New energy power prediction meteorological element screening method and related device

Abstract

This invention belongs to the field of power meteorological forecasting technology and discloses a method and related device for screening meteorological elements for new energy power prediction. The method includes: acquiring historical values ​​of installed capacity and historical characteristic values ​​of meteorological elements for each new energy power station within a region; based on the historical values ​​of installed capacity, performing a weighted average of the historical characteristic values ​​of meteorological elements for each new energy power station within the region to obtain the historical characteristic values ​​of each meteorological element for the region; acquiring historical values ​​of new energy power in the region and, in conjunction with the historical characteristic values ​​of each meteorological element, using several preset characteristic importance assessment methods to obtain several importance scores for each meteorological element; weighted fusion of the several importance scores for each meteorological element to obtain the final importance score for each meteorological element, and screening each meteorological element based on the final importance score. This method can screen out meteorological elements that are highly correlated with and representative of new energy power prediction, thereby supporting the accuracy of power prediction.

Description

Technical Field

[0001] This invention belongs to the field of power meteorological forecasting technology, and relates to a method and related device for screening meteorological elements for predicting new energy power. Background Technology

[0002] With the rapid growth of installed capacity of new energy sources such as wind power and photovoltaics, their power output exhibits significant randomness, intermittency, and strong weather dependence, placing higher demands on the safe and stable operation of the power system, power balance, and dispatch decisions. New energy power forecasting, as a core supporting technology for new energy grid connection and consumption management, directly impacts the formulation of grid dispatch plans, the allocation of reserve capacity, and the efficiency of the electricity market operation.

[0003] Existing renewable energy power prediction technologies mainly include physical mechanism models, statistical models, and artificial intelligence models. Among them, renewable energy power prediction methods based on artificial intelligence models can demonstrate good performance in short- and medium-term predictions by exploring the nonlinear relationship between meteorological factors and power. However, such methods are highly dependent on large-scale, high-quality renewable energy power prediction training data.

[0004] However, in practical applications, due to the wide variety of meteorological elements, when constructing training data for new energy power prediction, most rely on experience to select meteorological elements, lacking a unified and robust meteorological element screening mechanism. This results in a large data dimension in the constructed dataset and unnecessary noise interference, making it difficult to ensure that the dataset can truly reflect the complex relationship between meteorological changes and new energy power. Consequently, the prediction ability of the new energy power prediction model trained based on this dataset is low, which is not conducive to new energy regulation. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and related apparatus for screening meteorological elements for predicting new energy power.

[0006] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, the present invention provides a method for screening meteorological elements for predicting new energy power, comprising: acquiring historical values ​​of installed capacity of each new energy power station in a region and historical characteristic values ​​of each meteorological element; based on the historical values ​​of installed capacity, performing a weighted average of the historical characteristic values ​​of each meteorological element of each new energy power station in the region to obtain historical characteristic values ​​of each meteorological element in the region; acquiring historical values ​​of new energy power in the region, and combining them with the historical characteristic values ​​of each meteorological element in the region, using a preset method for assessing the importance of several features to obtain several importance scores for each meteorological element; weighting and fusing the several importance scores of each meteorological element to obtain a final importance score for each meteorological element, and screening each meteorological element based on the final importance score of each meteorological element.

[0007] Optionally, obtaining the historical characteristic values ​​of each meteorological element of each new energy power station in the region includes: obtaining the historical analysis values ​​of each meteorological element in the grid distribution of the region; and for each new energy power station, obtaining several neighboring grids corresponding to each new energy power station in the grid distribution, and obtaining the historical analysis values ​​of each meteorological element of several neighboring grids and averaging them to obtain the historical characteristic values ​​of each meteorological element of each new energy power station.

[0008] Optionally, the acquisition of historical analysis values ​​of meteorological elements in the grid distribution of the region includes: acquiring historical analysis values ​​of meteorological elements in the grid distribution of the region, identifying outlier values ​​and outlier values ​​based on spatiotemporal consistency verification, obtaining and removing several outliers, and performing interpolation completion processing based on the time resolution of the region's historical new energy power values.

[0009] Optionally, obtaining the historical value of renewable energy power in the region includes: obtaining the historical value of renewable energy power of each renewable energy power station in the region and completing the missing values; wherein, when the time interval of the missing value is less than a preset time interval threshold, the missing value is completed from both sides by constructing a long short-term memory network time series model; when the time interval of the missing value is not less than the preset time interval threshold, renewable energy power stations whose power correlation coefficient with the current renewable energy power station exceeds the preset power correlation coefficient threshold are obtained, several associated renewable energy power stations are obtained, and the historical value of renewable energy power of the several associated renewable energy power stations is completed using a normalized power method; based on the completion results of the historical value of renewable energy power of each renewable energy power station in the region, the historical value of renewable energy power of the region is obtained.

[0010] Optionally, the preset feature importance evaluation methods include at least two of the following: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance.

