Meteorological element screening method, system and device based on data-driven grid aggregation and causal sparsity screening and medium

Abstract

This invention belongs to the field of meteorological driving element screening technology, and discloses a method, system, equipment, and medium for meteorological element screening based on data-driven grid aggregation and causal sparse screening, to solve the problems of high-dimensional grid redundancy, spurious correlation interference, difficulty in identifying nonlinear time delays, cross-seasonal instability, and difficulty in verifying the necessity of elements. The method of this invention includes: acquiring and preprocessing target variable, grid meteorological, and calendar data to construct a standardized dataset; performing data-driven aggregation on the grid meteorological data to obtain regional-level sequences, and generating a candidate set based on correlation, partial correlation, and mutual information; performing time-delay causal discrimination and sparse screening on the candidate set, identifying time delays by combining cross-correlation and dynamic time warping, and estimating uncertainty through Bootstrap; performing split-Blu-ray and seasonal stability screening on key elements, quantifying contribution through counterfactual testing, and generating a callable meteorological element library.

Description

Technical Field

[0001] This invention belongs to the field of meteorological driving element screening technology, specifically relating to meteorological element screening methods, systems, equipment and media based on data-driven grid aggregation and causal sparse screening. Background Technology

[0002] In power systems with high penetration of new energy sources, wind power, photovoltaic power output, and load forecasting are significantly driven by meteorological factors. In engineering practice, meteorological data is usually represented as a high-dimensional gridded sequence covering the target area, which presents two prominent problems: First, the high dimensionality and redundancy of the grid points can introduce noise and lead to overfitting if used directly as input; second, simple averaging and other coarse-grained aggregation methods are insufficient to characterize the true differences in contributions from different spatial locations.

[0003] Existing feature selection methods often employ statistical correlation indicators such as correlation and mutual information. However, these indicators cannot distinguish between causal direction and action path, easily incorporating "spurious correlation" elements into the model, leading to significant fluctuations in generalization performance under seasonal changes or extreme weather scenarios. Furthermore, the impact of meteorological factors on wind and solar power output and load typically exhibits uncertain and nonlinear time lags, making it difficult to reliably identify time lag structures using traditional methods with fixed lags or single cross-correlation peaks. More critically, existing methods generally lack cross-seasonal stability constraints and counterfactual necessity verification, hindering the formation of an auditable, reusable, and standardized meteorological element database, thus restricting the standardization of predictive model input management.

[0004] In summary, existing technologies cannot simultaneously solve the problems of high-dimensional grid redundancy, spurious correlation interference, difficulty in identifying nonlinear time delays, lack of cross-seasonal stability, and unverifiable necessity of elements within a unified framework. Summary of the Invention

[0005] Based on the aforementioned shortcomings and deficiencies in the existing technology, one of the objectives of this invention is to at least solve one or more of the aforementioned problems in the existing technology. In other words, one of the objectives of this invention is to provide a meteorological element screening method, system, device, and medium based on data-driven grid aggregation and causal sparse screening that meets one or more of the aforementioned requirements. This aims to automatically complete the end-to-end construction from high-dimensional grid meteorological data to a callable meteorological element library under a unified framework, and to achieve spatial aggregation and dimensionality reduction of meteorological features, causal direction identification, nonlinear time delay identification, cross-seasonal stable screening, and necessity quantification verification. This will improve the interpretability, generalization stability, and engineering reusability of the input features of new energy power and load prediction models.

[0006] To achieve the above-mentioned objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a meteorological element screening method based on data-driven grid aggregation and causal sparse screening, comprising the following steps: S1. Acquire multi-source time series data of the target area and preprocess it to construct a standardized dataset under a unified time index. The multi-source time series data includes target variable data, gridded meteorological data and calendar feature data. S2. Perform data-driven aggregation on the gridded meteorological data to obtain a regional meteorological element sequence, and generate a candidate meteorological element set based on correlation, partial correlation and mutual information. S3. Perform time-delay causal discrimination and sparse screening on the candidate meteorological element set to determine the set of key meteorological elements with directionality, and identify the main time-delay relationships by combining cross-correlation and dynamic time warping. Estimate the uncertainty through Bootstrap and output the element-time-delay structured results. S4. Perform split-bar and seasonal stability screening on the set of key meteorological elements, and quantify the contribution of elements through counterfactual testing to generate a callable meteorological element library containing an element list, grid aggregation mapping, time lag parameters and stability indicators.

