A method and system for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output

Abstract

This invention belongs to the field of meteorological impact analysis technology for photovoltaic power generation, and relates to a method and system for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output. The system employs a micro-scale parameter correction module to output a corrected photovoltaic module power output calculation model at the micro-scale; a meso-scale key factor extraction module to extract meteorological factor combinations and dependency structures based on station-level meteorological and power output data at the meso-scale; a macro-scale spatiotemporal modeling and quantification module to output the influence weights of each meteorological factor and response curves of typical weather processes based on a spatiotemporal relationship model at the macro-scale; and a comprehensive analysis module to output conclusions on the influencing factors of multidimensional meteorological conditions on photovoltaic power output at the micro, meso, and macro scales. This invention can effectively improve the accuracy of photovoltaic power output impact mechanism analysis.

Description

Technical Field

[0001] This invention relates to the field of meteorological impact analysis technology for photovoltaic power generation, specifically to a method and system for analyzing the influence of multidimensional meteorological conditions on photovoltaic power output. Background Technology

[0002] Photovoltaic power generation is complexly influenced by various meteorological factors such as solar radiation, temperature, humidity, and wind speed. Accurately understanding the intrinsic relationship between meteorological conditions and photovoltaic output is a crucial prerequisite for improving the accuracy of photovoltaic power prediction and optimizing grid dispatch capabilities. Current technologies collect historical photovoltaic power generation and meteorological data, remove nighttime data and normalize it, and classify weather types according to the clear sky index. A photovoltaic power conversion model considering physical errors such as temperature and pollution is constructed. Simultaneously, a multilayer perceptron with L2 regularization and Gaussian initialization is used to predict irradiance and ambient temperature under different weather conditions. Finally, the predicted values ​​are substituted into the model to output photovoltaic power.

[0003] The following technical issues exist: There is a lack of systematic analysis of the impact mechanism of multidimensional meteorological factors on distributed power output at the micro, meso, and macro scales.

[0004] In view of this, it is very necessary to provide a method and system for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output, so as to solve the above-mentioned defects in the prior art. Summary of the Invention

[0005] The purpose of this invention is to address the technical problem in the prior art of lacking a systematic analysis of the influence mechanism of multidimensional meteorological factors on distributed photovoltaic power output at the micro, meso, and macro scales, and to provide a method and system for designing a method to analyze the influence factors of multidimensional meteorological conditions on photovoltaic power output, so as to solve the technical problems existing in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output, comprising the following steps: Step S1: At the microscale, multidimensional meteorological variables are introduced based on the physical model of the photovoltaic module to perform dynamic parameter correction, thereby obtaining the corrected photovoltaic module output calculation model. Step S2: At the mesoscale, based on station-level meteorological and power output data, autocorrelation analysis, cross-correlation analysis, maximum information coefficient, and characteristic synergy methods are used to obtain the combination of meteorological factors and their dependency structure. Step S3: At the macro scale, construct a spatiotemporal relationship model based on regional meteorological field and multi-station power data, and output the influence weights of each meteorological factor and the response curves of typical weather processes; Step S4: Based on the photovoltaic module output calculation model, the combination and dependency structure of meteorological factors, the influence weight of each meteorological factor and the response curve of typical weather processes, a cross-scale fusion analysis is performed to output the influencing factors affecting photovoltaic output.

[0007] Secondly, the present invention also provides a system for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output, comprising: The microscale parameter correction module is used to dynamically correct parameters at the microscale by introducing multidimensional meteorological variables based on the physical model of the photovoltaic module, so as to obtain the corrected photovoltaic module output calculation model. The mesoscale key factor extraction module is used to extract meteorological factor combinations and dependency structures at the mesoscale, based on station-level meteorological and power output data, by using autocorrelation analysis and cross-correlation analysis to evaluate the correlation of time series, and by combining the maximum information coefficient and feature synergy methods. The macro-scale spatiotemporal modeling and quantification module is used to construct spatiotemporal relationship models based on regional meteorological fields and multi-station power data at the macro scale, and output the influence weights of various meteorological factors and response curves of typical weather processes. The comprehensive analysis module performs cross-scale fusion analysis based on the photovoltaic module output calculation model, meteorological factor combination and dependency structure, influence weight of each meteorological factor and response curve of typical weather processes, and outputs the multi-dimensional meteorological conditions affecting photovoltaic output at the micro, meso and macro scales.

[0008] The modules work together to achieve everything from micro-component-level parameter correction and meso-level site-level key factor mining to macro-level regional-level impact quantification. Through cross-scale fusion analysis, the system systematically reveals the influencing factors of multi-dimensional meteorological conditions on photovoltaic power output at different spatial scales.

