Provincial wind power prediction method, system and equipment and storage medium
By calculating initial weights and dynamically correcting them in provincial wind power forecasting, and combining multi-dimensional input feature sets and LGBM model training, the problems of regional heterogeneity and data dependence in existing technologies are solved, achieving high-precision and flexible wind power forecasting that meets the multi-day forecasting needs of the power grid.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG GUODIAN POWER GENERATION MAINTENANCE CO LTD
- Filing Date
- 2025-11-21
- Publication Date
- 2026-04-24
AI Technical Summary
Existing provincial wind power forecasting methods fail to effectively consider the differences in geographical conditions and meteorological responses among prefecture-level cities, resulting in models that cannot reflect regional heterogeneity and are difficult to adapt to the forecasting needs of the power grid in weekly scheduling or multi-day power trading. They lack flexibility and rely on a single data source, leading to limited feature dimensions and poor data availability.
Initial weights are calculated by acquiring historical installed capacity data and average altitude data of the target province, and dynamically corrected based on historical NWP meteorological data. A multi-dimensional input feature set is constructed, trained using an LGBM model, and optimized by a rolling update mechanism to achieve adaptive adjustment of dynamic weights and feature sets.
It improves the accuracy and stability of provincial wind power forecasting, enhances the model's generalization ability in complex meteorological scenarios, ensures that the forecast results are consistent with the actual environment, and has the ability to continuously learn and dynamically optimize.
Smart Images
Figure CN121923080A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of wind power generation technology, specifically relating to a provincial-level wind power prediction method, system, equipment, and storage medium. Background Technology
[0002] In the power system, wind power generation, as an important component of clean energy, has the characteristics of wide energy distribution and strong renewability. Provincial wind power forecasting is an important technical link for grid security dispatch and power trading.
[0003] However, existing provincial-level wind power forecasting methods still have several shortcomings. First, most methods treat the entire province as a homogeneous region, using only average power or fixed installed capacity weights for modeling, without considering differences in geographical conditions, altitude, and meteorological response among prefecture-level cities. This results in models failing to reflect regional heterogeneity and limiting forecast accuracy. Second, existing models are generally limited to day-ahead (24-hour) forecasting scenarios, making it difficult to adapt to the grid's need for 2-7 day forecast results in weekly dispatching or multi-day electricity trading, thus lacking flexibility. Third, traditional models often rely on a single data source, with limited feature dimensions and poor data availability, restricting the widespread application of these models in practical operation. Summary of the Invention
[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a provincial wind power prediction method, system, device and storage medium to solve the above-mentioned technical problems.
[0005] In a first aspect, the present invention provides a provincial-level wind power prediction method, comprising: Obtain historical wind power data of the target province and preset data of various prefecture-level cities in the target province. The preset data includes historical installed capacity data, average altitude data and historical NWP meteorological data. The initial weights are calculated based on historical installed capacity data and average altitude data. The initial weights are then dynamically corrected based on historical NWP meteorological data and historical wind power data to obtain dynamic weights. A multi-dimensional input feature set is constructed based on dynamic weights and historical NWP meteorological data. The multi-dimensional input feature set is used as the training set to train the initial LGBM model, thus obtaining the provincial wind power prediction model. The latest NWP meteorological data is input into the provincial wind power prediction model, and the provincial wind power prediction results for the next N days are output.
[0006] In an optional implementation, initial weights are calculated based on historical installed capacity data and average altitude data, including: +
[0007] in, Let be the initial weight of the i-th prefecture-level city within the target province. This refers to the installed capacity data for the i-th prefecture-level city within the target province. For the total installed capacity data of the target province, Let be the average altitude of the i-th prefecture-level city within the target province. The average altitude of the target province is α, and the weighting coefficient is 0.5-0.8.
[0008] In an optional implementation, the initial weights are dynamically adjusted based on historical NWP meteorological data and historical wind power data to obtain dynamic weights, including:
[0009] in, Let be the dynamic weight of the i-th prefecture-level city within the target province. Let Pearson's correlation coefficient be the relationship between the wind speed of the i-th prefecture-level city in the target province and the total wind power of the province over the past T days. This represents the maximum absolute value of the correlation among all prefecture-level cities within the target province.