[0011] Optionally, the weighted fusion of several importance scores for each meteorological element includes: obtaining the standard deviation of the meteorological element importance ranking based on the cross-validation consistency method for each feature importance assessment method, as the reliability importance score of each feature importance assessment method; and obtaining the weight of each feature importance assessment method according to the reliability importance score of each feature importance assessment method using the following formula:

[0012] in, For the first m The weights of the feature importance assessment methods For the first m The reliability and importance score of the feature importance assessment method For the first k The reliability and importance score of the feature importance assessment method K This refers to the number of methods for assessing feature importance.

[0013] The importance scores of each meteorological element are standardized and then weighted and fused according to the weights of the importance assessment methods of each feature to obtain the final importance score of each meteorological element.

[0014] Optionally, the preset feature importance assessment methods include: a feature importance assessment algorithm based on mutual information, a feature importance assessment algorithm based on extreme gradient boosting, and a feature importance assessment algorithm based on random forest permutation importance; when weighting and fusing the importance scores of each meteorological element, the weight range of the feature importance assessment algorithm based on mutual information is [0.15, 0.25], and the weight range of the feature importance assessment algorithm based on extreme gradient boosting is... The weight range of the feature importance evaluation algorithm based on random forest permutation importance is [0.20, 0.30], and the sum of the weights of the three feature importance evaluation methods is 1.

[0015] In a second aspect, the present invention provides a meteorological element screening system for predicting new energy power output, comprising: a data acquisition module for acquiring historical values ​​of installed capacity and historical characteristic values ​​of meteorological elements of each new energy power station within a region; a regionalization module for weighted averaging of historical characteristic values ​​of meteorological elements of each new energy power station within the region based on historical values ​​of installed capacity, thereby obtaining historical characteristic values ​​of meteorological elements of the region; an importance scoring module for acquiring historical values ​​of new energy power output in the region, and combining them with historical characteristic values ​​of meteorological elements of the region, using a preset method for assessing the importance of several features, thereby obtaining several importance scores for each meteorological element; and a screening module for weighted fusion of the several importance scores of each meteorological element to obtain a final importance score for each meteorological element, and for screening each meteorological element based on the final importance score.

[0016] Optionally, obtaining the historical characteristic values ​​of each meteorological element of each new energy power station in the region includes: obtaining the historical analysis values ​​of each meteorological element in the grid distribution of the region; and for each new energy power station, obtaining several neighboring grids corresponding to each new energy power station in the grid distribution, and obtaining the historical analysis values ​​of each meteorological element of several neighboring grids and averaging them to obtain the historical characteristic values ​​of each meteorological element of each new energy power station.

[0017] Optionally, the acquisition of historical analysis values ​​of meteorological elements in the grid distribution of the region includes: acquiring historical analysis values ​​of meteorological elements in the grid distribution of the region, identifying outlier values ​​and outlier values ​​based on spatiotemporal consistency verification, obtaining and removing several outliers, and performing interpolation completion processing based on the time resolution of the region's historical new energy power values.

[0018] Optionally, obtaining the historical value of renewable energy power in the region includes: obtaining the historical value of renewable energy power of each renewable energy power station in the region and completing the missing values; wherein, when the time interval of the missing value is less than a preset time interval threshold, the missing value is completed from both sides by constructing a long short-term memory network time series model; when the time interval of the missing value is not less than the preset time interval threshold, renewable energy power stations whose power correlation coefficient with the current renewable energy power station exceeds the preset power correlation coefficient threshold are obtained, several associated renewable energy power stations are obtained, and the historical value of renewable energy power of the several associated renewable energy power stations is completed using a normalized power method; based on the completion results of the historical value of renewable energy power of each renewable energy power station in the region, the historical value of renewable energy power of the region is obtained.

[0019] Optionally, the preset feature importance evaluation methods include at least two of the following: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance.

[0020] Optionally, the weighted fusion of several importance scores for each meteorological element includes: obtaining the standard deviation of the meteorological element importance ranking based on the cross-validation consistency method for each feature importance assessment method, as the reliability importance score of each feature importance assessment method; and obtaining the weight of each feature importance assessment method according to the reliability importance score of each feature importance assessment method using the following formula:

[0021] in, For the first m The weights of the feature importance assessment methods For the first m The reliability and importance score of the feature importance assessment method For the first k The reliability and importance score of the feature importance assessment method K This refers to the number of methods for assessing feature importance.

[0022] Based on the weights of the importance assessment methods for each feature, the importance scores of each meteorological element are weighted and fused to obtain the final importance score of each meteorological element.

[0023] Optionally, the preset feature importance assessment methods include: a feature importance assessment algorithm based on mutual information, a feature importance assessment algorithm based on extreme gradient boosting, and a feature importance assessment algorithm based on random forest permutation importance; when weighting and fusing the importance scores of each meteorological element, the weight range of the feature importance assessment algorithm based on mutual information is [0.15, 0.25], and the weight range of the feature importance assessment algorithm based on extreme gradient boosting is... The weight range of the feature importance evaluation algorithm based on random forest permutation importance is [0.20, 0.30], and the sum of the weights of the three feature importance evaluation methods is 1.