[0009] The beneficial effects of this invention are as follows: This invention addresses the lack of systematic analysis in existing technologies regarding the impact of multidimensional meteorological factors on distributed photovoltaic (PV) output at micro, meso, and macro scales. At the micro scale, it introduces multidimensional meteorological variables based on the PV module's physical model for dynamic parameter correction, forming a modified PV module output calculation model. This model allows the module output to dynamically adjust with changes in meteorological conditions such as irradiance, ambient temperature, and wind speed, improving the physical consistency and accuracy of the output calculation results and laying a reliable foundation for subsequent mechanism analysis.

[0010] At the meso-scale, key meteorological factor combinations and their interdependence structures are extracted based on station-level meteorological data and power output data, enabling a quantitative expression of the synergistic relationship between multiple meteorological variables. This avoids the bias caused by isolated analysis of single factors, enhances the explanatory power for the causes of photovoltaic power output fluctuations at the station level, and improves the completeness and accuracy of influencing factor identification.

[0011] At a macro scale, a spatiotemporal relationship model is constructed based on regional meteorological fields and multi-station power data. The model outputs the influence weights of various meteorological factors and response curves of typical weather processes, thereby achieving a quantitative characterization of the degree of influence of different meteorological factors and revealing the spatiotemporal evolution law of photovoltaic power output under typical weather processes. This enhances the overall ability to grasp the trend of regional distributed photovoltaic power output changes.

[0012] Based on this, the modified photovoltaic module output calculation model, key meteorological factor combinations and their interdependence structure, the influence weight of each meteorological factor, and the response curves of typical weather processes are subjected to cross-scale fusion analysis to achieve the connection between micro, meso and macro scales, and output a unified expression of the multi-dimensional meteorological conditions affecting photovoltaic output. This effectively solves the technical problem of lacking cross-scale systematic analysis in the existing technology and improves the systematicness, accuracy and hierarchy of the photovoltaic output influence mechanism analysis.

[0013] Therefore, it is evident that the present invention has outstanding substantive features and significant progress compared with the prior art, and the beneficial effects of its implementation are also obvious. Attached Figure Description

[0014] 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 embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0015] Figure 1 This is a flowchart of a method for analyzing the influence of multidimensional meteorological conditions on photovoltaic power output; Figure 2 This is a schematic diagram of the principle of a system for analyzing the influence of multidimensional meteorological conditions on photovoltaic power output. Detailed Implementation

[0016] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The following embodiments are explanations of the present invention, but the present invention is not limited to the following implementation methods.

[0017] Example 1: like Figure 1 As shown in the figure, this embodiment provides a method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output, including the following steps: Step S1: At the microscale, multidimensional meteorological variables are introduced based on the physical model of the photovoltaic module to perform dynamic parameter correction, thereby obtaining the corrected photovoltaic module output calculation model. Step S2: At the mesoscale, based on station-level meteorological and power output data, autocorrelation analysis, cross-correlation analysis, maximum information coefficient, and characteristic synergy methods are used to obtain the combination of meteorological factors and their dependency structure. Step S3: At the macro scale, construct a spatiotemporal relationship model based on regional meteorological field and multi-station power data, and output the influence weights of each meteorological factor and the response curves of typical weather processes; Step S4: Based on the photovoltaic module output calculation model, the combination and dependency structure of meteorological factors, the influence weight of each meteorological factor and the response curve of typical weather processes, a cross-scale fusion analysis is performed to output the influencing factors affecting photovoltaic output.

[0018] In step S1: At the microscale, photovoltaic (PV) modules are modeled using a combination of physical modeling and simulation verification, focusing on the electrical characteristics of individual PV modules. Simulation verification is conducted using the PVsyst software simulation platform. After inputting parameters under standard test conditions, the basic output of the PV module, including the I-V curve and maximum power point, is simulated to verify the feasibility and consistency of the physical model under engineering conditions.

[0019] In terms of physical modeling, considering temperature T and irradiance S, an engineering approximation method is used to transform the implicit relationships of the equivalent circuit of a silicon solar cell into explicit I-V expressions. The simplified current-voltage equation is obtained as follows:

[0020] The maximum power point parameter Open circuit state parameters .

[0021] Through engineering approximation, what was originally difficult to measure directly... and reverse saturation current The internal parameters are converted into a form that can be expressed by measurable parameters, namely the short-circuit current. Open circuit voltage Maximum power point voltage and maximum power point current This improves the availability and computability of physical models in real-world engineering scenarios. Its electrical structure corresponds to the equivalent circuit of a single diode, as shown in the figure below.

[0022] To characterize the impact of more complex meteorological disturbances, multidimensional meteorological variables are introduced to the core parameters while maintaining the original form of expression. and Dynamic corrections are made. Multidimensional meteorological variables include diffuse radiation intensity. Direct radiation intensity Ambient temperature Surface temperature of photovoltaic modules The corrected physical model I–V for calculating the output of the photovoltaic module, taking into account wind speed V and humidity H, is as follows:

[0023] The correction relation satisfies: ; In the formula, and The original fitting parameters are those without considering multidimensional meteorological variables; ξ and η are the dynamic correction coefficients obtained by training the multilayer perceptron network based on multidimensional meteorological variables; the fitting parameters for the actual maximum power point are... Actual open-circuit state fitting parameters These are the core parameters after correction by multidimensional meteorological variables.