[0007] As a preferred approach, the data-driven aggregation of gridded meteorological data in step S2 specifically involves: For each meteorological element in the gridded meteorological data, non-negative aggregate weights are learned on the spatial grid, and spatial total variation constraints are applied to the non-negative aggregate weights to achieve spatial smoothing of the weights of adjacent grid points.

[0008] As a preferred approach, the specific steps for generating the candidate meteorological element set in step S2 are as follows: The correlation strength is calculated based on the correlation coefficient, partial correlation coefficient and mutual information between each element in the regional meteorological element sequence and each variable in the target variable data. If the calculated correlation strength exceeds the corresponding threshold, the element will be included in the candidate meteorological element set.

[0009] As a preferred approach, the time-delay causality determination in step S3 specifically involves: Perform a conditional independent test on each lag order of each element in the candidate meteorological element set, take the lag where the null hypothesis is rejected as the causal effective lag of the element, and include elements with at least one causal effective lag into the key meteorological element set.

[0010] As a preferred approach, the sparse screening in step S3 specifically involves: Group all the lagged terms of each meteorological element in the candidate meteorological element set, and perform joint compression on the coefficients of each group. Elements with a group coefficient of zero are eliminated, while elements with a group coefficient of non-zero are retained.

[0011] As a preferred approach, the identification of the main time delay relationships in step S3 specifically involves: The linear principal time delay is estimated using the cross-correlation function, the nonlinear time delay is estimated using dynamic time warping, and the confidence interval of the time delay is estimated using the Bootstrap resampling method.

[0012] As a preferred option, in step S4: The split-bar screening is based on the Wasserstein distance to construct an uncertain set and evaluate the worst-case performance of the key meteorological element set under distribution perturbation. The seasonal stability screening calculates the importance of each element in the set of key meteorological elements according to the season and constrains its cross-seasonal fluctuations. The counterfactual test quantifies the contribution of a single element by removing it from the set of key meteorological elements and then measuring the decrease in predictive performance.

[0013] Secondly, the present invention provides a meteorological element screening system based on data-driven grid aggregation and causal sparse screening, for implementing the meteorological element screening method described in the first aspect.

[0014] Thirdly, the present invention provides an electronic device, the computer device including a memory, a processor and a computer program, wherein when the computer program is executed by the processor, it implements the meteorological element screening method as described in the first aspect.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the meteorological element screening method as described in the first aspect.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This invention assesses the spatial contribution of high-dimensional meteorological grid data and performs data-driven aggregation processing. By learning non-negative aggregation weights and applying spatial total variation constraints, it can automatically identify key spatial locations and achieve spatial smoothing of the weights of adjacent grid points. Compared to directly using the original grid points or simple average aggregation, this invention effectively reduces the dimensional redundancy of the original meteorological grid data, reduces invalid noise input, and significantly improves the expressive efficiency and physical interpretability of meteorological features.

[0017] This invention constructs a causal-time-delay joint discrimination mechanism and employs conditional independence tests to distinguish between causal relationships and spurious correlations among variables, avoiding the shortcomings of traditional correlation indicators that easily introduce false associations. Based on this, it combines cross-correlation functions to identify linearly dominant time delays and introduces dynamic time warping methods to identify nonlinear time-shift structures. This enables a more accurate characterization of the actual impact paths of meteorological factors on wind power output, photovoltaic power output, and load changes, solving the problem that traditional fixed lags or single cross-correlation peak values ​​are difficult to stably identify time-delay structures.