[0024] In a third aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for screening meteorological elements for predicting new energy power.

[0025] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for screening meteorological elements for predicting new energy power.

[0026] Compared with the prior art, the present invention has the following beneficial effects: This invention presents a method for screening meteorological elements in new energy power prediction. Based on the historical characteristic values ​​of various meteorological elements at each new energy power station within a region, and using historical installed capacity values, a weighted average of the historical characteristic values ​​of each meteorological element at each new energy power station within the region is calculated to obtain the historical characteristic values ​​of each meteorological element for the region. This ensures a balanced match between meteorological elements and new energy power at the regional level. Simultaneously, multiple characteristic importance assessment methods are comprehensively considered for importance scoring, and a weighted fusion is performed to obtain the final importance score for each meteorological element. This effectively reduces the sensitivity of a single method to sample distribution and outliers, ensuring high stability of the final importance score of meteorological elements across different power stations, seasons, and weather patterns, thus improving the versatility of the input configuration for new energy power prediction models. Finally, based on the final importance score of each meteorological element, the model selects the most dynamic features contributing to renewable energy power from multi-source, multi-scale meteorological elements. This avoids omissions or biases caused by manual selection based on experience, improving the objectivity and consistency of feature selection. Furthermore, by eliminating low-importance or redundant features, the model's input dimensionality can be effectively reduced, training time and computational resource consumption can be decreased, and the risk of overfitting can be reduced, improving the generalization ability of the renewable energy power prediction model. The selected meteorological elements can be directly used for the configuration of key factors and quality control strategies for renewable energy power prediction, fully ensuring the accuracy of renewable energy power prediction, and effectively supporting the formulation of power grid dispatch plans, the configuration of reserve capacity, and the improvement of power market operation efficiency. Attached Figure Description

[0027] Figure 1 This is a flowchart of the meteorological element screening method for new energy power prediction according to an embodiment of the present invention.

[0028] Figure 2 This is a structural block diagram of the meteorological element screening system for new energy power prediction according to an embodiment of the present invention. Detailed Implementation

[0029] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] The present invention will now be described in further detail with reference to the accompanying drawings: See Figure 1 In one embodiment of the present invention, a method for screening meteorological elements for new energy power prediction is provided, which can screen out key meteorological elements that are highly relevant to and representative of new energy power prediction from a large number of meteorological elements, so as to improve the quality of the training dataset for new energy power prediction and provide effective input for subsequent modeling.

[0032] Specifically, the meteorological element screening method for new energy power prediction of the present invention includes the following steps: S1: Obtain the historical values ​​of the installed capacity of each new energy power station in the region and the historical characteristic values ​​of each meteorological element.

[0033] S2: Based on the historical value of installed capacity, the historical characteristic values ​​of each meteorological element of each new energy power station in the region are weighted and averaged to obtain the historical characteristic values ​​of each meteorological element in the region.

[0034] S3: Obtain the historical value of new energy power in the region, and combine it with the historical characteristic values ​​of various meteorological elements in the region. Then, use several preset characteristic importance assessment methods to obtain several importance scores for each meteorological element.

[0035] S4: Weight and merge the importance scores of each meteorological element to obtain the final importance score of each meteorological element, and then screen each meteorological element based on the final importance score of each meteorological element.

[0036] This invention presents a method for screening meteorological elements in new energy power prediction. Based on the historical characteristic values ​​of various meteorological elements at each new energy power station within a region, and using historical installed capacity values, a weighted average of the historical characteristic values ​​of each meteorological element at each new energy power station within the region is calculated to obtain the historical characteristic values ​​of each meteorological element for the region. This ensures a balanced match between meteorological elements and new energy power at the regional level. Simultaneously, multiple characteristic importance assessment methods are comprehensively considered for importance scoring, and a weighted fusion is performed to obtain the final importance score for each meteorological element. This effectively reduces the sensitivity of a single method to sample distribution and outliers, ensuring high stability of the final importance score of meteorological elements across different power stations, seasons, and weather patterns, thus improving the versatility of the input configuration for new energy power prediction models. Finally, based on the final importance score of each meteorological element, the model selects the most dynamic features contributing to renewable energy power from multi-source, multi-scale meteorological elements. This avoids omissions or biases caused by manual selection based on experience, improving the objectivity and consistency of feature selection. Furthermore, by eliminating low-importance or redundant features, the model's input dimensionality can be effectively reduced, training time and computational resource consumption can be decreased, and the risk of overfitting can be reduced, improving the generalization ability of the renewable energy power prediction model. The selected meteorological elements can be directly used for the configuration of key factors and quality control strategies for renewable energy power prediction, fully ensuring the accuracy of renewable energy power prediction, and effectively supporting the formulation of power grid dispatch plans, the configuration of reserve capacity, and the improvement of power market operation efficiency.

