolution discrete indices, and scale coupled evolution weights of each set target scale coupled evolution vector includes: By using the global trend evolution parameters and multi-source coupling evolution discrete index of each set target scale coupling evolution vector, the scale coupling evolution weights of each set target scale coupling evolution vector are embedded to obtain the scale coupling evolution embedding features of each set target scale coupling evolution vector. Obtain the vortex mode characterization vectors corresponding to the coupling evolution vectors of each set target scale; By utilizing the corresponding vortex mode characterization vectors of each set target scale coupled evolution vector and the scale coupled evolution embedding features of each set target scale coupled evolution vector, the state prediction viewpoint of each set target scale coupled evolution vector is determined. The confidence features of each state prediction viewpoint are determined by the vortex mode characterization vector corresponding to each set target scale coupling evolution vector and the scale coupling evolution embedding feature of each set target scale coupling evolution vector; or, the confidence features of each state prediction viewpoint are determined by the vortex mode characterization vector corresponding to each set target scale coupling evolution vector, the global trend evolution parameter of each set target scale coupling evolution vector, the multi-source coupling evolution discrete index, and the scale coupling evolution weight.

5. The method as described in claim 4, characterized in that, When the target meteorological observation data is wind field environmental sensing and monitoring data, the step of determining the state prediction viewpoint of each set target scale coupling evolution vector by utilizing the vortex model characterization vector corresponding to the scale coupling evolution weight of each set target scale coupling evolution vector and the scale coupling evolution embedding feature of each set target scale coupling evolution vector includes: When the target scale coupling evolution vector is set as a ground-scale environment vector or a cumulus-scale boundary vector, the state prediction view of the scale coupling evolution embedding feature of the target scale coupling evolution vector is determined based on the first vortex mode characterization vector and the scale coupling evolution embedding feature of the target scale coupling evolution vector. The first vortex mode characterization vector is used to indicate that the state prediction view and the scale coupling evolution embedding feature have a first quantization relationship. When the scale coupling evolution weight of the target scale coupling evolution vector is set to the cloud microphysical environment vector, the state prediction view of the scale coupling evolution embedding feature of the target scale coupling evolution vector is determined based on the second vortex mode characterization vector and the scale coupling evolution embedding feature of the target scale coupling evolution vector. The second vortex mode characterization vector is used to indicate that the state prediction view and the scale coupling evolution embedding feature have a second quantification relationship. When the scale coupling evolution weight of the target scale coupling evolution vector is set to one of the following: road surface mode weight, gravity wave fracturing mode weight, the ratio of the first radiative transfer mode weight to the second radiative transfer mode weight, the third radiative transfer mode weight, and the heat diffusion mode weight, the state prediction viewpoint of the scale coupling evolution embedding feature of the target scale coupling evolution vector is determined based on the third vortex mode characterization vector and the scale coupling evolution embedding feature of the target scale coupling evolution vector. The third vortex mode characterization vector is used to indicate the correlation between the state prediction viewpoint and the prediction offset, and the prediction offset is the comparison result between the scale coupling evolution embedding feature and the state probability statistics map.

6. The method as described in claim 4, characterized in that, The process of determining the confidence features of each state prediction viewpoint by using the corresponding vortex mode characterization vector of each set target scale coupled evolution vector and the scale coupled evolution embedding features of each set target scale coupled evolution vector includes: Obtain the state probability statistics graph vector; By embedding the scale coupling evolution of each set target scale coupling evolution vector into the vortex mode characterization vector corresponding to each set target scale coupling evolution vector, the state probability statistical graph vector is fully connected to obtain the fully connected feature value corresponding to each set target scale coupling evolution vector. The confidence features corresponding to each set target scale coupling evolution vector are determined by using the fully connected eigenvalues ​​corresponding to each set target scale coupling evolution vector and the vortex mode characterization vectors corresponding to each set target scale coupling evolution vector.

7. The method as described in claim 4, characterized in that, The confidence features of each state prediction viewpoint are determined based on the corresponding vortex mode characterization vector of each set target scale coupled evolution vector, the global trend evolution parameters of each set target scale coupled evolution vector, the multi-source coupled evolution discrete index, and the scale coupled evolution weight, including: The state heatmap corresponding to each set target scale coupling evolution vector is determined by the global trend evolution parameters of each set target scale coupling evolution vector and the multi-source coupling evolution discrete index. Based on the scale coupling evolution weights of each set target scale coupling evolution vector and the vortex mode characterization vectors corresponding to each set target scale coupling evolution vector, the state heatmap is fully connected to obtain the fully connected feature values ​​corresponding to each set target scale coupling evolution vector. The confidence features corresponding to each set target scale coupling evolution vector are determined by using the fully connected eigenvalues ​​corresponding to each set target scale coupling evolution vector and the vortex mode characterization vectors corresponding to each set target scale coupling evolution vector.

8. The method as described in claim 6 or 7, characterized in that, The step of determining the confidence features of each set target scale coupled evolution vector by using the fully connected eigenvalues ​​corresponding to each set target scale coupled evolution vector and the vortex mode characterization vector corresponding to each set target scale coupled evolution vector includes: When the vortex mode characterization vector corresponding to the target scale coupling evolution vector is set as the first vortex mode characterization vector, the fully connected feature value is determined as the confidence feature corresponding to the set target scale coupling evolution vector. When the vortex mode characterization vector corresponding to the set target scale coupling evolution vector is the second vortex mode characterization vector, the difference between the first set value and the fully connected feature value is determined as the confidence feature corresponding to the set target scale coupling evolution vector. When the vortex mode characterization vector corresponding to the set target scale coupling evolution vector is the third vortex mode characterization vector, the weighted result of the second set value and the fully connected feature value is determined as the confidence feature corresponding to the set target scale coupling evolution vector.

9. The method according to any one of claims 1 to 7, characterized in that, The process of identifying state trend description events in the target meteorological observation data included in the multi-source wind farm observation data yields description distribution labels for several target state trend description events in the target meteorological observation data, including: Data extraction is performed on the multi-source wind farm observation data to obtain the initial target meteorological observation data in the multi-source wind farm observation data; When the initial target meteorological observation data is determined to meet the optimization conditions, the optimization indication characteristics of the initial target meteorological observation data are determined. The initial target meteorological observation data is optimized by using the optimized indicator features to obtain optimized target meteorological observation data. The optimized target meteorological observation data is used to identify state trend description events, resulting in description distribution labels for several target state trend description events.

10. A meteorological condition prediction system, characterized in that, The method includes at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, causing the at least one processor to perform the method according to any one of claims 1-9.

Images