[0018] This invention employs a sparse screening mechanism to assess the importance of candidate meteorological elements. All lagged terms of each meteorological element are grouped and jointly compressed, automatically eliminating redundant elements. Based on this, Wasserstein bar analysis and cross-seasonal stability tests are combined to evaluate the performance of meteorological elements under different data distribution disturbances and seasonal conditions. This allows for the screening of key meteorological driving factors with long-term stable predictive contributions, significantly improving the generalization stability of features under seasonal changes and extreme weather scenarios.

[0019] This invention introduces a counterfactual testing mechanism, which quantifies the necessary contribution of each meteorological element to the performance of the prediction model by re-evaluating the change in prediction performance after removing a single element from the element set. This mechanism provides a quantifiable verification basis for the element selection results, overcoming the shortcomings of existing methods that only output importance rankings without verifying the necessity of elements, and significantly improving the interpretability and reliability of the selection results.

[0020] Through the organic integration of the above mechanisms, this invention can automatically complete the end-to-end construction from high-dimensional gridded meteorological data to a callable meteorological element library within a unified framework. The generated meteorological element library contains complete information such as element lists, grid aggregation mappings, time delay parameters, and stability indicators, which can provide standardized input features for wind power, photovoltaic, and load forecasting models, significantly improving the stability and reusability of forecasting models in engineering applications under complex meteorological conditions.

[0021] Further or more detailed beneficial effects will be described in conjunction with specific embodiments in the detailed implementation. 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, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the meteorological element screening method described in Embodiment 1 of the present invention.

[0024] Figure 2 This is a schematic diagram of the process of meteorological grid aggregation and candidate element generation as described in Embodiment 1 of the present invention.

[0025] Figure 3 This is a schematic diagram of the causal-time delay joint discrimination process described in Embodiment 1 of the present invention.

[0026] Figure 4This is a schematic diagram of the seasonal stability screening and counterfactual verification process described in Embodiment 1 of the present invention.

[0027] Figure 5 This is a schematic diagram of the meteorological element screening system described in Embodiment 2 of the present invention.

[0028] Figure 6 This is a structural diagram of the electronic device described in Embodiment 3 of the present invention.

[0029] Icon labels: 600. Electronic equipment; 601. Processor; 602. Communication bus; 603. User interface; 604. Network interface; 605. Memory. Detailed Implementation

[0030] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0031] In the following description, several embodiments of the present invention are provided. Different embodiments can be substituted or combined. Therefore, the present invention can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present invention should also be considered to include embodiments containing one or more other possible combinations of A, B, C, and D, even if such embodiments are not explicitly described in the following text.

[0032] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the described elements without departing from the scope of the invention. Various processes or components may be appropriately omitted, substituted, or added to the various examples. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Furthermore, features described with respect to some examples may be combined into other examples.

[0033] To facilitate a better understanding of the embodiments of the present invention, its application scenarios will be explained before providing a detailed explanation of the specific implementation methods.

[0034] The meteorological element screening method described in the embodiments of this specification is applied to meteorological feature engineering scenarios for wind power output forecasting, photovoltaic power output forecasting, and power load forecasting. In these scenarios, the application of the meteorological element screening method aims to automatically screen meteorological elements from high-dimensional gridded meteorological data that have a causal relationship with the target variable, match time lags, are stable across seasons, and contribute necessary meteorological elements. This provides a standardized and reusable input feature library for the prediction model, thereby improving the generalization ability and engineering stability of the prediction model under complex meteorological conditions.

[0035] The following is a brief explanation of the data-driven grid aggregation, causal sparse filtering, meteorological elements, target variable data, gridded meteorological data, calendar feature data, correlation, partial correlation, mutual information, time-delay causal discrimination, sparse filtering, bootstrap, split-Browsing bar, seasonal stability filtering, and counterfactual testing involved in several embodiments of this specification: Data-driven grid aggregation refers to the process of automatically assigning aggregation weights to each meteorological element on a spatial grid through optimization learning, compressing high-dimensional grid data into low-dimensional regional features. Unlike simple averaging or other fixed aggregation methods, data-driven aggregation can learn the contribution weights of different grid points based on the actual response of the target variable.