[0037] This study interprets the impact of regional meteorological factors' weighted regional installed capacity on the total power output of regional renewable energy sources. It calculates the importance of regional weighted meteorological factors for renewable energy output by considering the installed capacity of renewable energy power plants in the region. Using historical renewable energy power data from the past three years and monthly dynamically changing installed capacity, a weighting is applied based on the ratio of installed capacity to total installed capacity, as shown in the following formula:

[0038]

[0039] in, Indicates the first i The first new energy power station t Installed capacity at any given time Indicates the first i A new energy station t The newly added installed capacity at any given time indicates that the installed capacity of new energy power plants is constantly changing. Indicates the first i The first new energy power station t Installed capacity at time -1 This represents the weighted value of various meteorological elements in the region. Indicates new energy power station i exist t Historical characteristic values ​​of meteorological elements at any given time n This refers to the number of new energy power stations.

[0040] In one possible implementation, obtaining the historical characteristic values ​​of each meteorological element of each new energy power station in the region includes: obtaining the historical analysis values ​​of each meteorological element in the grid distribution of the region; and for each new energy power station, obtaining several neighboring grids corresponding to each new energy power station in the grid distribution, and obtaining the historical analysis values ​​of each meteorological element of several neighboring grids and averaging them to obtain the historical characteristic values ​​of each meteorological element of each new energy power station.

[0041] Explanatoryly, since historical analysis values ​​of meteorological elements are uniformly distributed across a grid, while renewable energy power data is discretely distributed, this embodiment employs a nearest-neighbor averaging method to match the spatial location of meteorological elements with renewable energy power. For example, for each renewable energy power station, four neighboring grids corresponding to each station in the grid distribution are obtained, and the historical analysis values ​​of each meteorological element from several neighboring grids are averaged to serve as the historical characteristic values ​​of each meteorological element for that renewable energy power station.

[0042] in, Representing new energy power stations P Historical characteristic values ​​of various meteorological elements Representing new energy power stations P The space near the first k Historical characteristic values ​​of each meteorological element in each grid.

[0043] In one possible implementation, the acquisition of historical analysis values ​​of meteorological elements in the grid distribution of the region includes: acquiring historical analysis values ​​of meteorological elements in the grid distribution of the region, identifying outlier values ​​and outlier values ​​based on spatiotemporal consistency verification, obtaining and removing several outliers, and performing interpolation completion processing based on the temporal resolution of the region's historical new energy power values.

[0044] Interpretive analysis of historical meteorological elements within a region's grid distribution is primarily achieved by collecting published meteorological reanalysis data covering the target area. For example, ERA5 data from the European Centre for Meteorological Research, FNL data from the United States, and CMA-RA data from China are collected, focusing spatially on the entire region and temporally covering the most recent five years of historical meteorological reanalysis data.

[0045] For interpretative purposes, outlier identification and removal are necessary for directly obtained historical meteorological element values. Specifically, identifying outliers includes setting thresholds based on the physical range of meteorological elements, identifying data exceeding reasonable ranges, such as traversing historical analysis values ​​of each meteorological element, marking outliers exceeding the threshold, and recording the outliers and their corresponding timestamps. Outlier identification based on spatiotemporal consistency verification includes: spatially, temperature variables at adjacent pressure levels (e.g., 850 hPa and 700 hPa pressure layers) at the same time must satisfy a vertical temperature lapse rate (e.g., a temperature decrease of approximately 6.5°C for every 100 hPa increase), with deviations exceeding 2°C considered outliers; temporally, hourly variables (e.g., hourly surface pressure) must have changes of less than 5 hPa between adjacent times (e.g., a sudden change exceeding 10 hPa is considered an outlier, such as a sudden change from 1010 hPa to 990 hPa), identifying outliers by calculating the difference between adjacent data.

[0046] Explanatoryly, to address the issue of differing time resolutions between meteorological element data and renewable energy power data, meteorological element data can be interpolated and supplemented according to the time resolution of the regional renewable energy power data. For example, based on the required synchronization time, spline interpolation can be used to achieve time synchronization of multi-source data and supplement outliers and missing data points.

[0047] In one possible implementation, obtaining the historical value of renewable energy power in the region includes: obtaining the historical value of renewable energy power of each renewable energy power station in the region and completing missing values; wherein, when the time interval of missing values ​​is less than a preset time interval threshold, the missing values ​​are completed from both sides by constructing a long short-term memory network time series model; when the time interval of missing values ​​is not less than the preset time interval threshold, renewable energy power stations whose power correlation coefficient with the current renewable energy power station exceeds the preset power correlation coefficient threshold are obtained, several associated renewable energy power stations are obtained, and the historical value of renewable energy power of several associated renewable energy power stations is completed using a normalized power method; based on the completion results of the historical value of renewable energy power of each renewable energy power station in the region, the historical value of renewable energy power of the region is obtained.

[0048] Interpretive historical values ​​of renewable energy power primarily include data from wind farms and solar PV plants. These historical values ​​are determined by collecting historical data from both. Specifically, historical wind farm data includes power data with a 15-minute time resolution, the installed capacity of different wind farms, and information on the hub height, wind speed, and latitude / longitude of the wind farms as measured by their anemometers. Historical solar PV plant data includes photovoltaic power data with a 15-minute time resolution, the installed capacity of different solar PV plants, and irradiance data measured by the irradiance meters at the plants, along with the latitude / longitude of the plants.