[0024] The dynamic correction coefficients ξ and η are solved by a multilayer perceptron network. The input is a multidimensional meteorological variable. The dynamic correction coefficients corresponding to C1 and C2 are solved through the multilayer perceptron network. The solution model is as follows:

[0025] The two output branches ξ and η correspond to C1 and C2 respectively, achieving independent parameter decoupling and correction. The values ​​of the correction coefficients ξ and η are limited to [0, 0.3], controlled within the weak correction range, to avoid excessive deviation from the basic physical expression, and to ensure that the photovoltaic module output physical model maintains stable engineering applicability and physical interpretability under complex weather conditions.

[0026] Through step S1, a physical model combining a physical expression based on measurable parameters and multidimensional meteorological dynamic correction is formed at the microscale of photovoltaic module output. This enables the photovoltaic module output physical model to have a clear physical basis and to respond to coupled disturbances from multiple factors such as scattered radiation, wind speed, and humidity.

[0027] In step S2: Within the meso-scale range, a meteorological-power output correlation analysis is conducted using site-level observation data within the photovoltaic module area of ​​the research object. Historical meteorological data and corresponding photovoltaic power generation data at the site level are obtained within the research area. Historical meteorological data includes site-level irradiance, temperature, humidity, wind speed, and air pressure; the photovoltaic power generation data represents the total photovoltaic power generation at the corresponding site.

[0028] In the data processing stage, historical meteorological data and corresponding photovoltaic power generation data are preprocessed to generate meteorological time series and power time series: For outlier data, statistical thresholding and the isolated forest algorithm were used for anomaly detection to identify and remove erroneous samples. For missing data, cubic spline interpolation, Hermitian interpolation, and machine learning-based interpolation methods were used for completion. After anomaly removal and missing data completion, the preprocessed historical meteorological data and photovoltaic power generation data were regularized into continuous, equally time-interval meteorological time series and power time series, providing a stable data foundation for subsequent related analyses.

[0029] In terms of time structure analysis, the autocorrelation function (ACF) is used to evaluate the autocorrelation characteristics of meteorological and power time series, respectively, and to identify the periodicity and lag characteristics of the meteorological and power time series. The cross-correlation relationship between the meteorological and power time series is analyzed to reveal the time delay characteristics and coupling strength of the photovoltaic power response to changes in meteorological variables. The autocorrelation characteristics of meteorological time series and photovoltaic power time series were evaluated separately. Let the meteorological time series be... The power time series is The sample length is N, and the means are respectively and The autocorrelation coefficient of a meteorological time series under a lag order of k is defined as follows: ; The autocorrelation coefficient of the power series under the condition of lag order k is ; An ACF curve is constructed by calculating the autocorrelation coefficient series for different lag orders k=1,2,…,K. When the autocorrelation coefficients corresponding to certain lag orders are significantly higher than the confidence interval ± When autocorrelation analysis is performed on meteorological and power time series, their respective periodic characteristics and lag properties can be identified. For example, a peak at k=24 at hourly resolution indicates a daily cycle; a peak at k=96 at 15-minute resolution indicates an intra-day cycle.

[0030] In the time-coupled analysis of meteorological and power time series, the cross-correlation function is used to characterize the lagged response relationship between variables. The cross-correlation function is defined as: ; When k > 0, it indicates that the meteorological variable leads the power series by k time steps; when k < 0, it indicates that the power series leads the meteorological variable. This is achieved by solving...

[0031] The optimal lag time delay of the impact of meteorological variables on power can be obtained. The corresponding ρXP ( The coupling strength is represented by the coefficient (C), which can quantitatively reveal the time delay characteristics and coupling strength of the photovoltaic power response to different meteorological variables, the magnitude of the delay effect, and the difference in lag between variables.

[0032] Based on autocorrelation and cross-correlation, a correlation analysis method combining maximum information coefficient (MIC) and characteristic synergy (VI) is constructed to quantitatively characterize the relationship between multidimensional meteorological variables and their corresponding photovoltaic power output. MIC is used to identify the strength of nonlinear correlations, while VI characterizes the degree of synergistic contribution of multiple variables. Through joint analysis, the combination of meteorological factors that significantly influence distributed photovoltaic power at the mesoscale and their dependency structure are obtained. This reveals the intrinsic correlation mechanism between changes in meteorological conditions and the power output of distributed photovoltaic modules at the mesoscale, providing a basis for meteorological factor selection and structural constraints for subsequent multi-scale prediction modeling.