[0036] Causal sparsity screening refers to the process of distinguishing causal relationships between variables rather than simply statistically analyzing correlations through methods such as conditional independence tests during feature selection, and eliminating redundant or irrelevant features through sparse regularization, while retaining key elements that have independent predictive contributions to the target variable.

[0037] Meteorological elements refer to various physical quantities that describe the state of the atmosphere, including but not limited to wind speed, wind direction, temperature, humidity, irradiance, air pressure, and precipitation.

[0038] The target variable data refers to the time series data of the object to be predicted. In the wind power prediction scenario, it is the wind power output; in the photovoltaic prediction scenario, it is the photovoltaic power generation; and in the load prediction scenario, it is the power system load value.

[0039] The gridded meteorological data index is the meteorological data output by the meteorological model, which is the observation or forecast data of meteorological elements on regular grid points covering the target area. Each grid point contains time series of multiple meteorological elements.

[0040] Calendar feature data refers to time-related auxiliary features, including hour, weekday, month, season, and holiday identifiers, which are used to control the impact of time changes on the target variable.

[0041] Correlation refers to the statistical measure of the degree of linear correlation between two variables, usually expressed by the Pearson correlation coefficient, which ranges from -1 to 1. The larger the absolute value, the stronger the linear correlation.

[0042] Partial correlation refers to the statistical measure of the net correlation between two variables when the influence of other variables is controlled. It can eliminate the interference of third-party variables and more accurately characterize the direct relationship between variables.

[0043] Mutual information refers to the information-theoretic measure of the amount of information shared between two variables. It can capture both linear and nonlinear relationships and has greater applicability than the correlation coefficient.

[0044] Lag causality discrimination refers to the process of determining, in time series data, whether a variable in the past has a directional causal influence on the target variable in the current time through methods such as conditional independence tests, and identifying the effective lag order.

[0045] Sparse filtering refers to compressing feature coefficients by introducing sparse regularization terms (such as L1 norm or group L1 norm), making unimportant feature coefficients zero, thereby achieving automatic reduction of feature dimensions.

[0046] Bootstrap is a nonparametric method that estimates the distribution of a statistic by resampling with replacement, and is used to estimate the uncertainty and confidence interval of parameters such as time delay.

[0047] The concept of robustness bar refers to evaluating the performance of a model or feature under the worst distribution, taking into account the uncertainty of the data distribution, in order to screen out robust elements that are not sensitive to distribution perturbations.

[0048] Seasonally stable screening refers to assessing the importance of meteorological elements in different seasons and constraining their cross-seasonal fluctuations in order to screen out elements that maintain a stable predictive contribution in all four seasons.

[0049] Counterfactual testing quantifies the necessary contribution of a feature to a predictive model by observing changes in predictive performance after removing it from the feature set. The more significant the performance decline, the more indispensable the feature is.

[0050] Example 1: This embodiment provides a meteorological element screening method based on data-driven grid aggregation and causal sparsity screening, such as... Figure 1 As shown, it includes the following steps: Step S1, data acquisition and preprocessing, such as Figure 2 As shown, it includes the following sub-steps: Multi-source time series data were acquired and preprocessed to construct a standardized dataset with a unified time index. The multi-source time series data included target variable data, gridded meteorological data, and calendar feature data.

[0051] Specifically, uniform time-granularity alignment and resampling are performed on the target variable data, gridded meteorological data, and calendar feature data to construct a sample set under a unified time index, thereby eliminating differences in sampling frequencies among multi-source data. The constructed standardized sample set is defined as follows: , In the formula: Represents a standardized sample set; Indicates the number of samples; Represents a unified time index set. This represents any point in time within the unified time index set. Indicates the first A unified time index corresponding to each sample; Indicates the first Each sample in the time index The target variable, where The term represents the set of real numbers, and its meaning remains the same throughout the following text. It should be noted, however, that in the general time index representation below, the target variable can be denoted as... , , and The only difference is the way the symbols are expressed; the meaning is the same. Indicates time index The gridded meteorological matrix, in which Indicates the number of grid points. Indicates the types and quantities of meteorological elements; Represents the characteristic vector of a calendar or control variable. Indicates the dimension of the control variable; The element is denoted as ,and For grid points In elements Above, Time The observed values, , .