[0010] In one optional implementation, The calculation formula is:
[0011] in, This represents the wind speed sample of the i-th prefecture-level city within the target province during the k-th hour; This represents the average hourly wind speed over T days in the i-th prefecture-level city within the target province; This represents the provincial power sample for the k-th hour; This represents the average hourly provincial power over days T; n is the total sample size. , where is the number of hours within day T.
[0012] In one optional implementation, a multidimensional input feature set is constructed based on dynamic weights, historical NWP meteorological data, and historical wind power data, including: Based on each type of meteorological data in historical NWP meteorological data, a weighted meteorological feature is constructed using dynamic weights as weighting coefficients, including:
[0013] in, Let m be the weighted average characteristic of the m-th type of meteorological data for the target province at time t. Let q represent the m-th type of meteorological data for the i-th prefecture-level city in the target province at time t, where q represents the number of prefecture-level cities in the target province. By progressively increasing the preset time step, the historical wind power data within the preset sliding window is traversed sequentially. The mean and maximum values of the historical wind power data within the preset sliding window are calculated to construct the historical power time series characteristics, including:
[0014]
[0015] in, For the current time t, the historical average wind power output over the preceding L hours is calculated. For the current time t, look back L hours to find the historical maximum wind power output. This represents the historical wind power data for the h-th hour, where t is the current calculation time, L represents the sliding window length, and h is the specific hour number in the time series. ; A multidimensional input feature set is constructed based on weighted meteorological features and historical power time series features.
[0016] In an optional implementation, it further includes: After each daily forecast, newly added measured power data and NWP meteorological data are added to the training set, and the dynamic weights and multi-dimensional input feature sets are recalculated to continuously update the provincial wind power forecast model.
[0017] Secondly, the present invention provides a provincial-level wind power prediction system, comprising: The data acquisition module is used to acquire historical wind power data of the target province and preset data of various prefecture-level cities in the target province. The preset data includes historical installed capacity data, average altitude data and historical NWP meteorological data. The weight calculation module is used to calculate the initial weight based on historical installed capacity data and average altitude data, and to dynamically correct the initial weight based on historical NWP meteorological data and historical wind power data to obtain dynamic weight. The feature set construction module is used to construct a multi-dimensional input feature set based on dynamic weights and historical NWP meteorological data. The multi-dimensional input feature set is used as a training set to train the initial LGBM model and obtain the provincial wind power prediction model. The model output module is used to input the latest NWP meteorological data into the provincial wind power prediction model and output the provincial wind power prediction results for the next N days.
[0018] Thirdly, a device is provided, comprising: The memory is used to store the provincial wind power forecasting program; A processor is configured to implement the steps of the provincial wind power prediction method as provided in the first aspect when executing the provincial wind power prediction program.
[0019] Fourthly, a computer-readable storage medium is provided, on which a provincial wind power prediction program is stored, wherein when the provincial wind power prediction program is executed by a processor, the provincial wind power prediction method as provided in the first aspect is implemented.
[0020] The beneficial effects of this invention are as follows: The provincial wind power prediction method, system, equipment, and storage medium provided by this invention calculate the initial weights based on the historical installed capacity and average altitude of each prefecture-level city in the target province, and then correct them to obtain dynamic weights. The use of dynamic weight modeling solves the problem of decreased accuracy caused by the homogenization of the entire province in existing provincial predictions. By integrating meteorological features, temporal features, and historical power features to construct a multi-dimensional input feature set, the model can simultaneously learn temporal patterns, enhancing its stability and generalization ability under complex meteorological scenarios. Furthermore, this application introduces a rolling update mechanism, which automatically incorporates the latest measured data daily after model deployment, recalculates the dynamic weights, and updates the training set, realizing adaptive retraining of the LGBM model. This enables continuous learning and dynamic optimization capabilities, ensuring that the prediction results always remain consistent with the actual operating environment.
[0021] Furthermore, the design principle of this invention is reliable, the structure is simple, and it has a very wide range of application prospects. Attached Figure Description
[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1 This is a schematic flowchart of a method according to an embodiment of the present invention.
[0024] Figure 2 This is a schematic block diagram of a system according to an embodiment of the present invention.
[0025] Figure 3 This is a schematic diagram of the structure of a device provided in an embodiment of the present invention. Detailed Implementation
[0026] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.