[0049] Meanwhile, outliers and missing values ​​that are clearly present in the historical renewable energy power values ​​of renewable energy power plants are marked. Then, outliers are deleted to achieve preliminary control of data quality, and missing values ​​are filled in to ensure the data quality of the final historical renewable energy power values.

[0050] Specifically, for missing values ​​within a long time window, based on relevant information from a single nearby renewable energy power station and multiple nearby renewable energy power stations, the time interval for the missing values ​​is determined. For missing values, data repair is performed from both sides by constructing an LSTM (Long Short-Term Memory) time series model. For missing values, new energy power plants with a correlation coefficient exceeding λ>0.8 in the nearest power range are used, and normalized power is used for completion. That is, the missing values ​​are predicted and completed based on the output per unit installed capacity.

[0051] In one possible implementation, the preset feature importance evaluation methods include at least two of the following: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance.

[0052] Explanatoryly, due to the limitations of a single method, the method of this invention considers multiple feature importance assessment methods, and obtains the final importance score of each meteorological element by integrating the assessment results of multiple feature importance assessment methods, thus ensuring the accurate analysis of the importance of each meteorological element.

[0053] Interpretive mutual information algorithms, based on information theory, effectively capture the nonlinear dependency between meteorological elements and power by calculating the difference between the joint probability distribution and the marginal probability distribution. They are not limited by dimensions or functional forms and provide a fundamental correlation assessment. Extreme gradient boosting (XGBoost) algorithms, based on a gradient boosting tree framework, quantify the actual contribution of features to prediction performance with high precision by statistically analyzing the reduction in loss function (i.e., gain) caused by each meteorological element splitting at a decision tree node and summing these values. They perform excellently in complex nonlinear relationships. Random forest permutation importance algorithms measure the importance of a meteorological element by randomly shuffling its value and observing the change in the model's prediction error. If the error increases significantly after shuffling, it indicates that the meteorological element is crucial for power prediction. This method is intuitive, robust, and effectively verifies the true influence of features and resists outlier interference.

[0054] For example, in the application of various feature importance assessment methods, for the feature importance assessment algorithm based on mutual information, the focus is on capturing the nonlinear mapping relationship between irradiance and photovoltaic power, as well as the shading effect of snowfall on irradiance. Hourly irradiance and photovoltaic power data can be processed through continuous variable kernel density estimation. For the feature importance assessment algorithm based on extreme gradient boosting, considering the diurnal intermittency of photovoltaic output, nighttime output data can be removed, and a weighted mean square error loss function can be used to increase the contribution of effective daytime data to feature importance. This can focus on quantifying the explanatory power of factors such as irradiance and snowfall on power fluctuations. For the feature importance assessment algorithm based on random forest permutation importance, the impact of derived features such as snowfall coverage duration and peak irradiance periods can be evaluated first by permutation importance, while retaining the evaluation of original meteorological elements, taking into account both direct and indirect effects, and realizing feature importance calculation.

[0055] In one possible implementation, the weighted fusion of several importance scores for each meteorological element includes: obtaining the standard deviation of the meteorological element importance ranking based on the cross-validation consensus method for each feature importance assessment method, as the reliability importance score of each feature importance assessment method; and obtaining the weight of each feature importance assessment method according to the reliability importance score of each feature importance assessment method using the following formula:

[0056] in, For the first m The weights of the feature importance assessment methods For the first m The reliability and importance score of the feature importance assessment method For the first k The reliability and importance score of the feature importance assessment method KThis refers to the number of methods for assessing feature importance.

[0057] The importance scores of each meteorological element are standardized and then weighted and fused according to the weights of the importance assessment methods of each feature to obtain the final importance score of each meteorological element.

[0058] Interpretive, the weights of each feature importance assessment method are determined based on its "reliability" in the new energy power prediction scenario. Reliability can be measured using a cross-validation consensus method. For example, in this embodiment, 5-fold cross-validation is used as the cross-validation consensus method.

[0059] For explanatory purposes, since the importance values ​​of different feature assessment methods vary considerably, the reliability importance scores of each method need to be min-max standardized to eliminate the influence of dimensions.

[0060] in, For the first m The first method for assessing the importance of features i Standardized reliability importance score for each meteorological element. For the first m The first method for assessing the importance of features i Reliability and importance score of each meteorological element q This refers to the quantity of meteorological elements.

[0061] Interpretively, the final importance score for each meteorological element is obtained by multiplying several standardized reliability importance scores of each meteorological element by the weights of the corresponding feature importance assessment methods and summing the results.

[0062] in, For the first i The final importance score of each meteorological element.

[0063] In one possible implementation, the preset feature importance evaluation methods include: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance.

[0064] When weighting and fusing the importance scores of various meteorological elements, the weight range of the feature importance evaluation algorithm based on mutual information is [0.15, 0.25], and the weight range of the feature importance evaluation algorithm based on extreme gradient boosting is... The weight range of the feature importance evaluation algorithm based on random forest permutation importance is [0.20, 0.30], and the sum of the weights of the three feature importance evaluation methods is 1.