[0033] Let the multidimensional meteorological variable be X, and the photovoltaic power be P. Mutual information is defined as:

[0034] The maximum information coefficient is the maximum normalized mutual information value found under different grid partitioning conditions, and its expression is: , Where x and y represent the number of grid divisions. B(N) represents the maximum mutual information under the optimal partition, and B(N) is the upper bound function of the sample size correlation. MIC ranges from [0,1] and is used to measure the strength of nonlinear association.

[0035] In the case of multiple variables, to characterize the synergistic contribution relationship among variables, a characteristic synergy index is introduced. Let the set of multidimensional meteorological variables be... Construct the synergy expression:

[0036] in, For univariate mutual information, This represents the mutual information of the joint variables regarding power. When VI > 0, it indicates a synergistic enhancement effect; when VI ≈ 0, it indicates independent contributions; and when VI < 0, it indicates redundant relationships between variables.

[0037] Based on this, a joint evaluation index is constructed. The degree of univariate correlation is defined as:

[0038] The bivariate combined evaluation index is defined as follows:

[0039] Where λ is the synergistic weighting coefficient. Based on the score ranking results, the combination of meteorological factors that have a significant impact on mesoscale distributed photovoltaic power and their dependency structure are selected.

[0040] By employing high-precision anomaly removal, hysteresis feature extraction, and nonlinear correlation analysis, the sensitivity of photovoltaic power to key meteorological factors at the mesoscale can be clearly identified, providing core input factors and structural constraints for multi-scale modeling.

[0041] In step S3: At the macroscopic level, a joint analysis of regional meteorological fields and multi-station power sequences is conducted around distributed photovoltaic (PV) sites within the target area. This involves acquiring reanalysis meteorological grid field data, distributed PV site power data, and site metadata within the target area.

[0042] Site metadata serves as the foundational support for spatial matching of meteorological data, normalization of power data, and analysis of regional photovoltaic output characteristics at a macro scale. It includes the latitude and longitude of photovoltaic sites, installed capacity, and module type. Its functions are threefold: First, it provides accurate geographic coordinates for interpolating reanalysis meteorological grid data to the corresponding photovoltaic site locations, ensuring spatial consistency between the meteorological information in the reanalysis grid and the actual site locations, thus guaranteeing the correlation of meteorological data. Second, based on the installed capacity and module type of photovoltaic sites, it normalizes the power data of distributed photovoltaic sites, such as converting it to output per unit installed capacity, eliminating output differences caused by different hardware configurations, and enabling comparability and fusion of power data from multiple sites in the region. Third, it provides site attribute features for the spatiotemporal relationship model of hybrid neural networks and weather scene gating mechanisms, characterizing the differences in response to meteorological factors among different types of sites.

[0043] During the meteorological data processing stage, bilinear interpolation was used on the reanalysis meteorological grid field data based on the latitude and longitude in the station metadata to obtain the station-level meteorological time series for the corresponding stations. This time series was then resampled to the same time resolution as the distributed photovoltaic (PV) station power data. Outlier removal and missing value imputation were performed on the distributed PV station power data to ensure the continuity and statistical stability of the time series.

[0044] The solar altitude angle and solar azimuth angle are calculated using a solar geometric model, and clear-sky radiation is calculated using the Ineichen clear skymodel. The clear-sky index is then derived from the clear-sky radiation and measured radiation to characterize the degree of cloud cover and the intensity of weather disturbances. The clear-sky index is calculated as follows: Calculate the solar altitude angle using a solar geometric model. and solar azimuth The solar altitude angle and azimuth angle are used to describe the spatial position of the sun relative to a geographical location and time. The formulas for calculating the solar altitude angle and azimuth angle are:

[0045] in, Here, δ represents the latitude of the observation location, h represents the solar declination angle, and δ represents the hour angle. Using these geometric parameters, solar radiation can be mapped onto the horizontal plane of the ground and the tilting surface of the photovoltaic modules.

[0046] Clear-sky radiation is calculated based on solar altitude angle and azimuth angle, combined with the Ineichen clear-sky model. : ; in The solar constant, This refers to atmospheric transmittance (direct radiation portion). denoted as the transmittance of scattered radiation.

[0047] Clear-sky radiation Compared with the measured radiation amount Combined, calculate the clear sky index. Quantifying the intensity of cloud cover and weather disturbances: Clear Sky Index The lower the value, the stronger the cloud cover or the more obvious the weather disturbance.

[0048] Based on the reanalysis of meteorological grid field data, lag features of 1 to 4 time steps and rolling statistical window features of 4 to 12 time steps are constructed to form a multi-scale meteorological feature matrix X, which is used to describe the temporal evolution characteristics of the regional meteorological field.