[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
[0028] The provincial wind power prediction method provided in this embodiment of the invention is executed by computer equipment, and correspondingly, the provincial wind power prediction system runs on the computer equipment.
[0029] Figure 1 This is a schematic flowchart illustrating a method according to an embodiment of the present invention. Wherein, Figure 1 The implementing entity can be a provincial-level wind power forecasting system. Depending on different requirements, the order of the steps in this flowchart can be changed, and some steps can be omitted.
[0030] like Figure 1 As shown, the method includes: S1. Obtain historical wind power data of the target province and preset data of various prefecture-level cities in the target province. The preset data includes historical installed capacity data, average altitude data and historical NWP meteorological data.
[0031] S2. Calculate the initial weights based on historical installed capacity data and average altitude data, and dynamically correct the initial weights based on historical NWP meteorological data and historical wind power data to obtain dynamic weights.
[0032] S3. Construct a multi-dimensional input feature set based on dynamic weights and historical NWP meteorological data. Use the multi-dimensional input feature set as the training set to train the initial LGBM model and obtain the provincial wind power prediction model.
[0033] S4. Input the latest NWP meteorological data into the provincial wind power prediction model and output the provincial wind power prediction results for the next N days.
[0034] In one embodiment of the present invention, based on step S1, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.
[0035] The system acquires historical wind power data for the target province and preset data for each prefecture-level city in the target province. The preset data includes historical installed capacity data, average altitude data, and historical NWP meteorological data.
[0036] Specifically, firstly, historical wind power data of the target province is retrieved from the power grid dispatch or energy management platform, preferably covering at least the past year with a time granularity of 15 minutes or 1 hour, and unified to the hourly time scale and a unified time zone (such as UTC+8) to represent the actual output of wind turbine units in the province at time t; then, for all prefecture-level cities in the target province, their preset data are collected, including (1) historical installed capacity data: the grid-connected installed capacity (MW) of each time period is given according to the prefecture-level city dimension, and if there is expansion, the effective time should be recorded and a segmented time series should be generated; (2) average altitude data: the average altitude (m) is obtained by statistically analyzing the administrative boundaries of prefecture-level cities based on publicly available geographic survey data, and if necessary, area weighting is introduced for areas with significant topographic relief. Or the weighted average of wind turbine locations; (3) Historical NWP meteorological data: hourly numerical weather forecast / reanalysis data of the corresponding grid of the prefecture-level city, including at least wind speed and wind direction. It is recommended to obtain temperature, humidity, air pressure, precipitation and other elements at the same time; In order to ensure spatial matching, the original latitude and longitude grid of NWP is aggregated to the boundary of the prefecture-level city through nearest neighbor / bilinear interpolation and weighted by area or wind turbine density to form a city-level time series meteorological sequence; In order to ensure time matching, the power and NWP are aligned with a unified time scale (if necessary, the power is resampled for 15→60 minutes, and the average or power preservation strategy is adopted), and basic quality control (missing measurement mark, 3σ anomaly removal and linear interpolation completion), unit unification and field standardization are completed. Finally, a structured dataset of "provincial power sequence + static characteristics of installed capacity / altitude of each city + hourly NWP meteorological sequence of each city" is output to provide reliable input for subsequent weight calculation, feature construction and model training.
[0037] In one embodiment of the present invention, based on step S2, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.
[0038] Initial weights are calculated based on historical installed capacity data and average altitude data. These initial weights are then dynamically adjusted based on historical NWP meteorological data and historical wind power data to obtain dynamic weights, including: +
[0039] in, Let be the initial weight of the i-th prefecture-level city within the target province. This refers to the installed capacity data for the i-th prefecture-level city within the target province. For the total installed capacity data of the target province, Let be the average altitude of the i-th prefecture-level city within the target province. The average altitude of the target province is α, and the weighting coefficient is 0.5-0.8.
[0040]
[0041] in, Let be the dynamic weight of the i-th prefecture-level city within the target province. Let Pearson's correlation coefficient be the relationship between the wind speed of the i-th prefecture-level city in the target province and the total wind power of the province over the past T days. This represents the maximum absolute value of the correlation among all prefecture-level cities within the target province.
[0042] The calculation formula is:
[0043] in, This represents the wind speed sample of the i-th prefecture-level city within the target province during the k-th hour; This represents the average hourly wind speed over T days in the i-th prefecture-level city within the target province; This represents the provincial power sample for the k-th hour; This represents the average hourly provincial power over days T; n is the total sample size. , where is the number of hours within day T.