[0065] Interpretive feature importance evaluation algorithms based on mutual information are appropriate weights as a basic statistical method; feature importance evaluation algorithms based on extreme gradient boosting are highly accurate, so their weights are set to the highest; feature importance evaluation algorithms based on random forest permutation importance are robust, so their weights are set to relatively high.

[0066] In one possible implementation, based on provincial-level renewable energy power data and renewable energy power station information for a specific province, the final importance score of each meteorological element is calculated for both the total provincial renewable energy power data and meteorological element data. For wind power meteorological elements, the final importance score ranking is: 100m wind speed (score 0.89) > 10m wind speed (score 0.56) > surface pressure (score 0.19) > temperature (score 0.12), which conforms to the physical correlation between wind power and wind speed and provides data support for feature selection in the power prediction model. For photovoltaic meteorological elements, the final importance score ranking is: irradiance (score 0.91) > snowfall (score 0.58) > surface pressure (score 0.21) > temperature (score 0.16), which also conforms to the physical correlation between photovoltaic power and irradiance.

[0067] In summary, the main purpose of the meteorological element screening method for new energy power prediction in this invention is to select key meteorological elements that are highly correlated with and representative of new energy power prediction from a large number of meteorological elements, thereby improving the quality of the dataset and providing effective input for subsequent modeling. It has the following characteristics: 1. It can remove irrelevant or weakly correlated meteorological elements, avoiding unnecessary noise interference, reducing data dimensionality, and improving the efficiency of model training. 2. By selecting the most influential meteorological elements, it ensures that the dataset can truly reflect the complex relationship between meteorological changes and new energy power, thereby improving the model's predictive ability. 3. Through the fusion of multiple feature importance evaluation algorithms, it ensures that the selected meteorological elements have good applicability and stability in different regions and different types of new energy scenarios, enhancing the model's generalization ability under different environments.

[0068] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not disclosed in the apparatus embodiments, please refer to the embodiments of the method of the present invention.

[0069] See Figure 2In another embodiment of the present invention, a meteorological element screening system for new energy power prediction is provided, which can be used to implement the above-mentioned meteorological element screening method for new energy power prediction. Specifically, the meteorological element screening system for new energy power prediction includes a data acquisition module, a regionalization module, an importance scoring module, and a screening module.

[0070] The system comprises the following modules: a data acquisition module for acquiring historical values ​​of installed capacity and meteorological features of each renewable energy power station within the region; a regionalization module for weighted averaging of historical meteorological features of each renewable energy power station within the region based on historical installed capacity values; an importance scoring module for acquiring historical renewable energy power values ​​within the region and, in conjunction with historical meteorological features of each meteorological feature, using several preset importance assessment methods to obtain several importance scores for each meteorological feature; and a screening module for weighted fusion of several importance scores for each meteorological feature to obtain the final importance score for each meteorological feature, and for screening each meteorological feature based on the final importance score.

[0071] In one possible implementation, obtaining the historical characteristic values ​​of each meteorological element of each new energy power station in the region includes: obtaining the historical analysis values ​​of each meteorological element in the grid distribution of the region; and for each new energy power station, obtaining several neighboring grids corresponding to each new energy power station in the grid distribution, and obtaining the historical analysis values ​​of each meteorological element of several neighboring grids and averaging them to obtain the historical characteristic values ​​of each meteorological element of each new energy power station.

[0072] In one possible implementation, the acquisition of historical analysis values ​​of meteorological elements in the grid distribution of the region includes: acquiring historical analysis values ​​of meteorological elements in the grid distribution of the region, identifying outlier values ​​and outlier values ​​based on spatiotemporal consistency verification, obtaining and removing several outliers, and performing interpolation completion processing based on the temporal resolution of the region's historical new energy power values.

[0073] In one possible implementation, obtaining the historical value of renewable energy power in the region includes: obtaining the historical value of renewable energy power of each renewable energy power station in the region and completing missing values; wherein, when the time interval of missing values ​​is less than a preset time interval threshold, the missing values ​​are completed from both sides by constructing a long short-term memory network time series model; when the time interval of missing values ​​is not less than the preset time interval threshold, renewable energy power stations whose power correlation coefficient with the current renewable energy power station exceeds the preset power correlation coefficient threshold are obtained, several associated renewable energy power stations are obtained, and the historical value of renewable energy power of several associated renewable energy power stations is completed using a normalized power method; based on the completion results of the historical value of renewable energy power of each renewable energy power station in the region, the historical value of renewable energy power of the region is obtained.

[0074] In one possible implementation, the preset feature importance evaluation methods include at least two of the following: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance.

[0075] In one possible implementation, the weighted fusion of several importance scores for each meteorological element includes: obtaining the standard deviation of the meteorological element importance ranking based on the cross-validation consensus method for each feature importance assessment method, as the reliability importance score of each feature importance assessment method; and obtaining the weight of each feature importance assessment method according to the reliability importance score of each feature importance assessment method using the following formula:

[0076] in, For the first m The weights of the feature importance assessment methods For the first m The reliability and importance score of the feature importance assessment method For the first k The reliability and importance score of the feature importance assessment method K This refers to the number of methods for assessing feature importance.