[0052] Anomaly detection and missing data repair are performed on each variable sequence, followed by standardization to ensure data sequence continuity and comparability. For any sequence... Implement standardization to meet the following requirements: , In the formula: This represents the original sequence value to be standardized. Can be or , Indicates time The target variable at the location; This represents the standardized sequence values; Represents the mean of the sequence; Represents the standard deviation of the sequence, where Represents a unified time index set. Indicate its size; This represents a constant used to prevent the denominator from being zero.

[0053] Step S2, grid aggregation and candidate feature generation, such as Figure 2 As shown, it includes the following sub-steps: Data-driven aggregation is performed on the gridded meteorological data to obtain a regional meteorological element sequence, and a candidate meteorological element set is generated based on correlation, partial correlation and mutual information.

[0054] Specifically, for each meteorological element in the gridded meteorological data, non-negative aggregation weights are learned on the spatial grid, and a spatial total variation constraint is applied to the non-negative aggregation weights to achieve spatial smoothing of the weights of adjacent grid points. For any element... Regional aggregation satisfies: , In the formula: Representation elements In time Regional-level aggregated values; Representing grid points In elements Aggregate weights on; Representing grid points Elements In time Observed values; Indicates the number of grid points; This represents a unified time index set.

[0055] Aggregate weights satisfy: .

[0056] Define the set of adjacent edges of grid points as and the weight vector Apply total variation constraints: , In the formula, The space variation of the weights is represented by the total variation. Indicates a pair of adjacent grid points; This represents the set of adjacent grid point pairs.

[0057] The above aggregation weights are obtained through learning the following optimization objective: , In the formula: Indicates the target variable over time Observed values; Represented as a region-level feature vector and control variables The output is the regression function that is the input; , Representation elements In time Regional-level aggregated values; For time The control variable feature vector, Its dimensions; as elements The weight vector; express Norm; The space variation of the weights is represented by the total variation. , is the regularization coefficient.

[0058] After obtaining the regional meteorological element sequence, the correlation strength is calculated based on the correlation coefficient, partial correlation coefficient and mutual information between each element in the regional meteorological element sequence and the target variable data. If the calculated correlation strength exceeds the corresponding threshold, the element is included in the candidate meteorological element set.

[0059] Specifically, calculation elements Correlation coefficient: , In the formula: Representation elements Correlation coefficient with the target variable; Representation elements In time Regional aggregate values, Indicates the target variable over time Observed values; Representation elements The sample mean; This represents the sample mean of the target variable; Indicates the size of the time index set.

[0060] In control variables To calculate partial correlation under given conditions, first construct a vector. Then calculate the covariance matrix. and precision matrix Partial correlation is defined as: , In the formula: Indicates control The partial correlation coefficient after; For the control variable vector; Representing the precision matrix The There are 12 elements, with index "1" corresponding to each element. Component, "2" corresponds to Quantity.

[0061] Calculate mutual information: , In the formula: Representation elements Mutual information with the target variable; Represents a sequence The set of values ​​after discretization; Represents a sequence The set of values ​​after discretization; Represents discrete values The marginal probability; Represents discrete values The marginal probability; Represents discrete value pairs The joint probability; Represents a logarithmic function.

[0062] Step S3, time-delay causality determination and sparsity screening, such as Figure 3 As shown, it includes the following sub-steps: The candidate meteorological element set is subjected to time-delay causal discrimination and sparse screening to determine the set of key meteorological elements with directionality. The main time-delay relationships are identified by combining cross-correlation and dynamic time warping. The uncertainty is estimated by Bootstrap and the element-time-delay structured results are output.