[0044] In one embodiment of the present invention, based on step S3, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.
[0045] A multi-dimensional input feature set is constructed based on dynamic weights and historical NWP meteorological data. This multi-dimensional input feature set is used as the training set to train the initial LGBM model, resulting in a provincial wind power prediction model, including: Based on each type of meteorological data in historical NWP meteorological data, a weighted meteorological feature is constructed using dynamic weights as weighting coefficients, including:
[0046] in, Let m be the weighted average characteristic of the m-th type of meteorological data for the target province at time t. Let q represent the m-th type of meteorological data for the i-th prefecture-level city in the target province at time t, where q represents the number of prefecture-level cities in the target province. By progressively increasing the preset time step, the historical wind power data within the preset sliding window is traversed sequentially. The mean and maximum values of the historical wind power data within the preset sliding window are calculated to construct the historical power time series characteristics, including:
[0047]
[0048] in, For the current time t, the historical average wind power output over the preceding L hours is calculated. For the current time t, look back L hours to find the historical maximum wind power output. This represents the historical wind power data for the h-th hour, where t is the current calculation time, L represents the sliding window length, and h is the specific hour number in the time series. ; A multidimensional input feature set is constructed based on weighted meteorological features and historical power time series features.
[0049] This application also includes adding newly added measured power data and NWP meteorological data to the training set after daily forecasts, recalculating dynamic weights and multidimensional input feature sets, and continuously updating the provincial wind power forecast model.
[0050] Specifically, firstly, the newly added provincial measured wind power data and NWP meteorological data from various prefecture-level cities from the previous day are imported into the database. All data are aligned and formatted according to hourly time granularity. If the power data is at a 15-minute granularity, it is converted to hourly data using a time averaging method; if the time resolution of the NWP data is higher than 1 hour, it is rounded down to the nearest whole number or processed using linear interpolation. Then, data quality control is performed, including deduplication, timestamp correction, outlier removal, and missing value completion. Outlier removal adopts the 3σ criterion, that is, power or wind speed data exceeding the mean ± 3 standard deviations are removed; missing value completion uses linear or spline interpolation methods to ensure the continuity of the time series. After cleaning, only the data from the most recent 30 to 60 days are retained as a rolling training set to ensure that the model can reflect the latest meteorological characteristics without causing training lag due to too many historical samples.
[0051] Next, the system recalculates the dynamic weights based on the updated training data. First, initial weights are calculated based on the installed capacity and average altitude of each prefecture-level city. The installed capacity percentage reflects power potential, while altitude affects wind speed intensity. These two factors are linearly combined using a weighting coefficient α (typically 0.5 to 0.8) to obtain the initial weights. Then, using the newly added data, the Pearson correlation coefficient between wind speed in each prefecture-level city and the total power of the province was calculated. The initial weights are then multiplied by their normalized correlation values across all prefecture-level cities to obtain new dynamic weights. After the weights are updated, the system regenerates the multidimensional input feature set using the latest weights and NWP data. The feature set consists of three parts: the first part is meteorological features, which categorizes elements such as wind speed, wind direction, temperature, humidity, and air pressure for each prefecture-level city according to... The first part is the weighted average, which forms the provincial representative features. The second part is the power time series features, including the power lag values of the past 1, 2, 3, and 7 hours, as well as the moving average and maximum values of the past 24 hours, used to reflect the inertia and fluctuation trend of wind power. The third part is the time features, which standardize the time of day into hourly fractions of 0 to 1 to characterize the diurnal cycle. All features are standardized, and the Z-score method is used to subtract the training set mean from each feature and divide by the standard deviation to ensure the stability of the input feature distribution. Subsequently, the system performs rolling model training based on the updated feature set and label data. The model uses the previous parameters as the initial state and only performs incremental training on new data to shorten the update time. If the prediction error of the validation set is significantly improved compared to the previous version (e.g., the mean absolute error (MAE) increases by more than 5%), full retraining is automatically triggered. The training uses the LGBM algorithm, with a learning rate of 0.05, 64 leaf nodes, and a feature sampling rate and sample sampling rate of 0.8. An early stopping mechanism is enabled to prevent overfitting. After model training is complete, the system saves the new parameter file, standardized parameters, dynamic weight matrix, and model evaluation metrics, marks the model version number as t+1, and deploys it as the prediction model for the next day. If the data quality check finds a missing rate exceeding 10% or an anomaly rate exceeding 5%, the system will stop updating and issue an alarm. Through this process, the system achieves a fully automated closed loop of automatic data supplementation, dynamic weight adaptation, multi-dimensional feature reconstruction, and rolling optimization of model parameters, enabling the model to continuously learn the latest meteorological and power change patterns and maintain high accuracy and robustness in provincial wind power prediction.