[0077] Based on the weights of the importance assessment methods for each feature, the importance scores of each meteorological element are weighted and fused to obtain the final importance score of each meteorological element.

[0078] In one possible implementation, the preset feature importance assessment methods include: a feature importance assessment algorithm based on mutual information, a feature importance assessment algorithm based on extreme gradient boosting, and a feature importance assessment algorithm based on random forest permutation importance; when weighting and fusing the importance scores of each meteorological element, the weight range of the feature importance assessment algorithm based on mutual information is [0.15, 0.25], and the weight range of the feature importance assessment algorithm based on extreme gradient boosting is... The weight range of the feature importance evaluation algorithm based on random forest permutation importance is [0.20, 0.30], and the sum of the weights of the three feature importance evaluation methods is 1.

[0079] All relevant content of each step involved in the aforementioned embodiments of the meteorological element screening method for new energy power prediction can be referenced to the functional description of the corresponding functional module of the meteorological element screening system for new energy power prediction in the embodiments of the present invention, and will not be repeated here.

[0080] The module division in this embodiment of the invention is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of the invention can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0081] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used for the operation of a method for screening meteorological elements for new energy power prediction.

[0082] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory). This computer-readable storage medium is a memory device within a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium of the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space containing the terminal's operating system. Furthermore, this storage space also contains one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the meteorological element screening method for new energy power prediction in the above embodiments.

[0083] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0085] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0086] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for screening meteorological elements for new energy power prediction, characterized in that, include: Obtain historical values ​​of installed capacity and historical characteristic values ​​of various meteorological elements for each new energy power station in the region; Based on the historical value of installed capacity, the historical characteristic values ​​of each meteorological element of each new energy power station in the region are weighted and averaged to obtain the historical characteristic values ​​of each meteorological element in the region. The historical values ​​of new energy power in the region are obtained, and combined with the historical characteristic values ​​of various meteorological elements in the region, several preset characteristic importance assessment methods are used to obtain several importance scores for each meteorological element. The importance scores of each meteorological element are weighted and fused to obtain the final importance score of each meteorological element, and the meteorological elements are screened based on the final importance scores of each meteorological element.

2. The method for screening meteorological elements for new energy power prediction according to claim 1, characterized in that, The historical characteristic values ​​of various meteorological elements of each new energy power station in the acquisition area include: Historical analysis values ​​of meteorological elements in the grid distribution of the region are obtained; and for each new energy power station, several neighboring grids corresponding to each new energy power station in the grid distribution are obtained, and the historical analysis values ​​of meteorological elements in several neighboring grids are obtained and averaged to obtain the historical characteristic values ​​of meteorological elements of each new energy power station.

3. The method for screening meteorological elements for new energy power prediction according to claim 2, characterized in that, The historical analysis values ​​of each meteorological element in the grid distribution of the acquired area include: Historical analysis values ​​of meteorological elements in the grid distribution of the region are obtained, and outlier identification and spatiotemporal consistency verification are performed. Several outliers are obtained and removed, and interpolation is performed based on the temporal resolution of the region's historical new energy power values.

4. The method for screening meteorological elements for new energy power prediction according to claim 1, characterized in that, The historical values ​​of renewable energy power in the acquired area include: The system acquires historical renewable energy power values ​​for each renewable energy power station within the region and completes any missing values. When the time interval of a missing value is less than a preset time interval threshold, a long short-term memory network time series model is constructed to complete the missing value from both sides. When the time interval of a missing value is not less than the preset time interval threshold, renewable energy power stations whose power correlation coefficient with the current renewable energy power station exceeds a preset power correlation coefficient threshold are acquired. Several associated renewable energy power stations are obtained, and the system completes the missing value using normalized power based on the historical renewable energy power values ​​of these associated renewable energy power stations. Based on the completion processing results of the historical renewable energy power values ​​of each renewable energy power station in the region, the historical renewable energy power values ​​of the region are obtained.

5. The method for screening meteorological elements for new energy power prediction according to claim 1, characterized in that, The preset methods for evaluating the importance of certain features include at least two of the following: Feature importance evaluation algorithms based on mutual information, extreme gradient boosting, and random forest permutation importance are all available.

6. The method for screening meteorological elements for predicting new energy power according to claim 1, characterized in that, The weighted fusion of several importance scores for various meteorological elements includes: The standard deviation of the meteorological element importance ranking based on the cross-validation consensus method is obtained for each feature importance assessment method, and used as the reliability importance score of each method. Based on the reliability importance score, the weight of each feature importance assessment method is obtained using the following formula: in, For the first m The weights of the feature importance assessment methods For the first m The reliability and importance score of the feature importance assessment method For the first k The reliability and importance score of the feature importance assessment method K The number of feature importance assessment methods; The importance scores of each meteorological element are standardized and then weighted and fused according to the weights of the importance assessment methods of each feature to obtain the final importance score of each meteorological element.