[0063] Specifically, a conditional independent test is performed on each lag order of each element in the candidate meteorological element set, and the lags in which the null hypothesis is rejected are taken as the causal valid lags of that element, and elements with at least one causal valid lag are included in the key meteorological element set.

[0064] The candidate feature set is defined as: , In the formula, Represents the set of candidate feature indices; , This is the threshold parameter.

[0065] For each With each lag Independent verification of execution conditions: , In the formula: Indicate the null hypothesis; Indicates conditional independence; Indicates time The target variable; Representation elements Lag The regional level value below; This indicates the maximum lag order considered. The feature vector of the condition set contains at least... Control variables And the lagged terms of other candidate elements.

[0066] Conditional independence test outputs significance probability Let the significance level be . Then the elements The causal effective lag set is defined as: , And define the set of key elements as: , In the formula, Representation elements The causal effective lag set; This represents the set of key element indices used for causal discrimination. This represents the empty set.

[0067] Step S3, identifying the main time delay relationships, specifically involves: estimating the linear principal time delay using a cross-correlation function, estimating the nonlinear time delay using dynamic time warping, and estimating the confidence interval of the time delay using the Bootstrap resampling method. Define the cross-correlation function: , And in The main time delay is obtained above: , In the formula: Representation elements Lag The cross-correlation value between the target and the target; Indicates the mutual correlation principal delay; , They are respectively , The sample mean; Indicates the number of samples; This represents the independent variable that makes the function reach its maximum value.

[0068] right Construct sequence , The alignment path is obtained through dynamic time warping: , And define the offset Then the nonlinear representative time delay is defined as: , In the formula: Represents the set of alignment paths between the feature sequence and the target sequence; Indicates the first in the path One alignment index pair; Indicates the path length; Indicates the index offset; Represents the median operator; This represents the nonlinear time delay of the DTW estimate.

[0069] Bootstrap resampling is performed on the time delay estimate. Next, samples were obtained. Its confidence interval is defined as: , In the formula: Indicates the number of resampling attempts; Indicates the first The time delay estimate obtained from the second resampling (can be taken as...) or (resampled estimation); Indicates the confidence level parameter; express Quantile operator; This represents the confidence interval for time delay.

[0070] Each element The lagged terms are grouped together to construct a set of eigenvectors: , And solve the set of sparse optimizations: , Output the final set of elements: , In the formula: Representation elements The group coefficient vector; Represents the group's sparse regularity coefficients; Represents the L2 norm; This represents the final set of element indexes after sparse filtering.

[0071] Step S4: Analyze the Blue Bar and seasonal stability screening, and perform a counterfactual test, such as... Figure 4 As shown, it includes the following sub-steps: The key meteorological element set is subjected to split-bar and seasonal stability screening, and the contribution of the elements is quantified by counterfactual testing to generate a callable meteorological element library containing an element list, grid aggregation mapping, time lag parameters and stability indicators.

[0072] The Wasserstein robustness screening method constructs an uncertain set based on Wasserstein distance and evaluates the worst-case performance of the key meteorological element set under distribution perturbation. The Wasserstein robustness objective is: , In the formula: Represents a function used for evaluation or prediction; Represent the function space; Represents an empirical distribution based on the sample; Indicated by Centered on, with radius The Wasserstein sphere is an uncertain set. Indicates the robust radius; Indicates distribution The following expectations; Represents the loss function; Represents the final set of elements The time-delay eigenvectors formed; Indicates time The target variable.

[0073] The seasonal stability screening calculates the importance of each element in the key meteorological element set according to the season and constrains its cross-seasonal fluctuations. (By seasonal set) Calculate the importance of elements And apply stability constraints: , And calculate the retention rate across quarters: , In the formula: Representation elements In the season Importance indicators; Represents the variance operator; This indicates the upper bound of importance fluctuation; Representation elements Cross-quarter retention rate; Indicates the size of the seasonal set; Indicates an indicator function; Indicates the importance threshold; Indicates a seasonal gathering. For seasonal indexing.