[0049] In spatial correlation analysis, by combining power data from distributed photovoltaic sites with multi-scale meteorological feature matrices, the delay correlation coefficients and correlation delays between sites are calculated. The spatial correlation radius is then determined based on the range curve of the correlation coefficient as a function of spatial distance. Using this correlation radius as a constraint, a multi-site coupling relationship structure constraint is constructed, providing a topological foundation for subsequent spatiotemporal joint modeling. By combining the power matrix P of distributed photovoltaic sites with the multi-scale meteorological feature matrix X, the inter-site delay correlation coefficient is calculated. and related delays : ; according to Determine the spatial correlation radius from the range curve that varies with spatial distance. As a constraint condition for constructing a multi-station coupling relationship structure, it provides a topological foundation for subsequent spatiotemporal joint modeling.

[0050] Under the constraint of multi-station coupling relationship structure, the power data of distributed photovoltaic sites and the corresponding multi-scale meteorological feature matrices are input into a spatiotemporal relationship model containing convolutional networks, graph convolutional networks, and long short-term memory networks to achieve the fusion of spatial structure modeling and temporal dynamic modeling. A weather scene gating mechanism based on cloud cover, clear sky index, wind speed, and convective available potential energy is adopted to adaptively adjust the feature channels under different weather types, quantifying the influence intensity of each meteorological factor on regional photovoltaic output under different weather processes. The specific operation is as follows: The power data P from distributed photovoltaic sites and the corresponding meteorological feature matrix X are input into the spatiotemporal relationship model to achieve the fusion of spatial structure modeling and temporal dynamic modeling. The spatiotemporal relationship model includes convolutional networks (CNN), graph convolutional networks (GCN), and long short-term memory networks (LSTM). Where A is the site adjacency matrix, which is composed of spatially related radii. Confirmed, ⊕ indicates feature fusion operation. This is the length of the scroll window.

[0051] A weather scene gating mechanism based on cloud cover, clear sky index, wind speed, and convective available potential energy is adopted to adaptively adjust the characteristic channels under different weather types, thereby realizing the dynamic weighting of meteorological factors on photovoltaic output. Where ⊙ denotes element-wise multiplication, σ is the Sigmoid function, and W is the trainable weight matrix.

[0052] The contribution of each meteorological variable to regional photovoltaic output was obtained through spatiotemporal relationship model network training and feature response analysis. The power response characteristics under typical weather processes are extracted. By combining spatial averaging, principal component analysis (PCA), empirical orthogonal function (EOF) decomposition and other methods, the total output of regional distributed photovoltaic power is structurally characterized and its features are compressed. The output results include the ranking of the influence weights of various meteorological factors, the response curve of photovoltaic power to typical weather processes, and the analysis conclusions of the regional-scale meteorological-power coupling mechanism.

[0053] By constructing a spatiotemporal relationship model and a weather scenario gating mechanism, the dynamic contribution of various meteorological factors to regional photovoltaic power output at a macro scale can be quantified, and typical weather process response characteristics can be generated, providing a quantitative basis for regional photovoltaic power prediction and optimization.

[0054] In step S4: Based on the photovoltaic module output calculation model, meteorological factor combination and dependency structure, influence weight of each meteorological factor and response curve of typical weather processes, cross-scale fusion analysis is carried out: the results of micro, meso and macro scales are integrated; weight mapping and collaborative correction are performed to achieve a unified measurement of the influence intensity of meteorological variables at different scales, cross-scale variable coupling relationship is constructed, the cross-scale transmission path of meteorological factors is analyzed in combination with typical weather processes, consistency test and sensitivity analysis are carried out, and the key meteorological factors affecting photovoltaic output are output.

[0055] The process of integrating results at the micro, meso, and macro scales is as follows: The I-V relationship of the photovoltaic module output calculation model, which is corrected by multidimensional meteorological variables at the micro scale, is used as the physical basis. The combination of meteorological factors selected at the meso scale is used as the input variable set. The influence weights of each meteorological factor identified at the macro scale are normalized to uniformly calibrate the intensity of the effects of different meteorological variables at different spatial scales.

[0056] The weight mapping and collaborative correction operation is performed as follows: The influence weights of various meteorological factors obtained at the macro scale are transferred to the meso and micro scales. Combined with the dependence structure of meteorological factors at the meso scale, the influence strength of variables is synergistically corrected to achieve a unified measurement of the degree of influence of meteorological variables across different scales.

[0057] The process of constructing cross-scale variable coupling relationships, analyzing the cross-scale transmission paths of meteorological factors in conjunction with typical weather processes, and performing consistency checks and sensitivity analyses are as follows: By combining the dependency structure of meteorological factors obtained at the mesoscale, a cross-scale variable coupling relationship is constructed. This cross-scale variable coupling relationship is then introduced into the parameter correction process of the microscale output calculation model to establish a cross-scale response relationship between meteorological factors and photovoltaic module output. Furthermore, by combining the response curves of typical weather processes at the macroscale, the variable interaction relationship under different weather scenarios is constrained.