[0052] In some embodiments, the provincial wind power forecasting system may include multiple functional modules composed of computer program segments. The computer programs for each program segment in the provincial wind power forecasting system may be stored in the memory of a computer device and executed by at least one processor to perform (see details). Figure 1 (Description) Function for provincial wind power forecasting.
[0053] In this embodiment, the provincial wind power prediction system can be divided into multiple functional modules according to its functions, such as... Figure 2 As shown. The module referred to in this invention is a series of computer program segments that can be executed by at least one processor and perform a fixed function, and is stored in memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.
[0054] The data acquisition module is used to acquire historical wind power data of the target province and preset data of various prefecture-level cities in the target province. The preset data includes historical installed capacity data, average altitude data and historical NWP meteorological data. The weight calculation module is used to calculate the initial weight based on historical installed capacity data and average altitude data, and to dynamically correct the initial weight based on historical NWP meteorological data and historical wind power data to obtain dynamic weight. The feature set construction module is used to construct a multi-dimensional input feature set based on dynamic weights and historical NWP meteorological data. The multi-dimensional input feature set is used as a training set to train the initial LGBM model and obtain the provincial wind power prediction model. The model output module is used to input the latest NWP meteorological data into the provincial wind power prediction model and output the provincial wind power prediction results for the next N days.
[0055] Figure 3 The provincial wind power prediction method provided in this application embodiment can be applied to equipment. Those skilled in the art will understand that the equipment structure involved in the embodiments of this invention does not constitute a limitation on the equipment. The equipment may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. In the embodiments of this invention, the equipment includes, but is not limited to, laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The equipment may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.
[0056] The device 300 may include a processor 310, a memory 320, and a communication unit 330. These components communicate via one or more buses. Those skilled in the art will understand that the server structure shown in the figure does not constitute a limitation of the present invention. It may be a bus topology or a star topology, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.
[0057] The memory 320 can be used to store execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 320 are executed by the processor 310, the device 300 is able to perform some or all of the steps in the above method embodiments.
[0058] The processor 310 serves as the control center of the storage device, connecting various parts of the electronic device via various interfaces and lines. It executes software programs and / or modules stored in the memory 320, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of integrated circuits (ICs), such as a single packaged IC or multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may consist only of a central processing unit (CPU). In this embodiment of the invention, the CPU may have a single processing core or include multiple processing cores.
[0059] The communication unit 330 is used to establish a communication channel, enabling the storage device to communicate with other devices. It can receive user data sent by other devices or send user data to other devices.
[0060] The present invention also provides a computer storage medium, wherein the computer storage medium may store a program, which, when executed, may include some or all of the steps provided in the embodiments of the present invention. The storage medium may be a magnetic disk, an optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0061] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other medium capable of storing program code. It includes several instructions to cause a computer device (which may be a personal computer, a server, or a second device, network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.
[0062] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.
[0063] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or modules may be electrical, mechanical, or other forms.
[0064] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0065] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.
[0066] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the present invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should also be covered within the protection scope of the present invention.
Claims
1. A provincial-level wind power prediction method, characterized in that, include: Obtain historical wind power data of the target province and preset data of various prefecture-level cities in the target province. The preset data includes historical installed capacity data, average altitude data and historical NWP meteorological data. The initial weights are calculated based on historical installed capacity data and average altitude data. The initial weights are then dynamically corrected based on historical NWP meteorological data and historical wind power data to obtain dynamic weights. A multi-dimensional input feature set is constructed based on dynamic weights and historical NWP meteorological data. The multi-dimensional input feature set is used as the training set to train the initial LGBM model, thus obtaining the provincial wind power prediction model. The latest NWP meteorological data is input into the provincial wind power prediction model, and the provincial wind power prediction results for the next N days are output.