[0074] The counterfactual test quantifies the contribution of a factor by measuring the decrease in predictive performance after removing a single factor from the set of key meteorological factors: , In the formula: Representation elements The degree of contribution; The performance index function is an error index or a goodness-of-fit index. Represents the final set of elements; Indicates from set Remove elements The set after.

[0075] The final output meteorological element database should include at least: the final element set. With candidate set ; Grid aggregation weight Adjacency set And spatial smoothing parameters; causal effective lag set Cross-correlation principal time delay DTW time delay Bootstrap confidence interval ; Sparse screening parameters Relevant thresholds Partial correlation threshold Mutual information threshold Significance level Robust radius Seasonal stability parameters Retention threshold Cross-quarter retention rate and contribution .

[0076] This embodiment completes the end-to-end construction from multi-source time series data to a callable meteorological element library. The selected meteorological elements have the characteristics of clear causal direction, time lag matching, cross-seasonal stability and quantifiable contribution, and can be directly used as feature inputs for wind power, photovoltaic power output and load forecasting models.

[0077] Example 2: like Figure 5As shown, this embodiment provides a meteorological element screening system based on data-driven grid aggregation and causal sparse screening, used to implement the meteorological element screening method described in Embodiment 1, including: The data acquisition and preprocessing module is used to execute step S1, acquire multi-source time series data of the target area and preprocess it, and construct a standardized dataset under a unified time index. The multi-source time series data includes target variable data, gridded meteorological data and calendar feature data. The grid aggregation and candidate generation module is used to perform data-driven aggregation on the grid meteorological data to obtain a regional meteorological element sequence, and generate a candidate meteorological element set based on correlation, partial correlation and mutual information. The causal discrimination and sparse screening module is used to perform time-delay causal discrimination and sparse screening on the candidate meteorological element set, determine the set of key meteorological elements with directionality, identify the main time-delay relationships by combining cross-correlation and dynamic time warping, estimate the uncertainty through Bootstrap, and output the element-time-delay structured results. The robust screening and verification module is used to perform robust and seasonally stable screening on the set of key meteorological elements, and to quantify the contribution of elements through counterfactual testing.

[0078] The element library generation and storage module is used to generate a stable set of meteorological elements based on the screening results and establish a meteorological element library. The meteorological element library includes an element list, grid aggregation mapping, time delay parameters and stability indicators for subsequent prediction models to call.

[0079] Example 3: like Figure 6 As shown, this embodiment provides an electronic device, which may include: at least one processor, at least one network interface, a user interface, a memory, and at least one communication bus.

[0080] The communication bus can be used to enable communication between the various components mentioned above.

[0081] The user interface may include buttons, and optional user interfaces may also include standard wired interfaces and wireless interfaces.

[0082] The network interface may include, but is not limited to, Bluetooth modules, NFC modules, Wi-Fi modules, etc.

[0083] The processor may include one or more processing cores. It connects various parts of the electronic device via various interfaces and lines, executing instructions, programs, code sets, or instruction sets stored in memory, and accessing data stored in memory to perform various functions and process data. Optionally, the processor can be implemented using at least one hardware form of DSP, FPGA, or PLA. The processor may integrate one or more of the following: CPU, GPU, and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor.

[0084] The memory may include RAM or ROM. Optionally, the memory may include a non-transitory computer-readable medium. The memory can be used to store instructions, programs, code, code sets, or instruction sets. The memory may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor. The memory, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a filtering application. The processor can be used to call the filtering application stored in the memory and execute the steps of the meteorological element filtering method mentioned in the foregoing embodiments.

[0085] Example 4: This embodiment provides a computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform the above-described instructions. Figure 1 One or more steps in the illustrated embodiment. If the constituent modules of the above-described electronic device are implemented as software functional units and sold or used as independent products, they can be stored in the computer-readable storage medium.