[0058] The response curves of typical weather processes are decomposed and analyzed, and the changes in weather processes at the macro scale are mapped step by step to the coordinated changes of meteorological factors at the meso scale, and then transmitted to the changes in component parameters at the micro scale. The transmission paths of meteorological factors at different spatial scales and their enhancing or weakening effects are identified.

[0059] Consistency and sensitivity analyses were conducted on the influence direction, magnitude, and weight ranking of the same meteorological factor at different scales. This included a consistency comparison between the correction coefficient and the weight ranking of each meteorological factor, an analysis of the correspondence between the correlation radius between stations and the synergistic strength of meteorological factors, and a stability analysis of the influence direction and magnitude of the same meteorological variable at different scales. The dominant and secondary intervals of meteorological variables at different scales were extracted to form a unified expression of the impact of multidimensional meteorological conditions on photovoltaic output at the module, station, and regional scales, and to output the meteorological factors affecting photovoltaic output.

[0060] Example 2: like Figure 2 As shown in this embodiment, a system for analyzing the influence of multidimensional meteorological conditions on photovoltaic power output includes: The microscale parameter correction module 1 collects multidimensional meteorological variables at the location of the photovoltaic module, performs standardization processing, and generates a multidimensional meteorological variable vector. It employs a multilayer perceptron structure with dual output branches, where each branch corresponds to dynamically correcting core parameters in the photovoltaic module's physical model. The correction coefficients obtained through dual-branch calculations independently adjust the core parameters of the photovoltaic module's physical model, achieving decoupled parameter correction and yielding the corrected photovoltaic module output calculation model. This corrected photovoltaic module output calculation model can dynamically respond to multidimensional meteorological disturbances in the photovoltaic module's environment at the microscale, ensuring its stability and interpretability even under complex meteorological conditions.

[0061] The mesoscale key factor extraction module 2 preprocesses historical meteorological and photovoltaic power data at the station level, including outlier removal, missing value imputation, and time series normalization, ensuring that the input data is continuous, equally spaced, and numerically stable. Anomalies are identified using a combination of statistical thresholding and the isolated forest algorithm. Missing value imputation integrates cubic spline interpolation, Hermitian interpolation, and machine learning interpolation methods to generate meteorological and power time series. The time series characteristics of the meteorological and power time series are calculated using the autocorrelation function (ACF) and cross-correlation function (CCF), extracting the periodicity, lag characteristics, and their delayed impact on power response, providing a time-series basis for subsequent factor selection. A correlation analysis method combining maximum information coefficient and feature synergy is used to quantify the relationship between multidimensional meteorological variables and corresponding photovoltaic power output, outputting the combination of meteorological factors affecting mesoscale distributed photovoltaic power and their dependency structure, providing meteorological factors and constraints for multi-scale modeling.

[0062] The macro-scale spatiotemporal modeling and quantification module 3 performs spatial matching and normalization processing on reanalysis meteorological grid field data and distributed photovoltaic site power data. It maps the grid field data to site locations using bilinear interpolation and standardizes the power data according to installed capacity and module type. A solar geometry model is used to calculate the solar altitude and azimuth angles, and a clear-sky index is generated by combining it with a clear-sky radiation model. Based on the reanalysis meteorological grid field data, multi-time-step lag features and rolling statistical features are constructed to form a multi-scale meteorological feature matrix, achieving a time-series representation of the regional meteorological field. By calculating the inter-site delay correlation coefficient and related time delay, and combining the range curve of the correlation coefficient changing with spatial distance, the spatial correlation radius is automatically determined, constructing a multi-station coupling relationship structure constraint. Under this constraint, the distributed photovoltaic site power data and the corresponding multi-scale meteorological feature matrix are input into the spatiotemporal relationship model, achieving a joint spatial structure modeling and temporal dynamic modeling. For different weather types, a weather scene gating mechanism based on cloud cover, clear sky index, wind speed, and convective available potential energy is adopted to adaptively weight the feature channels, realizing the dynamic adjustment of meteorological factors on photovoltaic power output. The dynamic influence weights of each meteorological factor are output, and the power response characteristics under typical weather processes are extracted. At the same time, by combining spatial averaging, principal component analysis, and empirical orthogonal function decomposition, the total regional distributed photovoltaic power output is structurally characterized and its features are compressed. The ranking results of the influence weights of each meteorological factor, the response curves of photovoltaic power output to typical weather processes, and the analysis conclusions of the regional-scale meteorological-power coupling mechanism are output.