2. The method according to claim 1, characterized in that, Initial weights are calculated based on historical installed capacity data and average altitude data, including: + in, Let be the initial weight of the i-th prefecture-level city within the target province. This refers to the installed capacity data for the i-th prefecture-level city within the target province. For the total installed capacity data of the target province, Let be the average altitude of the i-th prefecture-level city within the target province. The average altitude of the target province is α, and the weighting coefficient is 0.5-0.
8.
3. The method according to claim 1, characterized in that, The initial weights are dynamically adjusted based on historical NWP meteorological data and historical wind power data to obtain dynamic weights, including: in, Let be the dynamic weight of the i-th prefecture-level city within the target province. Let Pearson's correlation coefficient be the relationship between the wind speed of the i-th prefecture-level city in the target province and the total wind power of the province over the past T days. This represents the maximum absolute value of the correlation among all prefecture-level cities within the target province.
4. The method according to claim 3, characterized in that, The calculation formula is: in, This represents the wind speed sample of the i-th prefecture-level city within the target province during the k-th hour; This represents the average hourly wind speed over T days in the i-th prefecture-level city within the target province; This represents the provincial power sample for the k-th hour; This represents the average hourly provincial power over days T; n is the total sample size. , where is the number of hours within day T.
5. The method according to claim 1, characterized in that, A multi-dimensional input feature set is constructed based on dynamic weights, historical NWP meteorological data, and historical wind power data, including: Based on each type of meteorological data in historical NWP meteorological data, a weighted meteorological feature is constructed using dynamic weights as weighting coefficients, including: in, Let m be the weighted average characteristic of the m-th type of meteorological data for the target province at time t. Let q represent the m-th type of meteorological data for the i-th prefecture-level city in the target province at time t, where q represents the number of prefecture-level cities in the target province. By progressively increasing the preset time step, the historical wind power data within the preset sliding window is traversed sequentially. The mean and maximum values of the historical wind power data within the preset sliding window are calculated to construct the historical power time series characteristics, including: in, For the current time t, the historical average wind power output over the preceding L hours is calculated. For the current time t, look back L hours to find the historical maximum wind power output. This represents the historical wind power data for the h-th hour, where t is the current calculation time, L represents the sliding window length, and h is the specific hour number in the time series. ; A multidimensional input feature set is constructed based on weighted meteorological features and historical power time series features.
6. The method according to claim 1, characterized in that, Also includes: After each daily forecast, newly added measured power data and NWP meteorological data are added to the training set, and the dynamic weights and multi-dimensional input feature sets are recalculated to continuously update the provincial wind power forecast model.
7. A provincial-level wind power prediction system, characterized in that, include: The data acquisition module is used to acquire historical wind power data of the target province and preset data of various prefecture-level cities in the target province. The preset data includes historical installed capacity data, average altitude data and historical NWP meteorological data. The weight calculation module is used to calculate the initial weight based on historical installed capacity data and average altitude data, and to dynamically correct the initial weight based on historical NWP meteorological data and historical wind power data to obtain dynamic weight. The feature set construction module is used to construct a multi-dimensional input feature set based on dynamic weights and historical NWP meteorological data. The multi-dimensional input feature set is used as a training set to train the initial LGBM model and obtain the provincial wind power prediction model. The model output module is used to input the latest NWP meteorological data into the provincial wind power prediction model and output the provincial wind power prediction results for the next N days.
8. The system according to claim 7, characterized in that, The weight calculation module includes: The initial weight calculation unit includes: + in, Let be the initial weight of the i-th prefecture-level city within the target province. This refers to the installed capacity data for the i-th prefecture-level city within the target province. For the total installed capacity data of the target province, Let be the average altitude of the i-th prefecture-level city within the target province. The average altitude of the target province is α, and the weighting coefficient is 0.5-0.
8.
9. A device, characterized in that, include: The memory is used to store the provincial wind power forecasting program; A processor is configured to implement the steps of the provincial wind power prediction method as described in any one of claims 1-6 when executing the provincial wind power prediction program.
10. A computer-readable storage medium storing a computer program, characterized in that, The readable storage medium stores a provincial wind power prediction program, which, when executed by a processor, implements the steps of the provincial wind power prediction method as described in any one of claims 1-6.