[0086] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0087] Those skilled in the art will understand that all or part of the processes in the method of Embodiment 1 described above can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks. Unless otherwise specified, the technical features of this embodiment and the implementation scheme can be combined arbitrarily.

[0088] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0089] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0090] The above description is merely an exemplary embodiment of the present invention and should not be construed as limiting the scope of the invention. Any equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of embodiments of the invention upon considering the specification and practicing the disclosure herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of the invention are defined by the claims.

Claims

1. A meteorological element screening method based on data-driven grid aggregation and causal sparse screening, characterized in that, Including the following steps: S1. Acquire multi-source time series data of the target area and preprocess it to construct a standardized dataset under a unified time index. The multi-source time series data includes target variable data, gridded meteorological data and calendar feature data. S2. Perform data-driven aggregation on the gridded meteorological data to obtain a regional meteorological element sequence, and generate a candidate meteorological element set based on correlation, partial correlation and mutual information. S3. Perform time-delay causal discrimination and sparse screening on the candidate meteorological element set to determine the set of key meteorological elements with directionality, and identify the main time-delay relationships by combining cross-correlation and dynamic time warping. Estimate the uncertainty through Bootstrap and output the element-time-delay structured results. S4. Perform split-bar and seasonal stability screening on the set of key meteorological elements, and quantify the contribution of elements through counterfactual testing to generate a callable meteorological element library containing an element list, grid aggregation mapping, time lag parameters and stability indicators.

2. The meteorological element screening method according to claim 1, characterized in that, Step S2, which involves data-driven aggregation of gridded meteorological data, specifically includes: For each meteorological element in the gridded meteorological data, non-negative aggregate weights are learned on the spatial grid, and spatial total variation constraints are applied to the non-negative aggregate weights to achieve spatial smoothing of the weights of adjacent grid points.

3. The meteorological element screening method according to claim 1, characterized in that, The specific steps in step S2 for generating the candidate meteorological element set are as follows: The correlation strength is calculated based on the correlation coefficient, partial correlation coefficient and mutual information between each element in the regional meteorological element sequence and each variable in the target variable data. If the calculated correlation strength exceeds the corresponding threshold, the element will be included in the candidate meteorological element set.

4. The meteorological element screening method according to claim 1, characterized in that, The time-delay causality determination in step S3 is specifically as follows: Perform a conditional independent test on each lag order of each element in the candidate meteorological element set, take the lag where the null hypothesis is rejected as the causal effective lag of the element, and include elements with at least one causal effective lag into the key meteorological element set.

5. The meteorological element screening method according to claim 1, characterized in that, The sparse screening in step S3 specifically involves: Group all the lagged terms of each meteorological element in the candidate meteorological element set, and perform joint compression on the coefficients of each group. Elements with a group coefficient of zero are eliminated, while elements with a group coefficient of non-zero are retained.

6. The meteorological element screening method according to claim 1, characterized in that, The identification of the main time delay relationships in step S3 specifically involves: The linear principal time delay is estimated using the cross-correlation function, the nonlinear time delay is estimated using dynamic time warping, and the confidence interval of the time delay is estimated using the Bootstrap resampling method.

7. The meteorological element screening method according to claim 1, characterized in that, In step S4: The split-bar screening is based on the Wasserstein distance to construct an uncertain set and evaluate the worst-case performance of the key meteorological element set under distribution perturbation. The seasonal stability screening calculates the importance of each element in the set of key meteorological elements according to the season and constrains its cross-seasonal fluctuations. The counterfactual test quantifies the contribution of a single element by removing it from the set of key meteorological elements and then measuring the decrease in predictive performance.

8. A meteorological element screening system based on data-driven grid aggregation and causal sparse screening, characterized in that, Used to implement the meteorological element screening method as described in any one of claims 1 to 7.

9. A computer device, the computer device comprising a memory, a processor, and a computer program, characterized in that, When the computer program is executed by the processor, it implements the meteorological element screening method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the meteorological element screening method as described in any one of claims 1 to 7.

Images