[0063] Module 4 of the comprehensive analysis module performs cross-scale fusion analysis based on the photovoltaic module output calculation model, the combination and dependency structure of meteorological factors, the influence weight of each meteorological factor, and the response curves of typical weather processes. Using a photovoltaic module output calculation model as the basic expression, and a combination of meteorological factors selected at the mesoscale as the input variable set, the influence weights of each meteorological factor at the macroscale are normalized to uniformly calibrate the intensity of the effects of different meteorological variables at different spatial scales, thus integrating the results at the micro, meso, and macro scales. Through weight mapping, the influence weights of each meteorological factor at the macroscale are transferred to the meso and micro scales, and the influence intensity of variables is collaboratively corrected based on the dependency structure of meteorological factors at the mesoscale, achieving a unified measurement of the degree of influence of meteorological variables at different scales. Finally, cross-scale variable coupling relationships are constructed based on the dependency structure of mesoscale factors, and these cross-scale variable coupling relationships are introduced into the photovoltaic module. The parameter correction process of the photovoltaic module output calculation model establishes the cross-scale response relationship between meteorological factors and photovoltaic module output; combined with the response curves of typical weather processes, the interaction relationship of variables under different weather scenarios is constrained; through decomposition analysis of the response curves of typical weather processes, macro-scale weather process changes are mapped step by step to meso-scale meteorological factor co-changes, and then transmitted to micro-scale module parameter changes, identifying the transmission path of meteorological factors and their enhancement or weakening effects between different spatial scales; consistency tests and sensitivity analyses are carried out to extract the dominant and secondary intervals of meteorological variables at different scales, and output the multi-dimensional meteorological conditions affecting photovoltaic output at micro, meso, and macro scales.

[0064] The above-disclosed embodiments are merely preferred embodiments of the present invention, but the present invention is not limited thereto. Any non-creative variations that can be conceived by those skilled in the art, as well as any improvements and modifications made without departing from the principles of the present invention, should fall within the protection scope of the present invention.

Claims

1. A method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output, characterized in that, Includes the following steps: Step S1: At the microscale, multidimensional meteorological variables are introduced based on the physical model of the photovoltaic module to perform dynamic parameter correction, thereby obtaining the corrected photovoltaic module output calculation model. Step S2: At the mesoscale, based on station-level meteorological and power output data, autocorrelation analysis, cross-correlation analysis, maximum information coefficient, and characteristic synergy methods are used to obtain the combination of meteorological factors and their dependency structure. Step S3: At the macro scale, construct a spatiotemporal relationship model based on regional meteorological field and multi-station power data, and output the influence weights of each meteorological factor and the response curves of typical weather processes; Step S4: Based on the photovoltaic module output calculation model, the combination and dependency structure of meteorological factors, the influence weight of each meteorological factor and the response curve of typical weather processes, a cross-scale fusion analysis is performed to output the influencing factors affecting photovoltaic output.

2. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 1, characterized in that, In step S1, photovoltaic module modeling is performed using a combination of physical modeling and simulation verification, and the simulation verification is conducted based on the software simulation platform PVsyst.

3. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 2, characterized in that, The physical modeling includes: using engineering approximation methods to obtain the simplified current-voltage equation as follows: The maximum power point parameter Open circuit state parameters ; Introducing multidimensional meteorological variables to the core parameters and Dynamic corrections are made, and multidimensional meteorological variables, including diffuse radiation intensity, are included. Direct radiation intensity Ambient temperature Surface temperature of photovoltaic modules Given wind speed V and humidity H, the corrected physical model I–V for calculating the output of the photovoltaic module is as follows: The correction relation satisfies: ; In the formula, and The original fitting parameters are those without considering multidimensional meteorological variables; ξ and η are the dynamic correction coefficients obtained by training the multilayer perceptron network based on multidimensional meteorological variables; the fitting parameters for the actual maximum power point are... Actual open-circuit state fitting parameters These are the core parameters after correction by multidimensional meteorological variables.

4. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 3, characterized in that, The dynamic correction coefficients ξ and η are solved by a multilayer perceptron network. The input is a multidimensional meteorological variable. The dynamic correction coefficients corresponding to C1 and C2 are solved through the multilayer perceptron network. The solution model is as follows: The two output branches correspond to C1 and C2 respectively, and the values ​​of the correction coefficients ξ and η are limited to [0, 0.3].

5. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 4, characterized in that, In step S2: historical meteorological data at the station level within the research scope and corresponding photovoltaic power generation data at the station level are obtained. The historical meteorological data includes station-level irradiance, temperature, humidity, wind speed and air pressure. The photovoltaic power generation data is the total photovoltaic power generation of the corresponding station. Historical meteorological data and corresponding photovoltaic power generation data are preprocessed to generate meteorological time series and power time series. The autocorrelation function is used to evaluate the autocorrelation characteristics of the meteorological time series and power time series respectively, and to identify the periodicity and lag characteristics of the meteorological time series and power time series. The cross-correlation between meteorological time series and power time series was analyzed, and a correlation analysis method based on the combination of maximum information coefficient and feature synergy was constructed to quantitatively characterize the relationship between multidimensional meteorological variables and corresponding photovoltaic power output, thereby obtaining the combination of meteorological factors that have a significant impact on mesoscale distributed photovoltaic power and their dependency structure.

6. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 5, characterized in that, In step S3: acquire reanalysis meteorological grid field data, distributed photovoltaic (PV) station power data, and station metadata within the target area; construct a multi-scale meteorological feature matrix based on the reanalysis meteorological grid field data; input the distributed PV station power data and the multi-scale meteorological feature matrix into a spatiotemporal relationship model; adopt a weather scene gating mechanism based on cloud cover, clear sky index, wind speed, and convective available potential energy; adaptively adjust the feature channels under different weather types; quantify the influence intensity of each meteorological factor on regional PV output under different weather processes; extract power response characteristics under typical weather processes; perform structured characterization and feature compression on the total regional distributed PV output; output the ranking results of the influence weights of each meteorological factor, the response curve of PV output to typical weather processes, and the analysis conclusions of the regional-scale meteorological-output coupling mechanism.

7. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 6, characterized in that, The method for calculating the clear sky index is as follows: Calculate the solar altitude angle using a solar geometric model. and solar azimuth The formulas for calculating the solar altitude angle and azimuth angle are: in, The latitude of the observation location is given by δ, the solar declination angle is given by h, and the hour angle is given by h. Clear-sky radiation is calculated based on solar altitude angle and azimuth angle, combined with the Ineichen clear-sky model. : ; in The solar constant, Atmospheric transmittance. The transmittance of scattered radiation. Clear-sky radiation Compared with the measured radiation amount Combined, calculate the clear sky index. : .

8. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 7, characterized in that, In step S4: The results at the micro, meso, and macro scales are integrated; weight mapping and collaborative correction are performed to achieve a unified measurement of the influence intensity of meteorological variables at different scales, and to construct cross-scale variable coupling relationships; By combining typical weather processes to analyze the cross-scale transmission paths of meteorological factors, consistency tests and sensitivity analyses are conducted to output the meteorological factors affecting photovoltaic power output.

9. The method for analyzing the influencing factors of multidimensional meteorological conditions on photovoltaic power output according to claim 8, characterized in that, The process of integrating results at the micro, meso, and macro scales is as follows: The I-V relationship of the photovoltaic module output calculation model after correction by multidimensional meteorological variables at the micro scale is used as the physical basis expression, and the combination of meteorological factors selected at the meso scale is used as the input variable set. The influence weights of each meteorological factor identified at the macro scale are normalized. The weight mapping and collaborative correction operation is performed as follows: The influence weights of various meteorological factors obtained at the macro scale are transferred to the meso and micro scales. Combined with the dependence structure of meteorological factors at the meso scale, the influence intensity of variables is synergistically corrected. The process of constructing cross-scale variable coupling relationships, analyzing the cross-scale transmission paths of meteorological factors in conjunction with typical weather processes, and performing consistency checks and sensitivity analyses are as follows: By combining the dependence structure of meteorological factors obtained at the mesoscale, a cross-scale variable coupling relationship is constructed. This cross-scale variable coupling relationship is then introduced into the parameter correction process of the microscale output calculation model to establish a cross-scale response relationship between meteorological factors and photovoltaic module output. Furthermore, by combining the response curves of typical weather processes at the macroscale, the variable interaction relationship under different weather scenarios is constrained. The response curves of typical weather processes are decomposed and analyzed, and the changes in weather processes at the macro scale are mapped step by step to the coordinated changes of meteorological factors at the meso scale, and then to the changes in component parameters at the micro scale. The transmission paths of meteorological factors at different spatial scales and their enhancement or weakening effects are identified. Consistency analysis and sensitivity analysis are performed on the influence direction, amplitude and weight ranking of the same meteorological factor at different scales, and the meteorological factors affecting photovoltaic power output are output.

10. A system for analyzing the influence of multidimensional meteorological conditions on photovoltaic power output, characterized in that, include: Microscale parameter correction module (1), mesoscale key factor extraction module (2), macroscale spatiotemporal modeling and quantification module (3), comprehensive analysis module (4). The microscale parameter correction module (1) is used to dynamically correct parameters at the microscale by introducing multidimensional meteorological variables based on the photovoltaic module physical model, and output the corrected photovoltaic module output calculation model. The mesoscale key factor extraction module (2) is used to extract meteorological factor combinations and dependency structures based on station-level meteorological and power output data at the mesoscale. The macro-scale spatiotemporal modeling and quantification module (3) is used to construct a spatiotemporal relationship model based on regional meteorological fields and multi-station power data at the macro scale, and output the influence weights of each meteorological factor and the response curves of typical weather processes. The comprehensive analysis module (4) performs cross-scale fusion analysis based on the photovoltaic module output calculation model, meteorological factor combination and dependency structure, influence weight of each meteorological factor and response curve of typical weather process, and outputs the influence factors of multidimensional meteorological conditions on photovoltaic output at the micro, meso and macro scales.

Images