7. The method for screening meteorological elements for new energy power prediction according to claim 1, characterized in that, The preset feature importance evaluation methods include: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance. When weighting and fusing the importance scores of various meteorological elements, the weight range of the feature importance evaluation algorithm based on mutual information is [0.15, 0.25], and the weight range of the feature importance evaluation algorithm based on extreme gradient boosting is... The weight range of the feature importance evaluation algorithm based on random forest permutation importance is [0.20, 0.30], and the sum of the weights of the three feature importance evaluation methods is 1.

8. A meteorological element screening system for predicting new energy power, characterized in that, include: The data acquisition module is used to acquire historical values ​​of the installed capacity of each new energy power station in the region and historical characteristic values ​​of each meteorological element; The regionalization module is used to perform weighted averages of the historical characteristic values ​​of each meteorological element of each new energy power station in the region based on the historical value of installed capacity, so as to obtain the historical characteristic values ​​of each meteorological element in the region. The importance scoring module is used to obtain the historical value of new energy power in the region, and combined with the historical characteristic values ​​of various meteorological elements in the region, it uses several preset characteristic importance assessment methods to obtain several importance scores for each meteorological element. The filtering module is used to weight and fuse several importance scores of each meteorological element to obtain the final importance score of each meteorological element, and to filter each meteorological element based on the final importance score.

9. The meteorological element screening system for new energy power prediction according to claim 1, characterized in that, The historical characteristic values ​​of various meteorological elements of each new energy power station in the acquisition area include: Historical analysis values ​​of meteorological elements in the grid distribution of the region are obtained; and for each new energy power station, several neighboring grids corresponding to each new energy power station in the grid distribution are obtained, and the historical analysis values ​​of meteorological elements in several neighboring grids are obtained and averaged to obtain the historical characteristic values ​​of meteorological elements of each new energy power station.

10. The meteorological element screening system for new energy power prediction according to claim 9, characterized in that, The historical analysis values ​​of each meteorological element in the grid distribution of the acquired area include: Historical analysis values ​​of meteorological elements in the grid distribution of the region are obtained, and outlier identification and spatiotemporal consistency verification are performed. Several outliers are obtained and removed, and interpolation is performed based on the temporal resolution of the region's historical new energy power values.

11. The meteorological element screening system for new energy power prediction according to claim 8, characterized in that, The historical values ​​of renewable energy power in the acquired area include: The system acquires historical renewable energy power values ​​for each renewable energy power station within the region and completes any missing values. When the time interval of a missing value is less than a preset time interval threshold, a long short-term memory network time series model is constructed to complete the missing value from both sides. When the time interval of a missing value is not less than the preset time interval threshold, renewable energy power stations whose power correlation coefficient with the current renewable energy power station exceeds a preset power correlation coefficient threshold are acquired. Several associated renewable energy power stations are obtained, and the system completes the missing value using normalized power based on the historical renewable energy power values ​​of these associated renewable energy power stations. Based on the completion processing results of the historical renewable energy power values ​​of each renewable energy power station in the region, the historical renewable energy power values ​​of the region are obtained.

12. The meteorological element screening system for new energy power prediction according to claim 8, characterized in that, The preset methods for evaluating the importance of certain features include at least two of the following: Feature importance evaluation algorithms based on mutual information, extreme gradient boosting, and random forest permutation importance are all available.

13. The meteorological element screening system for new energy power prediction according to claim 8, characterized in that, The weighted fusion of several importance scores for various meteorological elements includes: The standard deviation of the meteorological element importance ranking based on the cross-validation consensus method is obtained for each feature importance assessment method, and used as the reliability importance score of each method. Based on the reliability importance score, the weight of each feature importance assessment method is obtained using the following formula: in, For the first m The weights of the feature importance assessment methods For the first m The reliability and importance score of the feature importance assessment method For the first k The reliability and importance score of the feature importance assessment method K The number of feature importance assessment methods; Based on the weights of the importance assessment methods for each feature, the importance scores of each meteorological element are weighted and fused to obtain the final importance score of each meteorological element.

14. The meteorological element screening system for new energy power prediction according to claim 8, characterized in that, The preset feature importance evaluation methods include: a feature importance evaluation algorithm based on mutual information, a feature importance evaluation algorithm based on extreme gradient boosting, and a feature importance evaluation algorithm based on random forest permutation importance. When weighting and fusing the importance scores of various meteorological elements, the weight range of the feature importance evaluation algorithm based on mutual information is [0.15, 0.25], and the weight range of the feature importance evaluation algorithm based on extreme gradient boosting is... The weight range of the feature importance evaluation algorithm based on random forest permutation importance is [0.20, 0.30], and the sum of the weights of the three feature importance evaluation methods is 1.

15. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the meteorological element screening method for new energy power prediction as described in any one of claims 1 to 7.

16. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the meteorological element screening method for new energy power prediction as described in any one of claims 1 to 7.

Images